flacs v9 manual

FLACS v9.0 User’s Manual

Copyright ©GexCon AS

Thursday March 12 2009

Contents

1 Introduction 1

1.1 About this publication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 About this manual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.5 Feedback from users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 Getting started 7

2.1 Prerequisites for users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Hardware and software requirements . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 Software installation and setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.4 Running FLACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.5 Help and support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.6 Introductory example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3 CASD 31

3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2 File menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3.3 Geometry menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.4 Object window in CASD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.5 Grid menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.6 Porosities menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.7 Scenario menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.8 Block menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

3.9 View menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

3.10 Options menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

3.11 Macro menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

3.12 Help menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

3.13 Potential bugs or problems with CASD . . . . . . . . . . . . . . . . . . . . . . . . . 110

ii CONTENTS

4 Flacs simulator 113

4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

4.2 The Run Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

4.3 Running several simulations in series . . . . . . . . . . . . . . . . . . . . . . . . . . 116

4.4 Output variables in FLACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

4.5 Files in FLACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

4.6 Input files to FLACS simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

4.7 Output files from FLACS simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 149

4.8 Potential bugs or problems with Flacs . . . . . . . . . . . . . . . . . . . . . . . . . . 153

4.9 Warning and error messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

5 Flowvis 155

5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

5.2 Creating a new presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

5.3 File menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

5.4 Edit menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

5.5 Page menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

5.6 Plot menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

5.7 Verify menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

5.8 Options menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

5.9 Help . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180

5.10 Flowvis examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180

6 Utility programs in FLACS 187

6.1 Geometry, grid and porosities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

6.2 Release source modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

6.3 Modifying simulation files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

6.4 Post-processing of simulation data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

7 Best practice examples 205

7.1 Combined dispersion and explosion simulations with FLACS . . . . . . . . . . . . 206

7.2 Simulation Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

7.3 Equivalent Stoichiometric Gas Cloud . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

7.4 Dispersion simulation with wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219

7.5 Hydrogen explosions and DDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219

8 Technical Reference 221

8.1 Definitions and gas thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . 222


CONTENTS iii

8.2 Stoichiometric reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

8.3 Governing equations for fluid flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224

8.4 Wall functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

8.5 Wind boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

8.6 Combustion modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229

8.7 Modelling of jet sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

8.8 Numerical Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

8.9 Linux Quick Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233

9 Nomenclature 237

9.1 Roman letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238

9.2 Greek letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239

9.3 Subscripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239

9.4 Dimensionless groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240

9.5 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240

9.6 FLACS variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241

10 References 243


Chapter 1

Introduction

1.1 About this publication


Copyright ©2009 GexCon AS

All rights reserved

Updated: January 26 2009

Typeset in Doxygen

Printed in Norway

Intellectual property notice

No part of this publication may be reproduced, stored in a retrieval system, or transmitted, inany form or by any means, electronic, mechanical, photocopying, or otherwise, without writtenpermission from GexCon AS.

GexCon AS hereby grants permission to use, copy, and print this publication to organizations orindividuals holding a valid licence for one or several of the software packages described herein.

For further information about GexCon AS, please visit the web site: http://www.gexcon.com

Exclusion of liability

GexCon AS has distributed this publication in the hope that it will be useful, but without anywarranty, without even the implied warranty of merchantability or fitness for a particular pur-pose.

Although great care has been taken in the production of this publication to ensure accuracy,GexCon AS cannot under any circumstances accept responsibility for errors, omissions, or advicegiven herein.

http://www.gexcon.com

2 Introduction

Registered trademarks

• FLACS, DESC, CASD, and Flowvis are registered trademarks of GexCon AS.

• Linux is a registered trademark of Linus Torvalds.

• Windows is a registered trademark of Microsoft Corporation.

Other product names mentioned herein are used for identification purposes only and may betrademarks of their respective companies.

1.2 Preface

Computational fluid dynamics (CFD) is a branch of fluid mechanics that uses numerical methodsand algorithms to solve and analyze problems that involve fluid flow, with or without chemicalreactions. Current use of CFD covers a broad range of applications, from fundamental theoret-ical studies involving models primarily derived from first principles, to practical engineeringcalculations utilizing phenomenological or empirical correlations.

Many of the hazards encountered in the society, and especially in the process industries, involveaccident scenarios where fluid flow in complex, large-scale, three-dimensional (3D) geometriesplay a key role. FLACS is a specialized CFD toolbox developed especially to address processsafety applications such as:

• Dispersion of flammable or toxic gas

• Gas and dust explosions

• Propagation of blast and shock waves

• Pool and jet fires

The development of FLACS started in 1980 at the Department of Science and Technology at Chris-tian Michelsen Institute (CMI). CMI established GexCon (Global Explosion Consultants) as a con-sultancy activity under the Process Safety Group in 1987. In 1992, the Science and Technologydepartment at CMI became Christian Michelsen Research (CMR), and CMR established GexConas a private limited company in 1998. GexCon AS is a wholly owned subsidiary of CMR, andholds the full proprietary rights to the CFD code FLACS.

The purpose of this manual is primarily to assist FLACS users in their practical work with thesoftware. In addition, the manual aims at documenting both the physical and chemical models,and the numerical schemes and solvers, implemented in the CFD code. Ample references topublished literature describe the capabilities and inherent limitations of the software.

1.3 Acknowledgements

The development of the FLACS software would not have been possible without the generouscontributions received from supporting companies and government institutions throughout theyears. The activity started at Christian Michelsen Institute (CMI) in 1980 with the Gas ExplosionProgrammes (GEPs), and FLACS-86 was the first version distributed to the supporting compa-nies.


1.3 Acknowledgements 3

Figure 1.1: The M24 compressor module represented in FLACS-86

The development of FLACS continued with the Gas Safety Programs (GSPs) and related projectsup to around 2000:

• BP, Elf, Esso (Exxon), Mobil, Norsk Hydro, and Statoil supported the development ofFLACS-86 during the First GEP (1980-1986).

• BP, Mobil, and Statoil supported the development of FLACS-89 during the Second GEP(1986-1989).

• BP, Elf, Esso, Mobil, Norsk Hydro, Statoil, Conoco, Philips Petroleum, Gaz de France, NVNederlandse Gasunie, Bundes Ministerium für Forschung und Technologie (BMFT), Healthand Safety Executive (HSE), and the Norwegian Petroleum Directorate supported the de-velopment of FLACS-93 during the First GSP (1990-1992).

• BP, Elf, Esso, Mobil, Statoil, Philips Petroleum, Gaz de France, HSE, and the NorwegianPetroleum Directorate supported the development of FLACS-94, FLACS-95, and FLACS-96during the Second GSP (1993-1996).

• BP, Elf, Exxon, Mobil, Norsk Hydro, Statoil, Philips Petroleum, Gaz de France, HSE,Agip, MEPTEC, and the Norwegian Petroleum Directorate supported the development ofFLACS-97, FLACS-98, and FLACS-99 during the Third GSP (1997-1999).

• BP, TotalElfFina (TEF), Norsk Hydro, Statoil, Gaz de France, Philips Petroleum, Mobil andsupported the LICOREFLA project (2000-2001).

Since 2000, various Joint Industry Projects (JIPs), funding from the European Commission (EU)and the Norwegian Research Council (NFR), and support and maintenance fees (S&M) from anincreasing number of commercial costumers have supported the development of the more recentFLACS releases, including several specialized versions of FLACS, such as DESC (Dust ExplosionSimulation Code), FLACS-Dispersion, and FLACS-Hydrogen:

• FLACS-Dispersion and FLACS-Hydrogen became available in 2001.• FLACS v8.0 came in 2003, including a test release of FLACS-Explo.• FLACS v8.1 came in 2005.


4 Introduction

• DESC 1.0 came in 2006.• FLACS v9.0 came in 2008, including a test release of FLACS-Fire.• GexCon also develops several in-house R&D tools, including FLACS-Explo, FLACS-

Aerosol, and FLACS-Energy.

GexCon is grateful to all companies, government institutions, and individuals that have partici-pated in the development of FLACS. We intend to honour these contributions by continuing todevelop the software, and thereby contribute to improved safety in the process industries.

1.4 About this manual

This User’s Manual describes a family of computational fluid dynamics (CFD) software productsfrom GexCon AS, generally referred to as FLACS:

• The preprocessor CASD• The CFD simulator Flacs• The postprocessor Flowvis• Utility programs in FLACS such as:

– geo2flacs, gm, and Porcalc– jet and flash– rdfile, cofile, and comerge– r1file, r3file, and a1file

These programs constitute a specialized CFD tool, FLACS, or ’standard FLACS’, designed tostudy releases of flammable gas and gas explosions in complex congested geometries, both on-shore and offshore. This manual also describes specialized versions of FLACS:

• FLACS-Hydrogen• FLACS-Dispersion• FLACS-Aerosol• FLACS-Energy• FLACS-Explo• FLACS-Fire• DESC

A full version of Standard FLACS exhibits the full functionality of FLACS-Hydrogen and FLACS-Dispersion, whereas DESC and FLACS-Fire are separate software products. FLACS-Energy,FLACS-Explo, and FLACS-Aerosol are still in-house R&D tools. The acronym FLACS (FLameACceleration Simulator) refers to the complete package of software products, whereas the termFlacs refers specifically to the numerical solver in the CFD code.

The latest release of FLACS is version 9.0 (FLACS v9.0). This version represents a major upgradeto the graphical user interfaces (GUIs), and is the first version that runs under both the Linux andWindows operating systems.

Getting started presents a detailed example for new users of FLACS, and Best practice examplescontains further examples that highlight various applications of FLACS, including some of thespecialized versions.

Technical reference contains technical reference material.


1.4 About this manual 5

1.4.1 Printing conventions in this manual

• The symbol ’>’ followed by text in typewriter font indicates command line input, e.g.:> command -options arguments (general syntax for commands)> find -name flacs (command line input in Linux)

• The symbol ’∗’ followed by text in typewriter font field input commands, e.g.:

∗ exit yes yes• The symbol→indicates a path through nested menu items or dialog box options, e.g.:

File→Save

Scenario→Ignition→Time of ignition• Certain features of the software may only be accessible through text file input, and the

content of a text file is also printed in typewriter font:THE FIRST LINE OF THE FILE ...THE SECOND LINE OF THE FILE ...... ...

• The format for describing keyboard and mouse input follows the pattern:

CTRL+C

CTRL+MOUSE+LEFT• The use of bold or italic font emphasizes specific words or phrases in the text.• The Nomenclature chapter lists the symbols and abbreviations adopted in this manual.

1.4.2 Special messages

Warning:

Look out for the potential pitfalls pointed out by this heading!

Attention:

Be aware of practical information pointed out by this heading.

Remarks:

Take notice of the points summarized under this heading.

See also:

Follow up the additional sources of information suggested by this heading if required.

1.4.3 Job numbers

The typical application of the FLACS software is to quantify potential consequences of industrialaccident scenarios involving compressible fluid flow, with or without chemical reactions. Propercharacterization of a particular problem may involve several simulations, and it is usually conve-nient to organize the files from related scenarios in a dedicated directory. The individual FLACSsimulations are assigned job numbers, or simulation numbers, or simply jobs. A user may forinstance type:

> run9 flacs 010100

on the command line in Linux to start a FLACS simulation for job number 010100.

The job numbers are constructed from a six-digit string ijklmn, where traditionally:


6 Introduction

• ij is the project number.• kl is the geometry number.• mn is the sequence number.

The default job number used in many of the examples in this manual is 010100, i.e. project 01,geometry 01, simulation 00. However, each of the six digits in the job number may in principletake on any integer value from zero to nine, and the references to project, geometry, and sequencenumbers only apply when the job numbers are derived from the file database in CASD.

Any updated version of this manual may be found on the FLUG web site.

1.5 Feedback from users

Feedback on the content in this manual is most welcome, and FLACS users may submit theircomments or suggestions by e-mail to: [email protected]

When submitting comments or suggestion to the content of the manual, or when pointing outmisprints in the text, please indicate the relevant page numbers or sections, and the correspond-ing version of the manual (date issued).


mailto:[email protected]

Chapter 2

Getting started

8 Getting started

This chapter describes the basics of setting up the FLACS software for new users, including rec-ommendations concerning the user threshold, typical hardware requirements, and proceduresfor installing FLACS on both Linux and Windows.

2.1 Prerequisites for users

Efficient use of FLACS does not require detailed knowledge about computational fluid dynam-ics (CFD). However, users should possess some experience in the application of computers forroutine tasks, such as text editing. Proper interpretation of simulation results requires adequateknowledge within the field of fluid dynamics. A suitable starting point for the novice in the fieldof gas explosions is the Gas Explosion Handbook (Bjerketvedt et al., 1993) from Christian MichelsenResearch (CMR), and new users of FLACS should attend a three-day introductory course arrangedby GexCon AS (http://www.gexcon.com).

2.2 Hardware and software requirements

FLACS v9 is available on Linux and on Microsoft Windows. The hardware requirements forrunning the FLACS software depend to some extent on the size of the problem in question, i.e.the number of grid cells required to resolve the computational domain properly. Most moderncomputers, be it desktops and laptops, will perform well for small or medium sized problems.A powerful screen card may be required to handle large geometries in CASD, extra memory(RAM) is necessary for simulating large problems, and storage of large amounts of simulationdata dictates the requirements for disk space.

Hardware requirements:

• Processor: Intel or AMD ix86 32 bit, Intel EM64T or AMD64. Intel IA64 is not supported.• Internal memory; 2GB or more recommended.• Free harddrive capacity: 350MB for software installation and typically 100GB simulation

space.• Graphics card using NVIDIA chip set. Graphics cards using for instance ATI or Intel

chipsets are in general not supported.• DVD-RW drive recommended.• High resolution colour screen (minimum 19", 1600x1200, 24 bit color depth).

FLACS v9 has been tested on the following platforms.

Linux:

• OpenSuse 10.0, 10.2, 10.3, 11.0• CentOS 4.6, 5.1• Ubuntu 7.10• Fedora 8

Microsoft Windows:

• XP (32 bit)• Vista (32 bit)



2.3 Software installation and setup 9

Red Hat Enterprise Linux 4.6, 5.1 is expected to be FLACS v9 compatible since it is compatiblewith CentOS 4.6, 5.1.

For updated hardware and software requirements, please refer to GexCon’s website,http://www.gexcon.com .

2.3 Software installation and setup

A license server is necessary for running FLACS. This section presents FLACS installation, theFLACS Licence Server and the FLACS Configuration Wizard that guides users through the basicsteps of setting up a FLACS Licence Server. All FLACS installations on a network acquire theirindividual licenses from a central licence server, and only one FLACS License Server shouldtherefore be running on a given network.

FLACS is distributed in a single setup file.

2.3.1 On Linux

On Linux FLACS can be installed system wide, in which case FLACS will be available to all users,or in a user’s home directory, in which case it will be available to this user only.

2.3.1.1 Installing in users home directory

If only one person will be using FLACS, the software can be installed in this users home directory.FLACS will by default be installed under /home/my_user/GexCon.

Save the installation package to a convenient location.

Make sure the file is executable:

> chmod u+x /home/my_user/flacs-v9.0-installer

Run the installation program:

> /home/my_user/flacs-v9.0-installer

Please follow the instructions given. It is recommended to keep the default parameters.

FLACS requires a license to run. The license is provided by a license server, which is installed ononly one machine on the local network. During the installation the user can choose to install:

1. Both FLACS software and FLACS license manager2. FLACS license manager only

For a user home directory installation option 1 should be selected.

The FLACS license manager must be set up before using FLACS. Please refer to the section aboutFLACS configure wizard.

2.3.1.2 Installing system wide as super user

To install FLACS system wide, access to the system super user ("root") is required./path/to/installation is the path to the location of the FLACS installation package.



10 Getting started

Change user to super user ("root"):

> su <give password>

Make sure the file is executable:

> chmod u+x /path/to/installation/flacs-v9.0-installer

Run the installation program:

> /path/to/installation/flacs-v9.0-installer

Please follow the instructions given. It is recommended to keep the default parameters.



Option 2 can be used to install a FLACS license manager on a system not running FLACS. Alter-natively one FLACS workstation in the network can be set up to serve licenses to all other FLACSinstallations in the network.

The FLACS license manager must be set up before using FLACS. Please refer to the section aboutFLACS configure wizard.

2.3.2 On Windows

To install FLACS on Windows please double-click the installation package "flacs-v9.0-installer.exe". This will start the installation wizard. Please follow the instructions given. It isrecommended to keep the default parameters.



Option 2 can be used to install a FLACS license manager on a system not running FLACS. Alter-natively one FLACS workstation in the network can be set up to serve licenses to all other FLACSinstallations in the network.

The FLACS license manager must be set up before using FLACS. Please refer to section Settingup the FLACS license server.

2.3.3 Setting up the FLACS license server

FLACS version 9.0 has a completely new license server/manager system, which operates througha network protocol. This means that the license manager can be installed anywhere on the net-work, as long as it is available to the FLACS clients through the local network. The license man-ager can be installed locally on the machine where the FLACS simulation software is installed, or



separately from the simulation software. Only one license manager should be running on yournetwork, and this is where the FLACS license is installed. All other FLACS installations shouldbe set up using this license manager.

After the installation is finished, the FLACS license configuration utility should start automat-ically. In the event that this does not happen please start the configuration utility as follows,depending on your installation.

Linux:

> /usr/local/GexCon/FLACS_v9.0/bin/run configureWizard

Windows:

> C:\Program Files\GexCon\FLACS_v9.0\bin\configureWizard.exe

Alternatively it can be started from the FLACS Runmanager Help→Start Configuration Wizard.

If FLACS is installed system wide (installed as root), on Linux, the license manager must berunning as user root.

The configuration utility will guide you through the setup of the license manager. The config-uration utility is also used to configure a FLACS installation that gets its license from a licensemanager on a separate machine.

2.3.3.1 Setting up the license server on client only FLACS installation

If a FLACS license server is installed and running somehwere on the local network, the FLACSinstallation must be configured to connect to the license server.

Figure 2.1: Setting up the license server on client only FLACS installation

2.3.3.2 Setting up the license server on a combined license server and client FLACS installa-tion

If there is no FLACS license server available on the local network, a license server must be in-stalled. To install a license server together with the FLACS simulation software, on the same


12 Getting started

machine, please use the following procedure. Alternatively a FLACS license server can be in-stalled on a separate machine, with or without FLACS software. Please refer to section Stan-dalone FLACS license manager installation.

Figure 2.2: Setting up the license server on a combined license server and client FLACS installa-tion (steps 1 and 2)

Figure 2.3: Setting up the license server on a combined license server and client FLACS installa-tion (steps 3 and 4)

2.3.3.3 Standalone FLACS license manager installation

It is possible to install the FLACS license manager only. This is useful if you would like to havethe license manager on a separate machine. To do this select the appropriate option during in-stallation (see Software installation and setup).



To configure a standalone FLACS license manager prompt the license manager for an activationkey, by running the following command in a terminal window.

Linux:

> /usr/local/GexCon/FLACS_LicenseManager/bin/FLMserver --get-ActivationKey

Windows:

> C:\Program Files\GexCon\FLACS_LicenseManager\bin\FLMserver.exe --get-ActivationKey

Send the activation key together with the IP address and license manager communication portnumber to <[email protected]>.

The communication port defaults to 25001. Please make sure that this port is available, and openon your system. If you are not sure about this please contact your system administrator.

GexCon will, based on the activation key, create a license text file. This file must be saved to:

Linux:

> /usr/local/GexCon/FLACS_LicenseManager/license/license-server.flm

Windows:

> C:\Program Files\GexCon\FLACS_LicenseManager\license\license-server.flm

Note that when using a standalone FLACS license manager, the license manager must be startedmanually each time the computer is restarted. This can be done using a startup script (not pro-vided).

2.3.3.4 Starting FLACS license manager as a service on Windows

The FLACS license manager can be started as a service on Windows using the following proce-dure.

1. Verify that the FLACS License Manager is working properly as a desktop application

(a) FLACS software and license key must be installed (see procedure above)(b) Test run FLMserver with the graphical user interface and then quit: > "C:\Program

Files\GexCon\FLACS_LicenseManager\bin\FLMserver.exe"(c) Make sure to quit FLMserver, the service will not function if there is a desktop FLM-

server running.

2. Download and install the Windows Resource Kit (rktools.exe)

(a) See the following links about Windows Services and related tools:

• http://search.microsoft.com/results.aspx?mkt=en-US&setlang=en-US&q=rktools.exe• http://support.microsoft.com/kb/137890

3. Install the FLACS License Manager service "FLMserver" using INSTSRV:

(a) > instsrv FLMserver "C:\Program Files\Windows ResourceKits\Tools\srvany.exe"

(b) The service can be removed with > instsrv FLMserver REMOVE(c) The path to srvany.exe might be different on your Windows installation



http://search.microsoft.com/results.aspx?mkt=en-US&setlang=en-US&q=rktools.exe

http://support.microsoft.com/kb/137890

14 Getting started

4. Run REGEDIT to set up the details of the service

(a) It is strongly advised to backup your current registry before editing(b) > regedit(c) Locate and select the FLMserver key:

• "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\FLMserver"(d) Add one new value for FLMserver: Description

• Edit→New→String Value : "Description"• Description = "FLACS License Manager service."

(e) Add one new key for FLMserver: Parameters

• Edit→New→Key : "Parameters"– Add two new values for FLMserver\Parameters: Application and AppParam-

eters* Edit→New→String Value : "Application"* Edit→New→String Value : "AppParameters"* Application = "C:\Program Files\GexCon\FLACS_-

LicenseManager\bin\FLMserver.exe"* AppParameters = –without-gui* IMPORTANT NOTE: options start with double dashes: –without-gui

• The service will start automatically on reboot, it can also be started/stopped man-ually:

– Control Panel→Administrative Tools→Services

2.3.4 Setting up the FLACS environment

After installation FLACS programs can be accessed from the system menu, in the following loca-tions:

Linux (KDE): Start→Applications→Edutainment→Construction

Linux (Gnome): Applications→Other

Windows: Start→All Programs→GexCon→FLACS_v9.0

Some systems may require the user to log out and restart before FLACS will appear in the systemmenu.

Desktops that do not follow the freedesktop.org standards will not install an icon in the Appli-cations menu. This will happen on older distributions. In these cases, the user may be able toinstall icons and associations manually. Refer to your GNU/Linux distribution vendor for detailson how to customize your desktop.

2.3.4.1 FLACS User setup on Linux

For easy access to FLACS from the command line add the following text to you startup file.

If you use the csh/tcsh shell, edit or create the .cshrc file:

alias run9 /usr/local/GexCon/FLACS_v9.0/bin/run

If you use the bash shell, edit or create the .bashrc file:

alias run9=/usr/local/GexCon/FLACS_v9.0/bin/run

FLACS programs can the be started by typing eg. run9 flowvis.


2.4 Running FLACS 15

2.3.5 Uninstalling FLACS

Linux: Run the program "/usr/local/GexCon/uninstall-GexCon.sh".

Windows: FLACS can be uninstalled using Control Panel/Add or Remove Programs.

2.4 Running FLACS

A typical simulation session with the CFD code FLACS involves several steps. Assuming FLACSis properly installed on the computer, including valid lisence files for the software, users caninitiate a FLACS session by clicking the FLACS icon on the desctop:

Figure 2.4: The FLACS icon

This should open the Run Manager window:

Figure 2.5: The FLACS Runmanager

Some of the main tasks of the Run Manager are:

• Starting the Licence Manager• Starting the preprocessor CASD• Running CFD simulations


16 Getting started

• Starting the postprocessors Flowvis

The preprocessor should start when clicking the CASD icon in the Run Manager:

Figure 2.6: The CASD icon

The CASD window looks like this:

Figure 2.7: FLACS preprocessor CASD

Work in CASD often involves opening the Database window from the Geometry menu:

Geometry→Database

The Database window looks like this:


2.4 Running FLACS 17

Figure 2.8: Geometry database window

Typical tasks performed from the Database window include:

• Creating a new database and new geometries

• Opening existing databases and geometries

• Creating new materials (i.e. colours), or modifying existing materials

• Creating new objects, or modifying existing objects

The New Object button, available in the Objects tab in the Database window, opens the Objectwindow:


18 Getting started

Figure 2.9: CASD object window

The main purpose of the Object window is to construct a new object, or to modify an existingobject. Users can build complex objects by adding or subtracting several insides (i.e. boxes orcylinders). Any geometry can consist of one or several objects, or assemblies of several objects.An alternative way of working with geometries involves geometry import using the geo2flacsutility . However, this requires that a representation of the geometry already exists on a compat-ible CAD format (typically Microstation or PDMS).

Apart from geometry building, the menus in CASD also perform the following tasks:

• Definition of the computational domain and the computational grid• Porosity calculations with the utility program Porcalc, as well as porosity verification• Scenario setup, including:

– Definition of monitor point locations, and selection of output variables– Specification of boundary conditions– Specification of vent panels and leaks– Specification of fuel type– Specification of ignition position and time of ignition

After defining the scenario, the next step is to run the actual FLACS simulations:

• Simulations can be started and monitored with the run manager• The same operations can be controlled from the command line in Linux

> run9 flacs 010100

Note that the Run Manager also monitors the simulations while they are running.


2.5 Help and support 19

The final step in a FLACS session is typically the presentation and verification of simulation re-sults with the postprocessor Flowvis, as well as data extraction and reporting. The postprocessorshould start when clicking the Flowvis icon in the Run Manager:

Figure 2.10: The Flowvis icon

The Flowvis window looks like this:

Figure 2.11: FLACS postprocessor Flowvis

Some of the most frequently used features in Flowvis include:

• Verifying porosities in a geometry• Creating scalar-time plots, 2D-plots, 3D-plots, ...• Creating animations

Data reporting may also include the extraction of numerical simulation results with the utilityprograms r1-file and r3-file. These programs run only from command line input in the currentversion of FLACS.

2.5 Help and support

FLACS users can get technical support by contacting GexCon software department:


20 Getting started

Email: [email protected]

Phone: +47 55574330

Commercial customers are entitled to support and maintenance:

Support: Up to 70 hours of email or phone support per year

Maintenance: New releases of FLACS as they are made available

In addition to the above the user has access to the FLACS User Group web site, which containsinformation about FLACS, including a FAQ (Frequently Asked Questions) and a self supportportal where the user can search for answers (as of November 2008 GexCon is working on im-plementing the self support portal, but a release date is not yet decided)

The support and maintenance requires the user to have a payed and valid support and mainte-nance contract.

2.6 Introductory example

This chapter contains an introductory example. It gives a first impression of how to set up andrun a simple FLACS explosion simulation. For additional examples see sections Best practiceexamples and Flowvis examples.

2.6.1 Things to keep in mind before you begin

FLACS is a CFD (Computational Fluid Dynamics) Explosion Simulator tool. The input to a CFDcalculation is:

• A geometry, either created manually for the specific purpose, or imported from a CADsystem

• A grid which divides the simulation domain into cells. In one cell a variable (eg. pressure)does not vary in space. FLACS use a regular, Cartesian grid, which means box grid cells.

• Various scenario parameter, such as boundary conditions, monitoring point locations, gascloud size, position and composition, and ignition location.

All of the above is normally handled in the FLACS pre-processor CASD. The geometry is saved toa file structure, called a file database. The file database file structure starts in a top level directorygiven a name with suffix ".db". The file database should not contain user files, or files otherthan those created by the file database interface in CASD.

In addition to the file database a number of other files are created before and during the simula-tion. All files contains the job number, a 6 digit number. The following files are created as input tothe simulation (010101 is the job number).

cg010101.dat3 The grid file

cs010101.dat3 The scenario file

co010101.dat3 The geometry file. This file contains a snapshot of the geometry contained in thefile database.

cp010101.dat3 The porosity file, which is created by Porcalc. Please see section and Porcalc fordetails.



2.6 Introductory example 21

During the simulation a set of result files will be created:

r1010101.dat3 Scalar-time output from monitor points

r3010101.dat3 Field output at selected times. Needed to create 2D and 3D plots

rt010101.dat3 Simulation log file

FLACS can also create and use other files. Please see section Files in FLACS for details.

Due to the number of files created by each simulation it is important to create a good file struc-ture of directories to keep track of the files. See section Files in FLACS for details and furtherrecommendations.

2.6.2 Initialising the work directory

As FLACS creates a relatively large number of files it is important to have a good system for bookkeeping. It is recommended to start out with an empty directory.

2.6.2.1 On Linux

Make a distinct directory (DIRECTORY_NAME) in which you perform the exercise:

> mkdir DIRECTORY_NAME

Move into this directory:

> cd DIRECTORY_NAME

Copy geometry files (notice the space before the ".").

> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/ex2/*00001* .

Start up the FLACS runmanager:

> run9 runmanager

2.6.2.2 On Windows

1. Make a distinct directory in which you perform the exercise: Open the file browser ("MyDocuments") and choose File→New→Folder.

2. Copy files from C:\Program Files\GexCon\FLACS_v9.0\doc\examples\ex2\∗00001∗(∗00001∗ means all files containing the text "00001").

3. Start the FLACS runmanager by clicking the desktop icon, or go to Start Menu→AllPrograms→GexCon→FLACS_v9.0→FLACS Runmanager.

2.6.3 Initialising and starting the preprocessor CASD

Use Run Manager → Tools → CASD (or click the FLACS pre-processor icon)


22 Getting started

2.6.3.1 Open and view the geometry in CASD (Move cursor to the CASD window)

1. choose OPEN in the FILE menu OR ∗ file open <CR> OR ALT-f o (<CR> means carrigereturn, ie. the enter key)

• CASD Ask for opening an existing job file

2. choose 100001.caj <OK>

• CASD: Open jobfile 100001, using MOUSE+LEFT

3. if any error message appears click <OK>

• CASD: Ignore error message => error message• CASD: Play with visualisation options, fly through geometry etc.

Figure 2.12: The geometry used in example 1

2.6.3.2 Make a grid for the simulation

Make a grid (mesh) for the simulation, calculate porosities (module dim.: 25.6m x 8m x 8m, originin corner below the control room).

1. Choose SIMULATION_VOLUME from GRID menu

• CASD: To enter the extension of the simulation domain

2. Enter -16 <TAB> -8 <TAB> 0 <TAB> 40 <TAB> 16 <TAB> 16 <OK>

• CASD: Volume is defined (16m out from vent, 8m to the sides; observe - sign)



3. In GRID menu, choose DIRECTION X4. In GRID menu, choose REGION and enter 56 <OK>

• CASD: 56 grid cells chosen (1.0m grid size).

5. Repeat steps for Y direction and use REGION 24

• CASD: 24 cells in Y-direction

6. Repeat steps for Z direction and use REGION 16

• CASD: 16 cells in Z-direction

7. In GRID menu, click INFORMATION, and <OK> to close window

• CASD: Check that grid dimension is 1.0m as intended

8. Choose SAVE from the FILE menu

• CASD: Save geometry and grid files

9. Choose CALCULATE from POROSITIES menu

• CASD: Map geometry information onto the grid, porcalc

10. Choose DISPLAY OFF in the GRID menus

• CASD: Don’t draw the grid anymore

Figure 2.13: Embedding the grid


24 Getting started

Figure 2.14: Porsity calculations using Porcalc

2.6.3.3 Define explosion scenario

1. Choose MONITOR_POINTS in SCENARIO menu OR ∗ scen mon <CR>

• CASD: Define where to measure variables

2. Click <ADD>, <EDIT> and 0.8 <TAB> 4.7 <TAB> 7.9 <OK>

• CASD: Add and define location of monitor point 1

3. Repeat this for point 2 (12.3, 4, 0.1) and point 3 (24, 7.9, 7.9)

• CASD: To edit a non-highlighted monitor, click on its number

4. Click <OK>

• CASD: Close MONITOR_POINT window

5. Choose SINGLE_FIELD_SCALAR from SCENARIO menu

• CASD: Define which variables to report at monitors

6. Click on <P>, drag mouse pushing MOUSE+LEFT across all monitors, <OK>

• CASD: Log pressure at all three transducers

7. Repeat for <PIMP> and <DRAG>

• CASD: Log pressure impulse and dynamic pressure, too

8. Click <OK> and choose SINGLE_FIELD_3D from SCENARIO menu

• CASD: Define variables for contour plots

9. Click on <P>, CTRL-<PROD>, CTRL-<VVEC>, <OK>

• CASD: Pressure, flame and velocity vectors. CTRL needed to select more than one(NB! deselect when using the scroll bar)

10. Choose SIMULATION in SCENARIO menu OR ∗ scen sim <CR>

• CASD: Choose output and simulation parameters

11. Click on <NPLOT>, enter 50 <OK>, <OK>

• CASD: Increase number of contour plots, return to main menu

12. Click on GAS_COMP... in SCENARIO menu OR ∗ scen gas_c <CR>

• CASD: Define gas cloud loc., size, comp. and concentration

13. Click on <POS...>, 0 <TAB> 0 <TAB> 0 <OK>

• CASD: Position of bounding box describing gas cloud



14. Click on <DIM...>, 25.6 <TAB> 8 <TAB> 8 <OK>

• CASD: Dimension of gas cloud equals module dimensions

15. Click on <VOL...>, <METHANE> 91.7 <OK> <ETHANE> 7 <OK> <PROPANE> 1.3<OK> <OK>

• CASD: Gas composition is defined

16. Click on <EQUI...> 1.05 <TAB> 0 <OK> <OK>

• CASD: Slightly rich gas mixture is chosen ER=1.05

17. Click on IGNITION in SCENARIO menu <POS...> 12.5 <TAB> 4.1 <TAB> 4.25 <OK><OK> OR ∗ scen ign pos 12.5 4.1 4.25 OK <CR>

• CASD: Define location of ignition (12.5, 4.1, 4.25)

18. Choose SAVE from the FILE menu

• CASD: Save all files, ready to run flacs

19. Minimize CASD

• CASD: Leave CASD for now, can be activated easily

Figure 2.15: Adding monitoring points


26 Getting started

Figure 2.16: Choosing variables for 3D output

Figure 2.17: Adding a gas cloud and choosing the gas composition



2.6.4 Start FLACS simulation

Select the job in Run Manager and click simulate (if job not visible, use add directory or if di-rectory is already added, right click and rescan), check how the simulation starts up (click logfile)

Figure 2.18: Running a simulation in the FLACS Runmanager

2.6.5 Study results in post prosessor Flowvis

Use Run Manager → Tools → Flowvis (or click the FLACS post-processor icon)

1. choose ADD from Page menu (or CTRL+a)

• FLOWVIS: Prepare first page

2. click MOUSE+RIGHT, choose PLOT_TYPE and SCALAR_TIME plot

• FLOWVIS: Plotting of time histories of variables

3. choose 100001 and P with MOUSE+LEFT, select all 3 monitors (drag mouse) <OK>

• FLOWVIS: Plot pressure time history at all monitors

4. <RESCAN>

• FLOWVIS: if simulation is running rescan will update plot

5. Choose MODIFY in the Page menu (or CTRL+m), enter <TAB> 1 <TAB> 2 <OK>

• FLOWVIS: divide page into 2 plots

6. Click at lower frame, then MOUSE+RIGHT, PLOT_TYPE, ANNOTATION_ST (or CTRL+0)

• FLOWVIS: show numerical values from pressure plots


28 Getting started

7. ADD page and do the same for the DRAG and PIMP variables

8. Choose ADD in Page menu (or CTRL+a), click MOUSE+RIGHT, PLOT_TYPE, 2D... (orCTRL+2)

• FLOWVIS: prepare 2D contour plot

9. Choose 100001, P, click <OK>

• FLOWVIS: contour plot of pressure

10. click MOUSE+RIGHT, choose PLOT_DOMAIN, change k-index to 5 <OK>

• FLOWVIS: choose XY-cut plane through ignition

11. Click MOUSE+RIGHT, choose VARIABLE_APPEARANCE change Value Range Setting toFixed

• FLOWVIS: choose a user-defined fixed scale for all time steps

12. Choose Min. Value as 0.05 and Max. Value as 2.0

• FLOWVIS: define the scale

Figure 2.19: Showing pressure-time curves with annotation in Flowvis



Figure 2.20: 2D cutplane plot showing over-pressures

Figure 2.21: Setting plot domain for a volume plot

Time steps can now be changed moving the bottom scroll bar to the right, page can be variedusing the right scroll bar.

1. Repeat this method for PROD and VVEC variables (these can be plotted on the same plot)


30 Getting started

• FLOWVIS: visualize flame and velocity vectors

Try to show PRESSURE and PROD on the same page using PAGE MODIFY (use a fixed scalefor PROD from 0.15 to 0.2 and change Min. Color Index to 9 and Max to 10) Now that you arefamiliar with Flowvis, try the volume plot menu to study the development of flame (PROD) andpressure Use PLOT DOMAIN to narrow the view window and see below the ceiling

2.6.6 Study the effect of ignition location

Enter CASD, open the 100001.caj job-file, save this as a new job number e.g. 100002.caj Changeignition location in order to study how pressures may vary with different ignition locations Endignition (0.5, 4.1, 4.25), (job number 100002) Your own assumed worst-case location (job number100003)

Report highest pressure achieved on monitor point

Make animation of either 2D or volume plots using the export menu (with all timesteps)


Chapter 3

CASD

32 CASD

The preprocessor CASD for the CFD simulator FLACS is used to prepare the input data, or jobdata , that defines a FLACS simulation: geometry model, computational grid, porosites, andscenario description. CASD is an acronym for Computer Aided Scenario Design.

CASD 4 released in 1994, use X11 graphics, but a new version is available based on QT

CASD 5 released in 2001, use Open Inventor graphics

CASD 6 released in 2008, use QT and Coin 3D graphics

This manual describes CASD 6, but the general functionality of CASD 6 is in principle the samefor CASD 4 and CASD 5. CASD 6 is fully backward compatible with CASD 4 and CASD 5.

3.1 Overview

This section provides a general overview of the functionality in CASD.

3.1.1 Starting CASD

Users start CASD by clicking the CASD icon in the run manager window:

Figure 3.1: The CASD desktop icon

or alternatively by executing the command:

> run9 casd6

on the command line in Linux.

3.1.2 CASD command line options

The following options can be given when starting CASD on the command line:

Option Description-macro macro file name Read input from specified macro file-numMat maximum number of materials Default is 50-numObj maximum number of objects Default is 10000-numAsis maxmimum number ofassemblies/instances

Default is 3500

-stackAsis maxmimu number of nested assemblylevels

Default is 8

-noLock Turns of locking on the database files. Mustnot be used if more than one user accesses thedatabase simultaneously. This option speedsup the database operations significantly.


3.1 Overview 33

-display and others Linux: options accepted by XTable 3.1: CASD command line options

Example:

Linux:

run9 casd -numObj 20000 -numAsis 20000 -noLock

Windows:

casd -numObj 20000 -numAsis 20000 -noLock

Alternatively the options can be set permanently in the FLACS Runmanager,Options→Preferences. This will only apply if CASD is started from the Runmanager.

3.1.3 The main window in CASD

Starting CASD 6 opens the main window.

Figure 3.2: The main window in CASD

The main window is divided into the following parts:

• The menu bar


34 CASD

• The icon bar

• The command input field

• The geometry window(s)

• The status field

These parts are described in the following subsections.

3.1.4 The menu bar

The menu bar contains the following menus:

• File

• Geometry

• Grid

• Porosities

• Scenario

• Block

• View

• Options

• Macro

• Help

The options on the various menus are described in separate sections in this chapter.

3.1.5 The icon bar

The icon bar contains the following toolbars:

• Main toolbar, provides shortcuts to several of the commands on the meny bar:

– New, Open, Save, Save as, Import, and Result on the File menu.– Database icon on the Geometry menu.– Calculate and Verify porosities on the Porosity menu.

• Graphics toolbar, controls various features of the geometry window(s).

– View splitting.– Rectangle zoom.– Spinning (toggle on/off).– Highlighting option, from filled only (0) to various degrees of contour highlighting

(1-5).

• Drawing toolbar, opens the plan drawing dialog box:

– Specifying file names for texture (e.g. drawings).– File formats: PNG, JPEG, GIF, TIFF


3.1 Overview 35

3.1.6 The command input field

The command input field represents an alternative interface between the user and CASD, inaddition to the regular menus on the menu bar. The control input field contains a scrollablecommand history list, and a current command context indicator (left side). The user controls thecommand history list from the keyboard:

• UP: retrieves the previous line from the command history list• DOWN: retrieves the next line from the command history list• RETURN: processes the content of the command input field

Hence, the user can choose whether to use a menu options on the menu bar, e.g: File→Exit→Yes(to exit and save) or to execute, after typing or retrieving, the following command in the com-mand input field:

∗ file exit yes yes Command line input will in many situations be the most efficient wayto work with CASD, and other sections in this chapter present additional examples on how touse this feature.

Examples: Using the command input field in CASD

• Select a box primitive in an object. The following command moves the box to (2, 2, 2), andwould cause the properties dialog to be shown

– ∗ edit properties 2 2– This is because the position is not completely specified. The user does not have to

specify all parameters, but must include all values for the parameter specified.

• If the user wants to edit one of the last parameters in the dialog, it is not necessary to specifyall the parameters in front. The parameter name can be used to indicate which parameterto edit

– ∗ edit properties size 2 2 2 vol_por 0.5

• The user can also supply the answer to a question in the input field. To delete an assem-bly/instance, CASD will ask to confirm the operation. To avoid the question dialog, typethe following command

– ∗ geometry delete yes– or shorter: ∗ ge de y

• To direct the output from a list to a file, append the file name after the list command. Forinstance, to list geometries in the database, enter the following command, which will createthe text file outfile.txt

– ∗ geometry list outfile.txt

3.1.7 The graphical area

The graphical area in the main window displays the geometry and the computational grid. Inaddition to the options on the View menu, there are several ways of manipulating the view:

• Rotation: MOUSE+LEFT• Panning: CTRL+MOUSE+LEFT• Zoom: MOUSE+SCROLL• Rectangle zoom: MOUSE+RIGHT+SELECT


36 CASD

• Splitting and closing views: MOUSE+RIGHT+SELECT

The use of these features are quite intuitive, and they will not be described in more detail in thismanual.

3.1.8 The message area

The message area in the main window contains information concerning the active database,project, geometry, grid, and units.

3.1.9 Files in CASD

CASD stores job data on a set of files. For the arbitrary job number 010100, the most importantfiles are:

• Header file, 010100.caj: ASCII file created by CASD; defines the co, cg, and cm files used byCASD.

• Geometry file, co010100.dat3: binary file created by CASD; contains a list of primitives froma CASD database that define the geometry; used by Porcalc and Flowvis.

• Grid file, cg010100.dat3: binary file created by CASD; defines the computational mesh; usedby CASD, Flacs, and Flowvis.

• Porosity file, cp010100.dat3: binary file created by Porcalc (typically from the Grid menu inCASD); defines the porosities for each grid cell; used by Flacs and Flowvis.

• Polygon file, cm010100.dat3: binary file created by CASD; defines the polygon model; usedby Flowvis (if the file exists).

• Scenario file, cs010100.dat3: ASCII file created by CASD; defines the general scenario (mon-itor points, output variables, fuel region, pressure relief panels, ignition position, etc.); usedby CASD, Flacs, and Flowvis.

The grid-file is also called the obstruction file, or co-file, and is not a direct input to the simulation; itis however used by Porcalc when generating the porosity file. The File menu in the main windowcontains commands for creating, opening, and saving the various job files. See section Files inFLACS for further information.

3.1.10 Working with geometries in CASD

To implement the geometry model in CASD can often be the most time consuming part of aproject. For modern process facilities it may be possible to import a geometry from an existingCAD model, but for many installations the geometry must be constructed manually from draw-ings, photographs, etc.

A large projects, such as a full probabililistic analysis, can involve hundreds of CFD simulations,and each simulation will typically produce 10-15 different files. Hence, it is very important toorganize the files in a well-structured manner.

The building blocks in a CASD geometry are instances of objects. The structure within an objectis a so-called Constructive Solid Geometry (CSG) model, where simple solid primitives (boxesand cylinders) are combined by Boolean operators (unions and left differences).

Objects in CASD can be either global or local. Several geometries can contain instances of thesame global object, whereas a local object can only be included in the geometry where it wascreated. It is generally recommended to use global objects, and avoid the use of local objects.


3.1 Overview 37

The list of information required to implement a typical process facility, such as an offshore oilplatform or an onshore process plant, is quite extensive:

• Plot plan• Sectional drawings• Piping plan• HVAC layout• Cable trays layout• Framing plans• Cladding• Deck plan

Most FLACS users find it convenient to define standardized axis directions, and the followingconvention is used by GexCon for typical process facilities:

• East-West along the x-axis, with positive x towards the east.• North-South along the y-axis, with positive y towards the north.• Up-Down along the z-axis, with positive z pointing upwards.

This results in a conventional right handed coordinate system, where the lower south-westerncorner of the facility coincides with the origin (0,0,0).

Each object in a CASD database is assigned a material property, and each ’material’ is assigned acolour hue from the 0-360° colour circle. Many FLACS users find it convenient to assign certainhues to various structural elements, and the following convention is used by GexCon for typicalprocess facilities.

Hue Colour Description0 Red solid walls and decks30 Orange pressure relief and and

louvred panels60 Yellow grated decks120 Green anticipated congestion180 Cyan equipment200 Light blue structure220 Medium Blue secondary structure250 Dark Blue piping300 Pink equipment

Table 3.2: Colour convention used by GexCon

A standardized colour scheme makes it more straightforward to review geometries from oldprojects.

3.1.11 About congestion, confinement, and vents

In order to have a good representation of the effect of obstacles it is important that they are wellrepresented geometrically by the chosen grid. In most practical situations it will not be possibleto represent the smaller obstacles on the grid, these should still be included since they may betreated by proper sub-grid models. Larger obstacles like the floor (or the ground), the ceiling, thewalls and larger equipment will be resolved on-grid. This means that they will be adjusted tomatch the grid lines.


38 CASD

The most challenging geometry to represent properly is repeated obstacles of the same size andspacing as the chosen grid resolution, in such cases the user should consider to change the grid toachieve a better representation. If this type of geometry is dominant it is of vital importance forthe accuracy of the result that the representation is good enough. In cases where such a geometryis not dominant one may pay less attention to how it is represented. For normal offshore modulesthere will be a range of subgrid sized obstacles which are more or less randomly distributed inspace.

In many experimental setups one will find repeated obstacles of the same size. The basic researchon gas explosions past many years now has focused on the effect of obstacle arrays, perhaps toa greater extent than on the effect of more realistic geometries. Both categories are important inorder to be able to validate tools like FLACS.

It is important to represent the vent openings of a semi-confined geometry properly. If obstaclesclose to the outer boundaries are adjusted to match the grid, the effective vent area may be af-fected. In order to verify that the representation of the vent openings is as good as possible theuser should check the porosity fields (using CASD or Flowvis).

3.2 File menu

3.2.1 New

Shortcut CTRL+N

Starts a new simulation job.

The New command in the File menu creates a new empty job. If there were unsaved changes tothe current job, a dialog box is displayed, asking about saving the changes.

3.2.2 Open

Shortcut: CTRL+O

This command opens an existing set of simulation files. The default selection is defined in a ∗.cajfile.

The Open command in the File menu opens an existing job.

If you enter the file name in the command input field, the path must be encapsulated in apostro-phes, for instance:

∗ open "../../Test/000000.caj"

If you select the command from the menu bar, or if no name is specified in the command inputfield, the Open dialog box is displayed, allowing you to specify a path and file name to open.

By default, the file filter is initiated for selecting CASD job header files (type ∗.caj). But youmay also select a geometry file (type co∗.dat3). CASD will then open all files with the same jobnumber.

If a geometry is open (in the database), the filter string will be constructed from the project andgeometry numbers. It is not possible to open a job that is not compatible with the open projectand geometry numbers.

If there were unsaved changes to the current job, a dialog box is displayed, asking about savingthe changes.

The geometry file is not read when a geometry is open in the database. If no geometry is open


3.3 Geometry menu 39

in the database, CASD will display the contents of the geometry file in the graphic area aftersuccessful open. The contents of the geometry file can be edited using the Edit File command inthe Geometry menu, see section Geometry menu.

3.2.3 Save

Shortcut: CTRL+S

Saves the current simulation job (i.e. the various files that define the job).

The Save command in the File menu saves the current job.

3.2.4 Save as

Shortcut: CTRL+SHIFT+S

The Save As command saves the current job under a new (user-defined) name (job number).

3.2.5 Import

Imports certain specifications from another simulation job (e.g. grid file, scenario file, etc.).

3.2.6 Exit

Shortcut: CTRL+Q

Exits the CASD software.

3.3 Geometry menu

CASD stores the geometry in a database, and on the geometry file (co-file). The commands inthe Geometry menu in the main window, except the Edit File command, are available whenconnected to a database. The Save and Save As commands in the File menu writes the geometryto the geometry file.

The building blocks in a CASD geometry are instances of objects. Objects can be global or local.Several geometries can contain instances of the same global object, while a local object only canbe included in the geometry where it was created.

Instances can be grouped under assemblies. Several levels of assemblies can be created. Eachinstance and assembly has a transformation matrix. The position, scale, and orientation of aninstance is the result of the matrices on all levels above the instance, in addition to the matrix forthe instance itself.

Each geometry is a member of a project. The project is the top level in the CASD data structures.A project can own a number of geometries.

Instances and assemblies can be made invisible and visible using the following commands:

CTRL+I Make the selected assembly/instance invisible

CTRL+SHIFT+I Make the selected assembly/instance visible.


40 CASD

Use the Position command in the Geometry menu to change the position of the selected assemblyor instance.

3.3.1 Geometry Database

The first option on the Geometry menu in CASD opens the Database dialog box.

Figure 3.3: The geometry database window in CASD

In the Database dialog box the user can:

• Create a new database, project, geometry, or object.• Connect to or save an existing database.• Open or save existing, projects, geometries, or objects.• Insert instances in a geometry.• Define new materials or edit existing materials.

3.3.1.1 Geometry tab

On the Geometry tab the user can create, open and manipulate projects and geometries. Projectscan be renamed and deleted, geometries can be renamed, copied and deleted.

3.3.1.2 Objects tab

The New Object button in the Database dialog box opens the Object window.

3.3.1.3 Materials tab

Each object in a CASD database is assigned a material property, and each ’material’ is assigneda colour hue from the 0-360° colour circle. To define a new material click the New Materialbutton. The new material is defined by a name and a hue, a value between 0 and 360.



3.3.2 Creating a CASD database

To create a database choose Geometry→Database or type ∗ geometry database. The Geome-try Database window is shown. Click the Connect button. A file selection dialog box is displayed.Move to the directory where the database should be created, and write the name of the database,e.g. my_database.db. Alternatively the database can be created using the command input: ∗database create my_database.db, which will create a database in the current directory.

If the Geometry Database window is not open, choose Geometry→Database. Use the New Projectbutton to create a new project, or the Open Project button to open an existing project.

When a project is opened, a new geometry can be created clicking the New Geometry button, oropen an existing geometry clicking the Open Geometry button.

When an existing geometry is opened, the assembly/instance structure and all objects andmaterials used are loaded into the CASD program. If the geometry contains many assem-blies/instances, you may get an error message indicating that there were not room enough inthe CASD data structures. See section CASD command line options for information on how youcan use command line options to allocate more memory for these structures.

3.3.3 Connecting to a database

To create a new database, see section Creating a CASD database.

To connect to an existing database choose Geometry→Database or type ∗ geometrydatabase. The Geometry Database window is shown. Click the Connect button. A file se-lection dialog box is displayed. Select the CASD_DB file on the database directory you want toconnect to.

If you enter the file name in the command input field, the path must be encapsulated in apostro-phes, for instance:

∗ database connect "MyCasdDB/CASD_DB"

3.3.4 Creating a new or opening an existing object

You can create a new object clicking the New Object button on the Objects tab in the GeometryDatabase window, or open an existing object using the Open button.

When you have completed the New or Open Object command, the object window is displayed.

3.3.5 Selecting a node and a subtree

At any time, a part of the binary tree is selected. It may be a single node, or a subtree containingseveral nodes. If a subtree is selected, the top node is referred to as the selected node. In thepostfix string, the top node is the rightmost node in the subtree.

The selected subtree is highlighted in the graphic window, and underlined in the message area.There are two different methods for selecting a subtree.

1. Click MOUSE+LEFT while pointing at a primitive. If several primitives are hit, they areplaced on a stack (list). Only one primitive is selected at a time. Press CTRL+TAB commandto parse this stack.

2. Use the following commands:


42 CASD

CTRL+L Select the previous instance

CTRL+R Select the next instance

3.3.6 Maintaining a CASD database

The dbfutil program is available for creating and maintaining CASD file databases.

Linux:

run9 dbfutil database command [option]

Windows:

dbfutil database command [option]

The usage of this program is described in table Using the the dbfutil program. Make sure that noother users are connected to the database when you execute this program.

Command Descriptioncreate Create databasedestroy Destroy databaseforce Destroy database, override any errorsdellock Delete all locks. Use this command if files in

the database are still locked after a crash inCASD

restoredep Restore dependencies. For each object in thedatabase, there is a file containing a list of allgeometries that contain instances of theobject. (Executing the Information commandin the File menu in the Object dialog lists thecontents of this file.) This file is used fordetermining if the object can be deletedwhen you execute the Delete Objectcommand in the Database menu. CASDupdates these files when required. But if aproblem should occur for some reason, therestoredep command might help. It updatesthe file mentioned above for all objects in thedatabase.

restorehead Restore header files. This command resetsthe process log file for the database. This filecontains a list of (CASD) processes currentlyconnected to the database.



list List the content of all table files, e.g. list Olists all objects:

P List the content of all project table files.

O List the content of all object table files.

M List the content of all material table files.

G List the content of all geometry table files.

L List the content of all local object tablefiles.

U List the content of all objects-used tablefiles.

A List the content of all asis table files.

Table 3.3: Using the dbfutil program

We strongly recommend that you make backups of your databases on a regular basis.

3.3.7 Local objects

Local objects consist simply of one box or one cylinder. Use local objects to define entities likewalls, floors etc. Define global objects for more complicated things.

The name of a local object must start with an underscore character (_).

The Local Object command in the Geometry menu creates a local object, and one instance of it.You can of course create several instances of the local object using the Instance command.

The Local Object command has two sub choices, Box and Cylinder. Select the appropriate primi-tive type.

CASD will first ask for the material name. Enter the name of an existing material. The materialdecides the colour of the object. If you haven’t defined any materials, use the New Materialcommand in the Geometry Database window to create one.

CASD will then ask for the sizes and porosities for the primitive. CASD creates an instance of theobject in (0, 0, 0). Use the Position or Translate command to move it to the correct position.

You can use the Properties command to edit material, sizes and porosities for a local object. TheRename command changes the name of the object.

3.3.8 Global objects

A global object is edited in a separate object window. All the commands described in this chapterrefers to the menus in the object window.

Global objects can have instances in several geometries. The structure within a global object is aconstructive solid geometry (CSG) model where simple solid primitives are combined by meansof Boolean set operations. The primitives and operations are nodes in a binary tree where theleaves are primitives and the internal nodes are operations.

Boxes, cylinders, ellipsoids, general truncated cones (GTC) and complex polyhedrons (CP8) are


44 CASD

the primitive types supported. The box primitive includes planes as a special case. Availableoperation types are union and difference.

Warning:

Only boxes and cylinders should be used in by default, but ellipsoids, general truncatedcones and complex polyhedrons can be used in special cases. These latter primitive typeshave the following important limitations:

• No subgrid models, thus not contribution to turbulence and drag force• Porosity calculation takes a long time for these primitive types. There should be no

more than 100-200 of these primitives in any given geometry

Figure 3.4: Supported primitive types

A root is a subtree that is not part of another subtree. The object typically contains several rootsduring editing. But it must contain only one root when it is saved.

The postfix string represents a way of visualising the binary tree defining the object.

The postfix string for the open object is displayed in the message area in the object window. Theselected subtree is highlighted.

A material is assigned to each object. The material decides the colour of the object.



Figure 3.5: The binary tree for an objects, and the corresponding postfix string

3.3.9 Assembly

Opens a dialog box where the user can specify an assembly of several instances.

3.3.9.1 Adding an assembly

Assemblies represents a way to group the instances in complicated geometries. The Assemblycommand in the Geometry menu adds an assembly to the geometry. CASD will ask for theassembly name. You must enter a name that doesn’t exist on the same level, see below. Theassembly is placed in (0, 0, 0). You can transform an assembly in the same way as an instance.

All geometries contains at least one assembly, called the top assembly. That assembly can not bedeleted.

When you create an assembly, it is placed in the geometry structure depending on what wasselected on forehand. If an instance was selected, the new assembly is placed after that instanceunder the same assembly. If an assembly was selected, the new assembly is placed under thatassembly.

You can later rename the assembly using the Rename command.

3.3.9.2 Selecting an assembly or instance

The selected instance, or all the instances in the selected assembly, are highlighted in the graphicwindow. The name of the selection is written in the message area. The name is concatenatedfrom the geometry name, the names of all assemblies above the selected assembly/instance, andthe name of the selected assembly/instance. Each level is separated by a period (.). An exampleis shown below.

Current Geometry Selection: M24.A1.COOLER-2


46 CASD

Here, M24 is the geometry name, A1 is an assembly. The last part of the string is the lowestlevel. In this example, it is an instance, identified by the object name, COOLER, and the instancenumber.

There are three different methods for selecting an assembly or instance. The first method is toselect from the graphic window. To do this, click MOUSE+LEFT while pointing at the instance. Ifseveral instances are hit, they are placed on a stack. Only one instance is selected at a time. InCASD4 use the CTRL+TAB command to parse this stack.

The second method is to use the following commands:

• Select the parent assembly: Press CTRL+U.• Select the child assembly/instance: Press CTRL+D.• Select the assembly/instance name: Press CTRL+F. You are asked to enter the concatenated

name to select.• Select the previous assembly/instance on the same level: Press CTRL+L.• Select the next assembly/instance on the same level: Press CTRL+R.

The third method is to use the List command in the Geometry menu to pop up a list of thecontents of the open geometry. You can use the mouse to select from the list.

3.3.10 Instance

Creates an instance in the current geometry and/or assembly.

3.3.10.1 Adding an instance

To add an instance of an object, use the Instance command in the Geometry menu. CASD willask for the object name. You must enter the name of an existing object. The instance is placed in(0, 0, 0). Use the Position or Translate command to move it to the correct position.

Alternatively the Instance button on the Objects tab in Geometry Datbase dialog can be used.

When a new instance is created, it is placed in the geometry structure depending on what wasselected on forehand. If an instance was selected, the new instance is placed after that instanceunder the same assembly. If an assembly was selected, the new instance is placed under thatassembly.

3.3.11 Local object

Creates a local object in the current geometry.

3.3.12 Delete

Deletes either the currently selected instance, local object, or the current assembly (must beempty).

3.3.13 List

Lists all assemblies and instances in the current geometry, including modified positions.



3.3.14 Duplicate

Duplicates the selected instances in the current geometry.

3.3.15 Position

Defines the position of an instance.

3.3.16 Translate

Translates the current instance.

3.3.17 Rotate

The Rotate command rotates the selected assembly or instance. Note that CASD only accepts axisparallel geometry. That means that the rotation angle must be a multiples of 90 degrees.

3.3.18 Scale

Scales the current instance by a certain factor in each spatial direction

3.3.19 Matrix

Specifies the transformation matrix of the current instance. This command is normally not useddirectly, but is available for macro reading and writing.

3.3.20 Making an assembly or instance visible or invisible

Shortcut: CTRL+I CTRL+SHIFT+I

This command lets the user make the current instance invisible/visible.

3.3.21 Select

Selects an instance in the current geometry through the following short cut options. See sectionSelecting an assembly or instance.

3.3.22 Substitute

Substitutes all instances of one object in the current geometry with instances of another object.The user specifies the name of the existing object and new objects.

3.3.23 Properties

Opens a dialog box where the user can observe and edit the properties of a local object.


48 CASD

3.3.24 Rename

Opens a dialog box where the user can rename assemblies or local objects.

3.3.25 Object

Opens the currently selected object.

3.3.26 Edit file

The Edit File command in the Geometry menu makes it possible to edit the geometry file (co file)for the open job. This command is only available when no geometry is open in the database.

The geometry is saved on the geometry file as one single object, when selecting Save in CASD.Upon the Edit File command, an object window is therefore shown for editing this object, ifthe geometry database is not available, or the user wants to make small modifications to thegeometry outside of the database.

Since the object structure lacks the assembly/instance mechanism, editing the geometry file di-rectly without using the database is recommended only for geometries with a relatively smallnumber of primitives. For geometries with many primitives, the postfix string is long and diffi-cult to manage.

Editing the geometry file for FLACS simulations may be advantageous when the user want totest the impact of small changes in the geometry on the simulation results. Note that there is noway to update the database from the geometry file.

3.4 Object window in CASD

The object window opens from the ’New Object’ button in the database dialog box.

The object window opens from the database window.


3.4 Object window in CASD 49

Figure 3.6: The object window in CASD

The message area in the object window shows the postfix string.

3.4.1 File menu in the Object window

The options on the file menu in the object window are explained below.

3.4.2 Save

If the user is editing an object in the database, the Save command in the File menu saves the objecton the database.

If the user is editing the geometry file, the changes are stored internally in the geometry database,and will be written to the file upon the Save and Save As commands in the File menu in the mainwindow. Exiting from the object window without saving, the changes are lost.

The object is stored only if it is consistent, that is if it has only one root. If the object is notconsistent, an error message is displayed, and a Union or Left Difference should be added.

3.4.3 Information

The Information command in the File menu displays a list of all geometries containing instancesof the open object.


50 CASD

3.4.4 Exit

Upon the Exit command, CASD asks about saving the object, and then whether to exit from theobject. If the answer is yes to the last question, the object window is closed.

3.4.5 Edit menu

The options on the edit menu in the object window are explained below.

3.4.5.1 Operations

The Operation command in the Edit menu changes the operation type if the selected node is anoperation.

3.4.5.2 Properties

The Properties command in the Edit menu changes the primitive properties if the selected node isa primitive.

If you have selected a subtree containing only one type of primitives, the Properties command canbe used for changing one or more parameters for all these primitives.

3.4.5.3 Translate

Use the Translate command to translate the selected assembly or instance a specified distance ineach axis direction.

Use the Translate command in the Edit menu to translate the selected subtree a specified distancein each axis direction.

3.4.5.4 From To

Use the From To command to translate the subtree so that one specified position, the base point,is moved to another, the target point. A dialog box for specifying the two positions is displayed.A circle is displayed in the graphic window, indication the position being edited. CASD keeps alist of positions used in the object. By pressing CTRL+L or CTRL+R, you can parse this list. Thecoordinates in the dialog box is updated.

3.4.5.5 Rotate

The Rotate command rotates the selected subtree. You must specify a base point for the rotation,and the rotation angle. As for the From To command, you can parse the position list using theCTRL+L or CTRL+R commands. Note that CASD only accepts axis parallel geometry. That meansthat the rotation angle must be a multiple of 90 degrees.

3.4.5.6 Scale

The Scale command is only legal when an instance of a local object consisting of a box is selected.



The Scale command scales the selected subtree. You must specify a base point for the scaling,and the scaling factor. You can parse the position list using the CTRL+L or CTRL+R commands

3.4.5.7 Delete

The Delete command in the Edit menu deletes the last node in the postfix string, the selectedsubtree or the current root. Note that if the postfix string for the object is consistent, it consists ofonly one root. Therefore deleting the current root deletes the entire object.

3.4.5.8 Mark

The Mark command is used in connection with the Substitute command. Select Mark commandto mark the subtree to be substitued with the subtree selected when the Substitute command isselected.

3.4.5.9 Substitute

The Substitute command in the Geometry menu substitutes all instances of one object with in-stances of another object. You are asked to specify the two object names.

The Substitute command in the Edit menu substitutes the selected subtree with another subtree.Use the Mark command to select the first subtree.

The substitute command implies the following steps. (Let subtree 1 denote the first subtree andsubtree 2 the second subtree.)

1. Make a copy of subtree 2 and give it a new identity, say subtree 3.2. Delete subtree 1 from the postfix string.3. Insert subtree 3 in the postfix string in the position where subtree 1 was situated.

3.4.5.10 Duplicate

The Duplicate command in the Geometry menu duplicates the selected instance. You are askedto enter the number of copies, and the distance between each copy in the three axis directions.Click on Ok, and a dialog box pops up for each copy, allowing you to edit the position.

The Duplicate command in the Edit menu duplicates the selected sub tree. You are asked to enterthe number of copies, and the distance between each copy in the three axis directions. Unionoperations are added automatically, so that the resulting sub tree includes the original one.

Creating pipe bundles Start with creating one cylinder with the appropriate diameter, lengthand direction. Use the Duplicate command in the Edit menu to duplicate the cylinder in onedirection. Use the same command once more to duplicate the resulting row of cylinders in theother direction.

If you need to change some parameters for all the cylinders, select the entire pipe bundle sub treeand use the Properties command. If you want to change the distances between the cylinders, thiscan be done by scaling the entire sub tree. Afterwards you can use the Properties command toreset the cylinder diameters and lengths.


52 CASD

3.4.5.11 Material

The Material command in the Edit menu edits the material name for the object. You must enteran existing material name.

3.4.5.12 Matrix

The Matrix command was introduced to make it simple to create and run macros for creatinggeometries.

Warning:

This command should normally not be used in interactive mode.

3.4.6 Add menu in the Object window

The options on the add menu in the object window are explained below.

3.4.6.1 Box

The Box command in the Add menu adds a box at the end of the postfix string. A dialog box fordefining the box parameters is displayed.

3.4.6.2 Cylinder

The Cylinder command adds a cylinder at the end of the postfix string. A dialog box for definingthe box parameters is displayed.

3.4.6.3 Ellipsoid

The Ellipsoid command adds an ellipsoid at the end of the postfix string. A dialog box for definingthe ellipsoid parameters is displayed. Note warning about the use of ellipsoid.

3.4.6.4 CP8

The CP8 command adds a complex polyhedron at the end of the postfix string. A dialog boxfor defining the complex polyhedron parameters is displayed. Note warning about the use ofcomplex polyhedron.



Figure 3.7: Definition of a complex polyhedron

3.4.6.5 GTC

The GTC command adds a general truncated cone at the end of the postfix string. A dialog boxfor defining the general truncated cone parameters is displayed. Note warning about the use ofgeneral truncated cone.

Figure 3.8: Definition of a general truncated cone

3.4.6.6 Union

The Union command adds an union operation at the end of the postfix string. This command isonly legal if the object contains at least two roots which can be connected by the operation.

3.4.6.7 Left Difference

The Left Difference command adds a difference operation at the end of the postfix string. Thiscommand is only legal if the object contains at least two roots which can be connected by theoperation. If using CASD4, use the Shade command in the View menu, to see the result of theoperation. Note that the right hand side operator of a difference operation must be a primitive.

3.4.6.8 Copy

The Copy command adds a copy of the selected sub tree at the end of the postfix string.


54 CASD

3.4.6.9 Object

The Object command adds a copy of a specified object at the end of the postfix string.

3.4.7 Select menu in the Object window

The options on select file menu in the object window are explained below.

3.4.7.1 Previous

Shortcut: CTRL+L

Selects the previous primitive or subtree

3.4.7.2 Next

Shortcut: CTRL+R

Selects the next primitive or subtree

3.4.7.3 Stack

Shortcut: CTRL+TAB

This command will parse (cycle through) the list of selected primitives or subtrees if more thanone is selected.

3.4.8 View menu in the Object window

The options on the view menu in the object window are explained below.

3.4.8.1 Print

The Print menu allows exporting a screenshot of the CASD window into different formats:

• Postscript• RGB• IV

3.4.8.2 Examiner Viewer and Fly viewer

The default and most widely used viewer is the Examiner viewer. The Fly viewer can be used tofly through the geometry.

3.4.8.3 The XY, XZ and YZ views

The option XY View and XZ View display a projection of the geometry in the XY and XZ planesrespectively. The options YZ East View and YZ West View display a projection of the geometryin the YZ plane along the positive and negative Y-axis respectively.



3.4.9 3D View

The 3D View option displays a default 3D view of the geometry.

3.4.9.1 Axis

The Axis option turns axis display on and off.

3.4.9.2 Maximize

The option Maximize maximizes the visible window to display the entire geometry and grid.

3.4.9.3 Grid Display

Three different options are available in the Grid Display menu:

• Off: The grid is not displayed. Only the geometry would be displayed.• Working Direction: The grid would be displayed in the working direction only.• All Directions: The grid would be displayed in the three directions.

3.4.9.4 Annotation

The options in this menu are currently not used.

3.4.9.5 Draw Style

Different options are available in this menu:

• Off: The geometry will not be displayed.• Wireframe: Only the edges of the objects that compose the geometry would be displayed.• Filled: Surfaces of the objects that compose the geometry would be displayed.• Scenario Wireframe: Only the edges of scenario objects (for example, a fuel region) would

be displayed.• Scenario Filled: Surfaces of scenario objects would be displayed.

3.4.9.6 LOD and Properties

The LOD (Level Of Details) and properties menus control the details of the geometry displayed.

3.4.10 Macro menu in the Object window

The options on the macro menu in the object window are explained below. for more infromationabout CASD macros see section Macro menu.

3.4.10.1 Run

Read a macro file defining a single object.


56 CASD

3.4.10.2 Record

Writes all commands given to a user specified macro file.

3.4.10.3 Write Object

Writes a macro file containing the current object.

3.5 Grid menu

The simulation volume is divided into a set of control volumes by three sets of grid planes, onein each axis direction.

There is always a current grid working direction, and a selected region of grid planes in thisdirection. The current working direction is shown in the message area. The lines indicating theselected region is highlighted.

3.5.1 Simulation volume

The Simulation Volume command lets you change the simulation volume extent in all three di-rections. If you increase the volume, the original grid planes are kept, but one additional plane isadded in each direction. If you decrease the volume, planes outside the new volume are deleted,and new planes are created on the volume borders.

3.5.2 Direction

The Direction command changes the working direction. Legal input is x, y or z. The Grid menucommands Region, Add, Position, Move, Delete, Smooth, Stretch and List affects the grid planesin the working direction.

3.5.3 Region

The Region command substitutes the selected grid planes by a new set of grid planes. CASD asksyou to enter the new number of control volumes in the region.

3.5.4 Add

The Add command adds a new grid plane in the working direction. You are asked to enter thecoordinate value for the new plane.

3.5.5 Position

The Position command lets you edit the position for the selected grid plane.


3.5 Grid menu 57

3.5.6 Move

The Move command moves the selected grid planes a specified distance.

3.5.7 Delete

The Delete command deletes the selected grid planes.

3.5.8 Smooth

The Smooth command substitutes the selected grid planes by a new set of grid planes.

For the Smooth command, the sizes of the control volumes at each end of the region is keptunchanged. The sizes of the control volumes between them varies gradually.

This function is typically used when refining the grid around a leak.

3.5.9 Stretch

The Stretch commands substitutes the selected grid planes by a new set of grid planes. This isparticularly useful when stretching the grid towards the outer boundaries.

The Stretch command has two sub-choices:

• Negative Direction (typically used at the boundaries at the negative end of the axis)• Positive Direction (typically used at the boundaries at the positive end of the axis)

You must enter the size of the control volume at one end of the region, default is the current size.Then you must enter a factor by which the sizes of the control volumes in the specified directionincreases/decreases.

Attention:

Note that stretching of the grid should be avoided in areas of the simulation domain wherethe main combustion is happening. The flame model in FLACS has been validated for cubicalcontrol volumes, thus the user should not stretch the grid in areas where accurate results arerequired. It is however good practice to stretch the grid towards the boundaries, to concervesimulation time and computer memory.

3.5.10 Information

The Information command displays status information about the defined grid, while the Listcommand lists the grid coordinates in the working direction.

3.5.11 List

The Information command displays status information about the defined grid, while the Listcommand lists the grid coordinates in the working direction.


58 CASD

3.5.12 Display

The Display command turns grid display off, displays the grid in the working direction only, ordisplays the grid in all three directions.

3.5.13 Select

The selected region of grid planes is limited by two planes, the lower and upper limit. If onlyone plane is selected, the upper and lower limit is the same grid plane. Grid planes are selectedusing the following commands:

• Lower boundary

– Select the next grid plane: CTRL+RIGHT– Select the previous grid plane: CTRL+LEFT

• Upper boundary

– Select the next grid plane: CTRL+UP– Select the previous grid plane: CTRL+DOWN

3.5.14 Grid-related operations

3.5.15 Importing the grid from another job

Use the Import command in the File menu to import the grid from another job.

If you enter the grid file name in the command input field, the path must be encapsulated inapostrophes, as described in section . If you select the command from the menu bar, or if noname is specified in the command input field, the Import dialog box is displayed, allowing youto specify the path and file name for a grid file. You will be asked to verify that the current gridis overwritten by the grid from the specified file.

3.5.16 Saving the grid

The Save and Save As commands in the File menu saves the grid, together with the rest of thejob data. If the grid is changed, you will need to recalculate the porosities.

3.5.17 Grid-related utilities

FLACS is deleivered with a command line tool for creating an manipulating the grid. This toolcan be used to quickly edit or get information about the grid. Please see section gm for furtherinformation.

3.5.18 Grid guidelines

The grid resolution should be chosen to obtain a simulation result within an acceptable timeframe. In most cases a reasonably good result can be obtained on a coarse grid within less thanone hour (in some cases 5 minutes), and high quality results can normally be generated in a fewhours (or at least over the night).


3.5 Grid menu 59

Never start a project with a calculation on a grid that will be running for days. If such long sim-ulation times are necessary, always start simulating on a much coarser grid [even if this violatesguidelines] to check that the scenario and setup are OK.

The user should keep the position of the grid lines in mind while defining the geometry. Thegeometry details such as walls and decks should be adjusted to the closest grid line when in-putted. Thereby the user keeps track of the positioning instead of having the geometry moved inan unwanted direction by the porosity calculation program.

In the grid embedding process, it is highly recommended to use Grid→Information in Casd tocheck different aspects of the grid. Grid sensitivity tests are also recommended.

3.5.18.1 Gas explosion simulations

Attention:

The user should always apply cubical grid cells in the combustion region. Deviations fromthis will give different flame propagation and pressures, and the validation work done isno longer valid. Deviations of the order 10% in aspect ratio is OK, deviations by a factorof 2 in aspect ratio is not OK. If one chooses not to follow this guideline, the results can besomewhat improved by setting a fixed control volume size for the time stepping routine (seesection The SETUP namelist, example TIME_STEPPING=" STRICT:L_FIX=1.0" ).

Channels and confined vessels and rooms (filled with gas from wall to wall) must always beresolved by a minimum of 5-6 grid cells in smallest direction if flame acceleration shall be modelled.This also applies for pipes where flame acceleration along the pipe is of interest.

A pipe connection from one vessel to the next may have less grid cells across the diameter (butpreferably more than 1 CV) if only flame transport by pressure difference and not flame accelera-tion along the pipe shall be modelled. Increase the inner diameter of angles and bends somewhatwhen modelling pipes with cylinder minus primitives. Remember that one full grid cell is re-quired inside the solid walls around " minus primitive holes" to ensure that the walls will not beleaking.

Unconfined gas clouds as well as partially filled clouds should have a minimum of 13 grid cellsacross the cloud if both sides are unconfined, and a minimum of 10 grid cells in directions wherecloud meets confinement on one side (example vertical direction for dense gas cloud in chemicalplant).

It is not recommended to use non-cubical grids for explosion simulations. As they are often usedfor dispersion simulations, the dispersion simulation results should be dumped, thereafter therdfile utility program should be used to transfer the results from the dispersion grid to a gridbetter suited for explosions, see example below:

> run9 rdfile rd111111.n001 rd222222.n001

Here 111111 is the dispersion calculation job number and 222222 is copy of the job, in which thegrid has been modified to follow explosion grid embedding guidelines. The grid of job 222222must be completely inside the grid of 111111.

The grid can be stretched outside the combustion region in directions where pressure recordingsare not of interest. In directions where pressure wave propagation is of interest, one should notstretch the grid because this will reduce the sharpness and quality of the pressures.

A proper distance to external boundaries is important. At least 5-10 grid cells from vent openingto external boundary should be used in situations where the external explosion is not important(small vent area or strong turbulence inside vessel).


60 CASD

In low-congested vessels with significant vent opening, external explosions and reversing of flowmay give a strong feedback into the vessel in connection with venting. To pick up this properly,the distance to the external boundaries should be significant (maybe 3-4 times the length of thevessel).

EULER boundary will reflect negative pressures, which can destroy results when simulating farfield pressure propagation. In this case PLANE_WAVE may be recommended (but then the bound-aries must be far away so no products from combustion reaches the boundaries). For unconfinedsituations, try to have the same distance to boundaries in all directions (use stretching in direc-tions with less interest in results).

3.5.18.2 Blast wave propagation in the far field

Maximum control volume size should be 10% of gas cloud diameter:

max CV = 0.1× (gascloudvolume)1/3 (3.1)

The grid cells must be approximately cubical in explosion simulations and the cell size should bemaintained in directions of interest for blast propagation simulations. For vessel burst the sameguidelines applies. If the pressure is much higher than 10 barg, somewhat larger grid cells thanthis criterion can be acceptable. Remember to use PLANE_WAVE boundary condition and properdistance to the boundaries.

3.5.18.3 Dispersion simulations near field

Calculations of flammable gas requires grid refinement near the leak. The area of the expandedjet (at ambient pressure) must be resolved by one grid cell (ACV < 2 × Ajet) except for low-momentum releases of highly buoyant gases such as hydrogen where the guideline (ACV <1.25× Ajet) should be followed. In most cases, a grid refinement near the leak helps in keepingmoderate calculation times while getting acceptable results.

Grid refinement guidelines for efficient simulation of high velocity jets recommend only refine-ment across jet direction (not along). CFLC should be increased by refinement factor (CFLVshould not be changed). Smoothing from fine grid cells near jet to normal grid cells furtheraway from jet is recommended.

If the jet is not along the axes, is impinging or has a low momentum with positive/negativebuoyancy, then extending the refined region of the grid in one or more directions may be re-quired. Refinement along the jet may then also be required.

If only far field concentrations are of interest, the refinement near the leak may not be needed.Quicker calculations and less stability problems will be seen without the refinement.

For dense gas calculations, it may be a good idea to use a finer resolution in the vertical directionnear the ground than in the other two directions. Increase CFLC by this refinement factor.

Outside the main area of interest, further stretching to the boundaries is recommended to min-imize the influence of the boundaries. If in doubt whether the distance to the boundaries in-fluences your results, increase the distance further to check the sensitivity. The general recom-mended procedure for setting up the grid is:

1. Cover the computational domain with a uniform grid2. Refine the grid in the near region of a jet (perpendicular to the jet axis)3. Stretch the grid outside the main region towards the boundaries


3.6 Porosities menu 61

3.6 Porosities menu

The Porosities menu supplies commands for calculating and verifying porosities.

Calculate This command starts the porosity calculation program, Porcalc

Verify This command starts Flowvis for porosity verification

3.6.1 Calculate

The Calculate command starts the porosity calculation program, Porcalc.

The version of Porcalc can be selected on the Preferences dialog in CASD. The default versionof Porcalc uses a resolution of 64, i.e. the control volume is divided into 64 parts when calculat-ing the porosities. It can be useful (and necessary) to use the porcalc_16 version if the porositycalculations are very slow, e.g. if the geometry contains slanted primitives.

If you are working with a multiblock simulation, only the porosities for the selected block arecalculated. Upon the Save and Save As command in the File menu, CASD will warn you aboutblocks where the grid and/or geometry is changed since the last time you calculated the porosi-ties.

Porcalc can also be started from the command line or the Runmanager.

3.6.2 Verify

The Verify command lets you view the calculated porosities. Flowvis is started for porosity ver-ification. A 2D Cut Plane plot for the appropriate job is automatically created with volume andarea porosities shown. The Plot Domain dialog box pops up. This dialog box lets you select otherplanes as wished.

By clicking inside a control volume, you can verify the porosity values for that volume.

If you are working with a multiblock simulation, only the porosities for the selected block areverified.

3.7 Scenario menu

The purpose of this chapter is to outline how to edit the scenario sections. A short description ofeach section and the impact on the FLACS simulations will be given. The scenario-file (cs-file) isan ASCII file and it is easy to edit manually as well as using CASD.

The items in the Scenario menu are read from a scenario definition file. There is one default sce-nario definition file, but several other choices can be activated by changing the scenario template.This can be done using the Preferences command in Options menu. For instance, the default+1template activates several advanced options, especially in the Simulation and Output Control sec-tion. The current job should then be saved, closed, and reopened in order for the new options tobe available.

When a section has been selected, the items in the section are displayed as a list. The method forediting a scenario section depends on the type of section.

Sections such as INITIAL_CONDITIONS and IGNITION contain a list of parameters, each withone or more values. A parameter is selected for editing by clicking on it, or by typing the param-eter name in the command input field in the main window.


62 CASD

Sections such as SINGLE_FIELD_SCALAR_TIME_OUTPUT and SINGLE_FIELD_3D_OUTPUTcontain a list of items, which can be selected. For some sections, each item has a subsection.For SINGLE_FIELD_SCALAR_TIME_OUTPUT the user must select from a list of monitor pointsfor each variable selected. An item is selected by clicking on it, or by typing the item name in thecommand input field in the main window. To select several items using the mouse apply CTRLor SHIFT keys. If a selected option shall be deselected without selecting an alternative option,deselect by clicking while pressing the CTRL key. Typing an item name in the command inputfield in the main window selects/deselects the item if previously not selected/selected.

Importing the scenario from another job

Use the Import command in the File menu to import the scenario from another job. The user willbe asked to verify that the current scenario is overwritten by the scenario from the specified file.

Saving the scenario

The Save and Save As commands in the File menu saves the scenario (together with the rest ofthe job data).

3.7.1 Monitor points

Monitor points are user defined locations in the simulation domain where one or more variablesare to be monitored during the simulation. The maximum number of monitor points allowed iscurrently 1000. Positions for monitor points are given in the unit selected in Options Preferences(normally meter)

When the user has defined all the desired monitor points, he/she may specify a list of variablesto be monitored and the relevant monitor numbers for each variable (see the next subsection).

FLACS identifies the 8 surrounding control volume centres and writes an interpolated value ofthe specified variables to the scalar-time output file (nodes on other side of wall or zero porositywill not be used when interpolating).

Attention:

The user should avoid putting monitor points exactly on a grid line or within fractions of agrid cell size from a wall. Usually it is best to enter monitor points according to the grid, notaccording to the geometry. E.g. if monitor points are placed on each side close to a solid wallthey may not necessarily be in two different control volumes (as was the intention). Duringthe porosity calculation the wall will be adjusted to the nearest control volume face (gridline) and might therefore move to the wrong side of a monitor point! It may also be a goodidea to ensure that none of the monitor points are inside fully blocked control volumes.


3.7 Scenario menu 63

Figure 3.9: Specification of Monitor Points


64 CASD

Figure 3.10: How to position the monitor points

3.7.2 Single field scalar time output

As indicated in the previous sub-section, the user must define all the monitor points and panelsbefore he/she can specify the list of output variables. For each output variable, the user may enterone or more numbers indicating the monitor point number(s), or panel number if it is a panelaveraged output variable, for which you want this variable to be measured. An example froma scenario-file section is shown below [NP identifies variable P (pressure) whereas NPP identifiesvariable PP (panel pressure) which gives average pressure load across a surface described bypanel]:

SINGLE_FIELD_SCALAR_TIME_OUTPUTNP 1 2 3 4 5NPP 1 2 3

This shows that pressure is reported for monitor points 1-5 and panel pressure is reported forpanels 1-3 (The definition of measurement panels is described in the section on Pressure ReliefPanels).

To select more than 1 monitor point use CTRL+MOUSE+LEFT or CTRL+SHIFT+MOUSE+LEFT, orsimply drag the mouse over the list of monitor points while pressing LEFT. If a variable and mon-itor points are selected by a mistake, these can be deselected by using the CTRL+MOUSE+LEFT todeselect the last monitor point of a variable.



The most commonly used monitor point variables for explosion simulations are pressure (P),dynamic pressure (DRAG), panel pressure (PP), pressure impulse (PIMP) and sometimes flowvelocity (UVW). For dispersion calculations volume gas concentration (FMOLE) and flow velocity(UVW) are among the most commonly used variables.

The monitor point results for job 010100 are written to the r1010100.dat3 file which can be readby Flowvis. If ASCII data is required, the r1file-utility program can be used (see the section onFLACS utilities).

The first section on the scenario file defines the names and identifiers for all the variables whichmay be selected for output from FLACS. In order to select alternative units for certain variables(e.g. psi or kPa for pressure or K for temperature) the scenario-file should be manually edited.The variable pressure is described in the top of the scenario-file as follows:

NP "P " 1 "(barg) " N"Pressure"

An output in a different unit (psig) can simply be obtained by editing this as follows:

NP "P " 1 "(psig) " N"Pressure"

Similar changes can be made for other variables and other units. Please note that the units oftime must always be seconds (however, it is possible to change them to ms in Flowvis).

A complete list of all variables available can be found in section Output variables in FLACS.

3.7.3 Single field 3D output

This is an output facility in FLACS which enables the user to generate plots of the spatial distri-bution of the variables (e.g. cut plane plots and volume plots) at different moments in time. Theuser needs to specify the list of desired variables for SINGLE_FIELD_3D_OUTPUT, an examplefrom scenario-file is shown below (here P, PROD and VVEC are selected in CASD):

SINGLE_FIELD_3D_OUTPUTNPNPRODNVVECNUNVNW

To select more than one variable press the CTRL-key selecting variable 2 and 3 etc.

Please observe that when velocity vectors are selected for output (VVEC), directional velocitiesU, V and W will automatically be selected. These should not be deselected while VVEC remainsselected (this is possible e.g. by manual editing of scenario-file), if this is done very strangeresults will be seen in Flowvis as result file is not consistent with scenario file used by Flowvis tointerpret results.

The user should be aware that this type of output may give very large files (r3-files). If the userwants to save disk space, the number of output variables and the number of time instants foroutput must be limited. The r3-files are binary, if ASCII data is wanted the utility program r3filecan be applied.

The most commonly used variables for explosion modelling are pressure (P), flame (PROD), some-times gas volume concentration (FMOLE), dynamic pressure (DRAG), maximum pressure (PMAX)


66 CASD

and velocity vectors (VVEC). For dispersion FMOLE and VVEC will be the most common variablesto report.

In certain situations the variable PMAX may not be written to even if specified (zero results ev-erywhere). This may happen if the simulation job requires more RAM than allowed (e.g. due toself-defined limits in Linux)

To specify output times DTPLOT and NPLOT will normally be used (see section Simulation andoutput control), and sometimes also cc-files (see section Runtime simulation control file). Tocreate animations it is normally recommended to have plots at 100-200 different moments intime. When creating results files to be used for animations a combination of DTPLOT and NPLOTis usually recommended for explosions, for dispersion only DTPLOT should be used.

Units of output variables can be changed using the technique described in the previous section.

A complete list of all variables available can be found in section Output variables in FLACS.

3.7.4 Simulation and output control

This section describes parameters for general simulation and output control. The defaultscenario-file setup which is suitable for gas explosion simulations is listed below:

TMAX -1LAST -1CFLC 5CFLV 0.5SCALE 1MODD 1NPLOT -1DTPLOT -1GRID "CARTESIAN"WALLF 1HEAT_SWITCH 0

In addition the following entries are available in the default+1 template:

TSTART -1TMIN -1LOAD -1STEP ""KEYS ""

For dispersion simulations, higher value of CFL numbers (20 and 2) are recommended. NPLOTshould be -1 (it has no meaning) and a finite value of DTPLOT should be given.

A detailed description of each parameter is given below.

3.7.4.1 TMAX

This is the maximum time interval (seconds) that the simulation will last. For explosion simu-lations, default value in CASD can typically be used. The default value set by CASD is -1, thismeans there is no maximum time specified, and automatic stop criteria will be applied. Auto-matic stop criteria will usually work well for explosion calculations. The simulation stops 20 %after either

• 90 % of fuel is burnt or pushed out of the domain• 50 % of fuel is burnt or pushed out of the domain (and average pressure becomes negative)



Automatic stop criteria can not be used for dispersion, and will also be less useful for somespecial situations. If pressure If far-field blast pressures are of interest, the automatic stop criteriashould not be used as it may stop the simulation before blast waves have hit their target.

In a gas dispersion simulation TMAX will typically be from a few seconds to a few minutes. Sim-ulation results are not affected by variation of this parameter.

3.7.4.2 LAST

This is the maximum number of iterations allowed for the simulation. The default value set byCASD is -1 that means that there is no limitation to the number of iterations. This value may bechanged if additional control of when to stop a simulation is required but this is generally notused.

3.7.4.3 CFLC

This is a Courant-Friedrich-Levy number based on sound velocity. The value of CFLC connectssimulation time step length to control volume dimension through signal propagation velocity (inthis case the velocity of sound), in the following way:

Each time step length is chosen so that sound waves may propagate only a limited distance,which is the average control volume length multiplied by the value of CFLC. The default valueset by CASD is 5.0. The time step limit imposed by this criterion is normally dominant in the earlyphase of an explosion, when flow velocities and combustion rate are still low (see also CFLV).

Simulation results may change with this parameter. Therefore, it is not recommended to changethis value for explosion simulations as the validation work is nullified. If convergence problemsoccur (a rare occurrence), CFLCmay be reduced by a factor of 2. However, other problems shouldfirst be ruled out. Extreme changes of the CFL numbers (i.e. by an order of magnitude) are neverrecommended for normal simulations.

Note that for multi-block simulations a maximum CFLC=0.5 should be used for the BLASTblocks. It is recommended to use CFLC=0.2 in the BLAST blocks in order to ensure numericalstability and good representation of the blast wave.

For dispersion simulations, a default value of 20 is normally recommended. This can be increasedby the grid refinement factor (if applicable) i.e. if the grid is refined near the leak by a factor of5, a CFLC number of 20∗5 = 100 may be used (a lower value should be used in case of stabilityproblems).

For far-field blast simulations, this should be combined with STEP="KEEP_LOW" in order tokeep the time step short even after the explosion is outside the "core" area (more information isgiven below).

3.7.4.4 CFLV

This is a Courant-Friedrich-Levy number based on fluid flow velocity. The value of CFLV con-nects simulation time step length to control volume dimension through signal propagation ve-locity (in this case the fluid flow velocity), in the following way:

Each time step length is chosen so that the fluid may propagate only a limited distance, whichis the average control volume size multiplied by the Courant number. The default value set byCASD is 0.5. The time step limit imposed by this criterion is normally dominant in the later phaseof an explosion, when flow velocities and combustion rate are high (see also CFLC).


68 CASD

Simulation results may change with this parameter. Therefore, it is not recommended to changethis value for explosion simulations as the validation work is nullified. If convergence prob-lems occur (rare), CFLV may be reduced by a factor of 2. However, other problems should firstbe ruled out. Extreme changes of the CFL numbers (i.e. by an order of magnitude) are neverrecommended for normal simulations.

For dispersion simulations, a default value of 2 is normally recommended (a lower value shouldbe used in case of stability problems).

3.7.4.5 SCALE

This parameter is used to scale all linear dimensions in a scenario. This means that a 10 m longexplosion vessel is calculated as being 20 m long if SCALE is set to 2.0. Positions and sizes ofequipment, gas cloud, ignition region, monitor points, panels etc. are scaled accordingly. TheCASD default value is 1.0. This parameter will influence simulation results, typically explosionpressures increase with increasing scale.

This is practically never used for realistic geometries.

3.7.4.6 MODD

This is a parameter that may be used to determine how often data for scalar-time plots are writtento the results file during a simulation: data are namely stored every MODD timesteps. CASDdefault is set to 1. This variable does not influence simulation results, only the amount of datastored.

This is normally not used in explosion simulations, but a value of MODD=10 (or higher) may beused for long dispersion simulations.

3.7.4.7 NPLOT

This is a parameter that may be used to determine how often data for field plots are written tofile during a simulation: data are namely stored at given fuel levels where NPLOT

is the number of fuel levels equally spaced between zero and a maximum. Fuel level is definedas the current total mass of fuel divided by the initial total mass of fuel. This output mechanismis not active in the case of a gas dispersion simulation (leaks are specified). This variable does notinfluence simulation results, only the amount of data stored.

3.7.4.8 DTPLOT

This is the time interval (in seconds) for field output. This is useful in gas dispersion simulationsand also in gas explosion simulations when frequent output is required. Note that the field outputfile will become very large if DTPLOT is set small. This variable does not influence simulationresults, only the amount of data which is stored.

3.7.4.9 WALLF

This is a control switch that specifies the use of wall-functions in FLACS. Wall-functions are usedto resolve the effect of momentum and thermal boundary layers on the momentum and energyequations in near wall regions. The following choices are available:



• 0 = OFF• 1 = ON

The CASD default value of WALLF is set to 1. This parameter will influence the simulation results.

When WALLF equals 1 wall, functions are employed based on theory explained in [Sand andBakke, 1989]. A slightly modified version of this wall-functions procedure is employed whenWALLF equals 2. However, no validation work for WALLF=2 is available.

The default value of WALLF=1 should always be used.

3.7.4.10 HEAT_SWITCH

This parameter is meant to control the activation of thermal attributes on objects in Flacs.

Default choice is zero (0) as large scale explosions is not much influenced by heat loss. Choos-ing one (1) will let walls and objects have background temperature, and if gas temperaturechanges, some heat transfer into or out of gas will take place. This is useful for small-scaleconfined explosions and dispersion with important heat effects. This can be combined withKEYS="RADIATE=04" in order to activate radiation heat losses (see below).

If heat switch is activated all solid surfaces will now be initialized with ambient temperature (inprevious versions of FLACS a cs [jobno].HEAT file had to be written). Further heat objects can bespecified at different temperatures, see manual.

Two models for radiation heat loss can be activated. One simplified model can be activated usinga "cs \e [jobno].RAD" file, see description in previous FLACS manual. Alternatively a 6-fluxmodel can be activated with the KEYS-string in the scenario-file or setup-file:

KEYS = " RADIATE=04"

This model will calculate gas heat loss (and absorption/scatter) from radiation as well as radi-ation from hot objects around. Walls absorb 100% of the incoming radiation and emit radiationbased on its own temperature. Symmetry planes will reflect 100%.

3.7.4.11 TSTART

Remarks:

Only available in the default+1 template.

This variable makes it possible to specify a start time for simulation (-1 means not applied =>default is zero or time of dump-file). If dump-file exists, one can still adjust TSTART, but theprevious history of the simulation can then not be kept using the KEEP_OUTPUT in a setup-file.

3.7.4.12 TMIN

Remarks:


This variable makes it possible to define a minimum time for simulation. Automatic stop criteriawill not activate before TMIN has been reached (-1 means not applied). This can be useful in blastsimulations.


70 CASD

3.7.4.13 LOAD

Remarks:


This makes it possible to load a dump file directly from CASD (instead of using the cc-file). Thenumber of dump file should be specified here.

3.7.4.14 STEP

Remarks:


It is possibility to give time stepping input (ref. Manual, options for TIME_STEPPING). Theoptions include KEEP_LOW that is recommended for calculations of far-field blast propagation,and effectively prevents time step from growing when explosion calms down. Another option is

STRICT:L_FIX=1.00”

that instructs the simulation to use 1m grid size as basis for timestep (and ignore local grid re-finement). This can be used instead of increasing the CFLC number as a result of grid refinement.

3.7.4.15 KEYS

Remarks:


This provides the option for entering setup-file options directly in CASD, e.g.

RADIATE=04

for enabling radiation heat loss.

3.7.5 Boundary conditions

In the Boundary condition menu, the user must specific boundary conditions for the outer bound-aries of the simulation domain. The lower boundaries in X- Y- and Z-direction are denoted byXLO, YLO and ZLO respectively, and the upper boundaries likewise by XHI, YHI, ZHI. Recom-mended boundary conditions are as follows:

EULER: Euler equations

NOZZLE: Nozzle formulation

PLANE_WAVE: Plane wave condition

WIND: Wind inflow or outflow

SYMMETRY boundary condition is generally not used for realistic geometries. In addition,it is also possible to choose EQCHAR, and BERNOULLI, but these boundary conditions are notrecommended.



Remarks:

• For most explosion simulations, EULER can and should be used.• For wind and dispersion simulations, NOZZLE boundary condition (similar to EULER)

is more robust.• PLANE WAVE boundary condition is recommended for explosion in low confinement

and for far field blast propagation. Boundary must be extended far outside the explo-sion (Flames should not reach boundary).

• Different boundaries do not need to have the same condition.• Boundary conditions try to model what happens beyond the boundary. Except for solid

walls, this is not straightforward. Sometimes the boundary condition will disturb oreven destroy a simulation. Then the user should:

1. Ensure that the chosen boundary conditions are those that fit best to the problem.2. Consider to increase the Simulation volume and move the boundaries to regions

where less steep gradients will cross the boundaries.

The details of various boundary conditions are given below:

3.7.5.1 Euler

The inviscid flow equations (Euler equations) are discretized for a boundary element. This meansthat the momentum and continuity equations are solved on the boundary in the case of outflow.The ambient pressure is used as the pressure outside the boundary. A nozzle formulation is usedin the case of inflow or sonic outflow.

Warning:

EULER boundary condition may give too low explosion pressures in unconfined situations.In such cases, the Simulation volume should be extended and the Plane wave boundarycondition should be applied.

3.7.5.2 Nozzle

A nozzle formulation is used for both sub-sonic inflow and outflow and sonic outflow. Thiscondition is suitable for porous areas with small sharp edged holes or grids (e.g. louvres andgratings). A discharge coefficient is calculated from the area porosity and a drag coefficient.NOZZLE condition has shown to give a bit higher explosion pressures than EULER, but it is morerobust.

Warning:

NOZZLE boundary condition may give too low explosion pressures in unconfined situations.In such cases, the Simulation volume be extended and the Plane wave boundary conditionshould be applied.

3.7.5.3 Plane wave

This boundary condition was designed to reduce the reflection of the pressure waves at openboundaries which occurs when using EULER or NOZZLE. The pressure wave reflection is causedby setting a fixed pressure at the boundary. PLANE_WAVE boundary condition extrapolates thepressure in such a way that reflections are almost eliminated for outgoing waves.


72 CASD

Attention:

The pressure might stabilize at a slightly elevated level after an explosion. For low confine-ment scenarios it is recommended to use Plane wave boundary condition and to extend thedomain such that the total volume is about 100 times larger than the volume of the initial gascloud.

Warning:

In semi-confined situations where the boundaries are close to the vents, PLANE_WAVE shouldnot be applied.

3.7.5.4 Wind

WIND boundary condition models an external wind field. Velocity and turbulence profiles arespecified at the wind boundaries, either by setting some turbulence parameters manually or bychoosing one of the atmospheric stability classes, see Pasquill class . WIND boundary conditionsare particularly applicable to dispersion scenarios. It is possible to apply WIND on both inflowand outflow boundaries and on boundaries where the flow is parallel to the boundary.

Warning:

In cases where a generated internal flow has a strong impact on the boundary flow, e.g. gasexplosions, WIND should not be used.

Figure 3.11: Specification of Wind boundary condition

Wind speed WIND_SPEED, U0, is the velocity on the boundary at a given Reference height. It ispossible to set WIND_SPEED to positive, zero or negative values, but GexCon recommends to seta postive value and use the Wind direction parameter to specify the direction of the wind. In caseof no wind, the user should consider to use another boundary conditions. A uniform velocity



profile is obtained by setting the Reference height equal to zero. Then the total volumetric fluxover the boundary is as follows:

V = U0 ∑n

Anβ2n (3.2)

Wind direction WIND_DIRECTION is a vector and each component may be given a positive,zero or negative value. The sign of this parameter determines the flow direction. A positive valuemeans wind flow in positive direction, that is inflow over the lower boundaries and outflow atthe upper boundaries. Wind at an angle different from axis directions may be specified using theWIND_DIRECTION vector.

Relative turbulence intensity RELATIVE_TURBULENCE_INTENSITY, IT , is the ratio betweenthe isotropic fluctuating velocity, u′, and the mean flow velocity U0:

IT =u′

U0(3.3)

IT will typically have a value in the range 0.0 to 0.1. This parameter is used to calculate the valuefor turbulent kinetic energy, k = 3/2u′2, at the boundary.

Attention:

It is not necessary to set RELATIVE_TURBULENCE_INTENSITY for inflow boundarieswhen a Pasquill class is specified. When a Pasquill class is set, FLACS will automaticallycreate profiles for velocities and turbulence parameters at the boundary.

Turbulence length scale TURBULENCE_LENGTH_SCALE, `LT is a typical length scale on theboundary. It is used to calculate the rate of dissipation of turbulent kinetic energy, ε at the bound-ary:

ε =Cµk3/2

`LT(3.4)

For internal flows, the length scale should be about half of the hydraulic diameter. It is notnecessary to give a turbulent length scale when a Pasquill class is set.

Wind buildup time WIND_BUILDUP_TIME is the time velocities on the boundaries used to risefrom zero to WIND_SPEED. A value for WIND_BUILDUP_TIME larger than zero gives a smoothstart of the simulation. GexCon recommends to use WIND_BUILDUP_TIME and to start eventualleaks after the wind field has reached steady state.

3.7.5.5 Fluctuating wind

One can also specify fluctuating wind field from the boundaries using a setup-file. Two differentfrequencies in horizontal directions and one in vertical direction are applied. This can be definedwith:

VERSION 1.1$WINDGUST

USE=.TRUE.$END


74 CASD

To change the frequencies of the fluctuations, more options must be specified:

VERSION 1.1$WINDGUST

USE = .TRUE.AMP = 2.40, 1.90, 1.84

, 2.40, 1.90, 0.00TAU = 15.0, 10.0, 10.0

, 70.0, 50.0, 0.0$END

Two different fluctuating periods are assumed along (15s and 70s) and across (10s and 50s) thewind direction. In the vertical direction one period of 10s is used. Fluctuations are done as har-monic periods with average velocity fluctuation equal to 2.4 (along), 1.9 (across) and 1.3 (verticaldirection, here constant is multiplied by square root of 2 since only one period is used) times thefriction velocity, u∗.

3.7.5.6 Using TRACER mode when simulating dispersion

In flacs2.2.6 a new simulation mode called ’tracer’ has been implemented. In this mode Flacs willonly solve a passive transport equation. Below is an example of how to enable the tracer mode:

VERSION 1.1$SETUP

KEYS="TRACE:T=100,DT_MUL=Y:5"$END

In the new scenario templates there is an option to define KEYS within the scenario file (no needto use a separate setup-file). The above setup-file (or the KEYS-line defined in the scenario file)will simulate normally until time=100s, thereafter only the fuel transport equation will be solved(flow field will be kept constant). When the flow field equations are switched off at 100s, the timestep is at the same time increased by a factor of 5 if using the DT_MUL=Y:5 string.

This option can be useful when a dispersion of neutral gas (or small quantities of gas) shall besimulated in an established wind field. This option should be used with care!

3.7.5.7 Effect of temperature gradients

To simulate the effect of an inversion layer, it is possible to define a cold or hot temperature region(layer) in FLACS. One can for instance define a cold valley by using the setup-file as described inprevious sections. Boundary conditions can not take a temperature profile.

3.7.5.8 General considerations for boundary conditions

It is generally advantageous to place the outer boundaries of the simulation domain far awayfrom the geometrical extent, but limitations of memory and computing speed may restrict thepractical size of the problem, and in most cases one is forced to compromise between quality andcost.

Solid wall boundary The solid wall boundary condition is straightforward to model. The ve-locity vectors are zero at solid walls, both in the tangential and the perpendicular directions.Hence, a zero gradient perpendicular to the boundary, or a fixed value, works well for the scalarvariables. Furthermore, wall-functions may improve the modelling of the flow in near-wall re-gions, both at the outer boundaries and in the interior space.



External influences In cases where there are obstacles outside the vent openings of a semi-confined volume they should be included in the total simulation volume, because they may havean effect on the explosion. One effect may be that the total venting is reduced due to the externalobstacles, especially if they are large and are placed close to the vents. Since the vent flow ischanged, also the internal flow past obstacles is modified and the explosion becomes different(higher or lower pressure results). An effect which also may be important is the appearance ofan external explosion which will start when the flame reaches any unburnt gas which may haveescaped through the vent openings. The pressure waves from the external explosion will prop-agate back into the semi-confined volume and give rise to higher pressures there. The strengthof the external explosion will depend on the local turbulence in the external space, this againdepends on the properties of the vent openings and on any obstacles which may be positionedin the external space.

3.7.6 Initial conditions

Initial conditions set values for turbulence fields, temperature and pressure at the beginning ofthe simulation. Information about the gravity condtions, parameters for the atmospheric bound-ary layer and the composition of the air is also set here.

The default values are as follows:

UP-DIRECTION 0 0 1GRAVITY_CONSTANT 9.8CHARACTERISTIC_VELOCITY 0.0RELATIVE_TURBULENCE_INTENSITY 0.0TURBULENCE_LENGTH_SCALE 0.0TEMPERATURE 20.0AMBIENT PRESSURE 100000AIR "NORMAL"GROUND_HEIGHT 0GROUND ROUGHNESS 0REFERENCE_HEIGHT 0LATITUDE 0SURFACE_HEAT_P1 0 0 0SURFACE_HEAT_P2 0 0 0MEAN_SURFACE_HEAT_FLUX 0PASQUILL_CLASS "NONE"GROUND_ROUGHNESS_CONDITON "RURAL"

Attention:

GexCon recommends to use the default values for explosion scenarios.

3.7.6.1 Up-direction

This is a three-component vector defining the upward direction, i.e. opposite of the accelerationdue to gravity. The three components denote the spatial components x, y, and z, respectively. It ispossible to define any direction, not only directions aligned with the main axes. The vector doesnot need to be of unit length.

3.7.6.2 Gravity constant

GRAVITY_CONSTANT is the magnitude of the gravitational acceleration, normally 9.8m/s2. Set-ting a zero value for this parameter means that there are no gravitational influences on the flow.Note that panels with inertia are also influenced by this parameter if the direction of panel releaseis in the direction of the Up-direction, see also Pressure relief panels .


76 CASD

3.7.6.3 Characteristic velocity

CHARACTERISTIC_VELOCITY, U0, is used to find values for initial turbulence fields and itshould take a positive or a zero value.

3.7.6.4 Relative turbulence intensity

RELATIVE_TURBULENCE_INTENSITY, IT , is the ratio between the isotropic fluctuating veloc-ity, u′, and the mean flow velocity U0:

IT =u′

U0(3.5)

IT will typically have a value in the range 0.0 to 0.1. This parameter is used together with Charac-teristic velocity to calculate the value for the turbulent kinetic energy, k = 3/2u′2 at the beginningof the simulation.

3.7.6.5 Turbulence length scale

TURBULENCE_LENGTH_SCALE, `LT , is the length scale of the initial turbulence. A reasonablechoice for TURBULENCE_LENGTH_SCALE is a typical value for the grid size ∆g. `LT is used tocalculate an initial value for dissipation of turbulent kinetic energy, ε:

ε =Cµk3/2

`LT(3.6)

3.7.6.6 Temperature

The initial temperature, T0 in C◦. The default value is T0 = 20.0 C◦

3.7.6.7 Ambient pressure

AMBIENT PRESSURE, P0 is the initial pressure in the simulation and the pressure outside thesimulation volume. The default ambient pressure is P0 = 100000 Pa = 1 bar.

3.7.6.8 Air

AIR is used to define the composition of the air. "NORMAL" denotes a standard composition of20.95% oxygen and 71.05% nitrogen in mole fractions. The composition of air can be changed bysetting an other oxygen content, either as a mole fraction:

AIR "25%MOLE"

or mass fraction:

AIR "10%MASS"

Changing the air composition will influence the lower flammability limit and upper flammabilitylimit.



3.7.6.9 Wind boundary layer parameters

In the Wind boundary condition, the reference wind speed, U0 and the direction of the windare specified. In additions, information about the ground conditions is needed to setup inletprofiles for velocity and turbulence parameters. It is possbile to specify an atmospheric boundarylayer stability class, Pasquill class. Then turbulence parameter profiles are generated at the windboundaries. If no Pasquil class is given, uniform values for k and ε are obtained on the boundaryaccording to the expressions in Relative turbulence intensity and Turbulence length scale.

The velocity profile at the wind boundaries is given by the following expression:

u(z) =

{u∗κ

(ln(

z+z0z0

))if z0 > 0

U0 if z0 = 0(3.7)

where z is the height abouve the ground, z0 is the atmospheric roughness length and u∗ is afriction velocity. For the neutral and none Pasquill class, u∗ is defined by:

u∗ =U0κ

ln( zre f

z0

) (3.8)

Ground Height GROUND_HEIGHT is height above the ground where the boundary layer ac-tually starts, for instance due to vegetation. Usually GROUND_HEIGHT=0.

Ground roughness GROUND_ROUGHNESS, z0, refers to the aerodynamic roughness length .Typical values for z0 are given in table Typical values for aerodynamic roughness length. z0should in not be larger than the control volume height close to the surface. aerodynamic rough-ness length should not be mixed with pipe roughnes etc., but a rule of dump is to relate z0 to theaverage height εg of the surface irregularities by z0 = εg/30.

Terrain description z0 (m)Open water, fetch at least 5 km 0.0002Mud flats, snow; no vegation, no obstacles 0.005Open flat terrain; grass, few isolatedobstacles

0.03

Low crops; occasional large obstacles 0.10High crops, scattered obstacles 0.25Parkland, bushes, numerous obstacles 0.5Regular large obstacle coverage (suburb,forest)

1.0

Table 3.4: Typical values for aerodynamic roughness length

Reference height REFERENCE_HEIGHT, zref is the height relative to the ground where the ve-locity equals the wind speed.

Latitude Due to the rotation of the earth, the height of the atmospheric boundary layer is muchlarger at equator than at the poles. LATITUDE will only have an effect if a Pasquill class is chosenand the simulation volume is very large (> 200 m) in the z direction.


78 CASD

Surface heat P1 If the temperature and velocity are known at two different altitudes, for in-stance from experimental data, it is possible to estimate the surface heat flux.

SURFACE HEAT P1 is a vector Z1 U1 T1, where Z1 is the altitude, U1 is the velocity at Z1 andT1 is the temperature at Z1

Surface heat P2 SURFACE HEAT P2 is a vector Z2 U2 T2, where Z2 is the altitude, U2 is thevelocity at Z2 and T2 is the temperature at Z2. Z2 must larger than Z1 and U2 must larger thanU1.

Mean surface heat flux MEAN_SURFACE_HEAT_FLUX is the heat flux in W/m2 from theground to the flow. This is a parameter in the boundary layer profiles for the unstable Pasquillclass A, B, and C. MEAN_SURFACE_HEAT_FLUX will not apply as heat contribution from theground to the flow in the simulations.

Pasquill class Pasquill atmospheric stability classes is a method of categorizing the amount ofatmospheric turbulence present. Pasquill (1961) made six stability classes where:

A is very unstable

B is unstable

C is slightly unstable

D is neutral

E is slightly stable

F is stable

An overview when to apply the different Pasquill classes is given in table: Pasquill stabilityclasses.

Wind speed Day, strong sun Day, moderatesun

Night, clouds>50%

Night, clouds<50%

< 2m/s A B E F2 - 3 m/s A-B B-C E F3 - 5 m/s B B-C D E5 - 6 m/s C C-D D D> 6 m/s C D D D

Table 3.5: The Pasquill stability classes

Ground roughness conditions At present the only option is:

GROUND_ROUGHNESS_CONDITON "RURAL"

3.7.7 Pressure relief panels

Pressure relief panels are commonly used in the process industry as a mitigating device in thecase of an explosion. When the pressure forces on the panel exceed a certain limit, the panel yieldsand the pressure is relieved. In FLACS, special planar elements inside the simulation volume are



used to model the effect of explosion relief panels and yielding walls. They are specified as boxeswhere one, and only one, of the three side lengths must be zero. In addition several parameterssuch as the yield pressures, area porosities and drag factor may be specified.

A panel may be active or passive. An active panel will initially modify the porosity in the regionit covers, and again when the pressure difference over the panel exceeds the given limit. If theinitial porosity is set to 0 and the final porosity is set to 1 the panel will start as being closed andend up as being open. This is how the behaviour of a real yielding panel is imitated. A passivepanel will never modify the porosity, but it may be used to monitor the same data as an activepanel. There are 5 different panel types (active panels):

Panel type DescriptionUNSPECIFIED Panel with linear displacement, full

parameter setPOPOUT Panel with a linear displacement, limited

parameter setHINGED Panel with a rotational displacement, limited

parameter setPLASTIC Simulates the presence of plastic sheets

(commonly used in experiments)OVERLAY Panel properties are ’blended’ with existing

equipmentTable 3.6: Panel types in FLACS

Note that the PLASTIC and OVERLAY panels can not be chosen directly in CASD. It is possibleto edit the scenario file manually to use these panel types, by initially creating an UNSPECIFIEDpanel and change it to the desired type. It is advised that only experts attempt to do this.

The edges of a panel will be automatically adjusted to the closest grid line. It is advised that theuser specify panels whose dimensions match the grid, either by adjusting the grid or modifyingthe panel.

One topic which needs special attention is the presence of solid structures close to the panel area.For example structural beams may constitute quite large blockages which must be accounted for,especially since they also occur at the vent openings of typical offshore modules. In such casesthe panel area must be defined using several panels with solid beams in between. Smaller beams,often an integral part of the support frame of the pressure relief panels, may be accounted for byadjusting the final panel porosity for the panel itself.

In addition to different types of pressure relief panels, an INACTIVE panel used to monitor vari-ables is incorporated in the FLACS code. Use an INACTIVE panel to measure for instance averagepressure across surfaces such as decks and walls. The panel will be available under Single fieldscalar time output in the Scenario menu.

Note that if several panels are specified, they should not overlap each other anywhere (in themodel it is assumed that the panels are non-overlapping).

The full list of parameters from the scenario file is shown below. The type of the panel determineswhich of these parameters are relevant.

INSERT 1POSITION 0.0, 0.0, 0.0 (m)SIZE 1.0, 1.0, 0.0 (m)MATERIAL BLUEPANEL_TYPE UNSPECIFIEDOPENING_PRESSURE_DIFFERENCES 0.0, 0.0 (bar)INITIAL_AND_FINAL_POROSITY 0.0, 1.0 (-)


80 CASD

WEIGHT 0.0 (kg/m2)DRAG_COEFFICIENT 1.0 (-)MAXIMUM_TRAVEL_DISTANCE 0.0 (m)SUB_SIZES 1.0, 1.0 (m)

The parameters for specifying panel properties are described in the following sections.

3.7.7.1 Insert

Integer identifying the panel.

3.7.7.2 Position

Cartesian coordinates [m] of the corner of the panel (the corner with lowest value of the coordi-nate in each axis direction).

3.7.7.3 Size

The panel is assumed to be a plane of rectangular shape. The dimension [m] in each of the axisdirections is given. One dimension should be zero, this shows how the panel is oriented. If e.g.the dimension in x-direction is zero, the normal of the panel points in x-direction. The other twodimensions should be positive. The panel edges will be adjusted

to match the grid lines perfectly in the FLACS code, so it is advised that you only specify panelswhich coincide with the grid lines, in order to avoid any confusion concerning the geometricalrepresentation of the panel.

3.7.7.4 Material

Colour used when visualizing the panel as part of the geometry considered.

3.7.7.5 Panel type

There are 6 different panel types described here. A panel is either active or passive. They arelisted and described in more detail below.

Inactive is a passive panel that does not affect the geometry or the flow field. such panels areused to monitor certain variables

Unspecified is a panel with linear displacement when the panel yields

Popout is a panel with a linear displacement when the panel yields

Hinged is a panel with a rotational displacement when the panel yields

Plastic is a panel that imitates a plastic sheet (commonly used in experiments)

Overlay is a panel superimposed on other equipment



Inactive panel A passive panel, corresponding to the case when PANEL_TYPE equalsINACTIVE, does not affect the numerical simulation in any respect (neither the geometrical de-scription including the porosities, nor the flow field), it is only included in order to monitorvariables related to the area which the passive panel occupies (e.g. area-averaged pressure overall the control-volume faces of the panel). Only the parameters INSERT, POSITION, SIZE, andMATERIAL are relevant when the PANEL_TYPE is INACTIVE.

Unspecified, Popout and Hinged panel An active panel (all the panel types except INACTIVE)will in general affect both the porosities representing the geometry and the flow field. For theactive panels both an initial and a final area porosity value is given (open area divided by totalarea). The initial porosity corresponds to the state of the panel before it has started to yield dueto external pressure forces. The final porosity corresponds to the state of the panel after it hasyielded completely due to pressure forces.

Note that each control-volume face covered by the panel is treated separately. Thus it is possible,depending on the scenario, that one part of the panel is open, while at the same time other partsare closed.

The way of using the initial and final porosities in the FLACS code depends on the type of panelconsidered. In the case of the panel types UNSPECIFIED, POPOUT, and HINGED; for each control-volume face which is covered by the panel, the porosity representing the panel is also the porosityof the control-volume face, regardless of any other equipment or structure included in the geom-etry.

Overlay and Plastic panel For a panel of type OVERLAY the value of the area porosity deter-mined by the panel is multiplied by the value of the area porosity determined by other geometri-cal objects (in the absence of any panel) giving the effective area porosity, for each control-volumeface covered by the panel. Let us consider an example where the final porosity of the panel is 1(fully open). After the panel has yielded completely for all the control-volume faces covered bythe panel, the area porosities have the same values that they would have if no panel was includedin the geometry. So in this case other geometrical objects than the panel itself can not "blow out"together with the yielding panel (cf. the comments above).

For a panel of type PLASTIC, and in the case of the initial area-porosity; for each control-volumeface which is covered by the panel, the porosity representing the panel is also the porosity of thecontrol-volume face, regardless of any other equipment or structure included in the geometry(as in the case of the panel types UNSPECIFIED, POPOUT, and HINGED). Usually the initial area-porosity of a PLASTIC panel is zero (completely blocked). If the PLASTIC panel has yieldedcompletely, the area porosities have the same values that they would have if no panel was in-cluded in the geometry (as in the case of an OVERLAY panel with final area-porosity 1). Notethat for a PLASTIC panel the final area-porosity is always assumed to be 1, regardless of theuser-given value.

Panels of the types PLASTIC or OVERLAY are assumed to be light panels, and inertial forcesare neglected by the FLACS code during the dynamical process when the panel yields due topressure forces. In the case of panels of types UNSPECIFIED, POPOUT, or HINGED, inertial forcesare in general taken into account.

Attention:

The PLASTIC and OVERLAY panels cannot be chosen in the present version of CASD. It ispossible to edit the scenario file manually to use these panel types, by initially creating anUNSPECIFIED panel and change it to the desired type. It is advised that only experts attemptto do this.


82 CASD

The +IMP panel types The standard panel types will yield at a give negative or positive pres-sure. In some situations it can of necessary for panels to also take into account the pressureimpulse. To activate this the user can add the suffix +IMP to the end of the panel type string, e.g

PANEL_TYPE POPOUT+IMPOPENING_PRESSURE_DIFFERENCES 0.05 0.005 (bar bar*s)

When using this the input to OPENING_PRESSURE_DIFFERENCES changes to

• Parameter 1 is yield pressure for both negative and positive direction [bar]• Parameter 2 is yield pressure impulse both negative and positive direction [bar∗s]

The panel will yield if both the pressure and pressure impulse yield values are exceeded.

This is valid for panel types HINGED and POPOUT.

Attention:

This functionality is not available through CASD. The user must manually edit the cs-file.

3.7.7.6 Opening pressure differences

When the net pressure over the panel (pressure on the negative side relative the coordinate axisminus pressure on the positive side, i.e. the net pressure from the fluid acting on the panel)exceeds limits given by the user, the panel starts to yield. Two opening pressure-differences, inunits of [bar], are given by the user, the first corresponds to the case when the net pressure acts innegative direction, the second when the net pressure acts in positive direction. If for example theOPENING_PRESSURE_DIFFERENCES are given by -0.1 and 0.2, the panel starts to yield when thenet pressure is less than -0.1bar (negative direction) or greater than 0.2 bar (positive direction).Note that the FLACS code makes the same interpretation of the input data, regardless of thesign of the opening pressure-differences given by the user. Hence, if the OPENING_PRESSURE_-DIFFERENCES were given instead by 0.1 and 0.2, or by 0.1 and -0.2, the effect on the numericalsimulation would be exactly the same. The opening pressure-differences must be given for theactive panels (all the panel types accept INACTIVE).

3.7.7.7 Initial and final porosity

The initial area porosity [-] of the panel (open area divided by total area) corresponds to thestate of the panel before it has started to yield due to external pressure forces. Similarly the finalporosity corresponds to the state of the panel when it has yielded completely. In the case ofa PLASTIC panel, the final area-porosity should be set to 1 by the user in order to correspondto how the FLACS code works (the final area-porosity is always assumed by FLACS to be 1,regardless of the user-given value). Note that the way of using the initial and final porosities inthe FLACS code depends on the type of panel considered. See the comments above. The initialand final porosity must be given for the active panels (all the panel types accept INACTIVE). Thevalue of the porosity range from 0 (totally blocked) to 1 (fully open). It is expected that the initialporosity is less than the final porosity (the net pressure opens up the panel). If this is not the case,the FLACS code gives a warning.

3.7.7.8 Weight

The parameter WEIGHT is specified for the panel types UNSPECIFIED, POPOUT, and HINGED (forthe other panel types it is not relevant to specify this parameter). This parameter is the mass per



unit area [kg/m2] of the part of the panel that is moved when the net pressure acting on the panelexceeds the opening pressure-difference given by the user. The value of WEIGHT is either zero orpositive. If the value of WEIGHT is zero, it is assumed in the FLACS code that the panel is so lightthat inertial forces can be neglected during the dynamical process when the panel yields due topressure forces. If the value of WEIGHT is positive, the dynamics when the panel yields and partof the panel is accelerated and move away from its initial position, is modelled in the FLACScode. Note that when you specify the panel type POPOUT, the inertial forces must be there whenthe panel yields, that is the WEIGHT must be a positive number. Note that it is only the part ofthe panel that is accelerated and move away from its initial position, that should contribute tothe mass per unit area. The mass of a rigid frame that does not yield, should not be included.

3.7.7.9 Drag coefficient

A dimensionless drag-coefficient [-] is given by the user for the panel type UNSPECIFIED (forthe other panel types it is not relevant to specify this parameter). The drag coefficient is usedwhen the drag force from the panel on the fluid is modelled (both before, during, and after thepanel has yielded). The value of the DRAG_COEFFICIENT is zero or positive. A typical value is2.0 (the value 2.0 is set by the FLACS code as a fixed preset value for the panel types POPOUT andHINGED).

3.7.7.10 Maximum travel distance

The maximum travel-distance [m] is given by the user for the panel type UNSPECIFIED (forthe other panel types it is not relevant to specify this parameter). The use of the maximumtravel-distance is based on a rough approximation. The maximum travel- distance is the small-est distance from the initial position of the panel to the position where the yielded panel nolonger affects significantly the effective area-porosity at the initial position of the panel. Themaximum travel-distance is used to model the effective area-porosity at the initial position of thepanel during the dynamical process when the panel yields. A typical value of the MAXIMUM_-TRAVEL_DISTANCE is in the order of 1m (for the panel types POPOUT and HINGED there is anin-built model in the FLACS code for the effective area-porosity that does not need a value of theMAXIMUM_TRAVEL_DISTANCE as input from the user).

3.7.7.11 Sub sizes

The parameter SUB_SIZES is given by the user for the panel types POPOUT or HINGED (for theother panel types it is not relevant to specify this parameter). These panel types are assumed toconsist of sub-panels mounted on a frame. The width [m] and the height [m] of the sub-panelsare given by the user (it is assumed in the numerical model that all the sub-panels are of uniformsize). Both the width and the height should be positive. In the case of the panel type HINGED itis assumed that each sub-panel turns on a hinge when it yields. The width of the sub-panel isdefined as the dimension in the direction normal to the axis of the hinge. It is important (sinceit affects the model in FLACS) to specify the width of a HINGED panel first and then the height,in the parameter SUB_SIZES (for a POPOUT panel it is of no importance which dimension that isdefined as the width).

3.7.8 Gas composition and volume

This section allows you to define a box shaped cloud region and the gas concentration and com-position. The menu is shown below:


84 CASD

POSITION_OF_FUEL_REGION 0.0, 0.0, 0.0 (m)DIMENSION_OF_FUEL_REGION 0.0, 0.0, 0.0 (m)TOXIC_SPECIFICATION ""VOLUME_FRACTIONS <MENU>EQUIVALENCE_RATIOS_(ER0_ER9) 1.0, 0.0

The position of the cloud is the location of the minimum point (xmin, ymin, zmin) of the box, and thesizes (xsiz, ysiz, zsiz) give the side lengths of the box (only positive values allowed for the sizes).The gas composition is specified by entering volume fractions of the listed gas components, andthe concentration is given by the equivalence ratios (ER0 inside the gas cloud and ER9 outside)

METHANE 1.0ACETYLENE 0.0ETHYLENE 0.0ETHANE 0.0PROPYLENE 0.0PROPANE 0.0BUTANE 0.0HYDROGEN 0.0CO 0.0CO2 0.0 (inert)H2S 0.0TOXIC 0.0

The sum of the volume fractions does not need to be 1, FLACS will interpret the values as volu-metric parts (normalization is done in FLACS by dividing each value by the sum of all values).

For a mixture of two or more gas components, the combustion properties are calculated as a sortof average of all the selected components, taking into account the mole (volume) fraction and therelative consumption of O2 (parameter a in the table above). This has shown to give the properblending of the component properties into mixture properties.

The laminar burning velocity of a gas mixture in FLACS depends on the concentration of the fuelrelative to the concentration of oxygen as well as on the type of fuel. A widely used measure forthe relative fuel-oxygen concentration is the Equivalence ratio (ER) which is defined as follows:

ER = θ =

(m f uel/moxygen

)actual(

m f uel/moxygen

)stoichiometric

=

(Vf uel/Voxygen

)actual(

Vf uel/Voxygen

)stoichiometric

(3.9)

Now, the dependency of the laminar burning velocity (Slam) is illustrated for methane in thefigure below. The factor, ERfac, which is a function of ER and ranges from 0 to 1 is multipliedwith the tabular value of the laminar burning velocity for the gas component, thus a relationbetween Slam and ER is established.

Each gas component has its own ERfac curve in FLACS. The next figure shows the laminar burn-ing velocity curve as function of ER for all the gas components separately. Note that hydrogen,acetylene and propylene have the three highest laminar burning velocities, and notably also thewidest range of flammability.

For a gas mixture the resulting curve is based on the blended properties of the components ashave been described initially in this section.

3.7.9 New gas data

The gas-data for flacs2.2.7 are found in " ∼flacs/FLACS/bin/files/gasdata2.2.7/" , these havebeen adjusted for better conformance with published values (especially for the LFL and UFL



values). In addition, the model for the effect of inert gases on the flammability limits and burningvelocity has been revised.

The effect of inert gas on the flammability limits can be better understood by looking at the plotbelow:

Notice that He or N2 cannot be modelled as " fuels" in FLACS yet. H2O can now be modelled.

The two plots below show the effect of different inert gases on the laminar burning velocity ofmethane at varying equivalence ratio (ER):

3.7.10 Leaks

Gas leaks occur quite frequently in many offshore and onshore installations and elsewhere, eitherduring production, processing or transport of the gas. Hydrocarbons and many other types ofhighly flammable gases are present all over the industry and may be a source for potentiallydangerous gas explosions. In addition, spills of flammable liquids may cause accumulation ofexplosive gas clouds under certain conditions. Experiments have clearly shown that even smallamounts of gas mixed with air at the right concentration can result in very strong explosions.By using the dispersion and explosion simulation capabilities of FLACS it is possible to analysequite realistic scenarios, where the effect of a single gas leakage developing into an explosivegas cloud and finally resulting in several possible gas explosion scenarios may be studied. Oneprimary interest may be to vary the leakage itself, changing the flow rate, location, and duration.Other sensitivities may also be studied such as modifying the ventilation and wind conditionsfor a given geometrical layout. Finally, the variation of ignition time and location may be madeduring the explosion studies.

The release model for leaks in FLACS ensures that the desired mass flow or velocity of gas is setat the control volume where the leak is located. Also other properties such as the temperatureand relative turbulence intensity and length scale must be specified. The figure below shows howa jet is represented on the grid:

Figure 3.12: Leak definition


86 CASD

The area porosity of the open control volume face is adjusted to ensure that the flow rate iscorrect. If the grid resolution is coarse compared to the size of jet orifice, a good representation ofthe detailed flow in the near-orifice region is not possible. This may lead to excessive smearingof the gradients of the flow. In such cases (which may be the normal practical case) one mustrealize that the simulated gas concentration is locally underestimated, although the total amountis correct. The grid resolution may be especially important for jets impinging on nearby wallsand jets with a strong cross-wind influence.

It is possible to specify one or more leakage points in a FLACS simulation. The leakage is in-tended to model the flow of gas from a reservoir into the simulation volume, the available pa-rameters controlling the leakage are shown next:

LEAKSINSERT 1TYPE "JET"POSITION 1.0, 1.0, 1.0 (m)OPEN_SIDES "+X"START_TIME 1.0 (s)DURATION 10.0 (s)OUTLETVESSEL

Attention:

Remember to set the ER0 parameter under GAS_COMPOSITION_AND_VOLUME to a largenunber if the leak is a pure fuel leak.EXCESS LEAK RATE

3.7.10.1 Insert

This command is used to insert a new leakage, the parameters for this leakage will be explainedin the following next pages. The maximum number of leakages is 50.

3.7.10.2 Type

The type of leakage must be specified as one of the following:

DIFFUSE Low momentum point leak (deprecated)

JET Low and high momentum point leak (preferred)

AIR Similar to "JET", but leak mixture is air (ER9)

SUCTION Negative point source, i.e. removal of gas/air mixture

FAN Fixed momentum in CV

The pool leak is not specified in this menu but is rather defined using the cs.POOL and thecl.POOL files (see section Pool setup file).

Diffuse The DIFFUSE leak is a no momentum (same velocity as surrounding flow) release ofgas into the grid cell chosen. One must give a direction, but this direction will not be applied.DIFFUSE should normally not be used.

Jet Low and high momentum point leak.



Air Low and high momentum point leak of air (ER9).

Suction Negative point source, i.e. removal of gas/air mixture.

Air If a FANwith direction +X is specified, this will be translated into the string "!J+X=X:fan"in the leak file. Fans should be along the axis directions.

3.7.10.3 Leak build-up

Leaks previously were built up in a time equal to 1% of the leak duration, and thereafter immedi-ately shut down when killed. Now the maximum build-up time has been set to 1s, the shut-downhas also been made smoother, with a gradual shut-down over 1s. These values can be changedmanually in cl-files.

3.7.10.4 Leak concentration between ER0 and ER9

Concentration between ER0 and ER9 can be specified for a leak. The control string must bechanged manually.

Example: To get a concentration of 60% ER0 and 40% ER9 one must change leak control string incl-file.

From:

"!J+X"

to:

"J+X:mix=0.6"

3.7.10.5 Q8 and Q9 output

The rt.FUEL and rt.MON files have now two columns, called Q8 and Q9. These files are gen-erated from GAS_MONITOR_REGION" in cs-file or cs-MON files. Q8 and Q9 is used to createequivalent stoichiometric gas clouds from real gas clouds.

Q8 This column reports expansion weighted volume, i.e. closed volume equivalent cloud atconcentration for maximum expansion (normally near stoichiometry).

Q9 This column gives an improved version of Q5 for gases where maximum flame speed andmaximum expansion deviates. Q9 will for other gases be marginally higher than Q5.

For enclosed situations, and situations where combustion is much quicker than venting(including quasi-detonation and flames faster than speed of sound ahead) Q8 will be arecommended equivalent cloud size.

For well-vented situations we generally recommend to use Q9 (to replace Q5) as equivalentstoichiometric cloud.

3.7.10.6 Position

This menu is used to specify the leak position (in metres).


88 CASD

3.7.10.7 Open sides

The open sides indicates which side(s) of the control volume affected by the leak should be madeto open up as the leak starts (direction of the leak). OPEN_SIDES is a text string containing char-acters taken from "+-XYZ". A leakage directed in the positive x-direction should have OPEN_-SIDES = "+X". Several sides may be open for a single leak, e.g. OPEN_SIDES = "+X-X" orOPEN_SIDES = "+XY-ZX". This will give two or more identical leaks

3.7.10.8 Start time

This is the time (in seconds) when the leakage should start. Normally one would allow a certaintime for wind build-up prior to starting any leaks.

3.7.10.9 Duration

This is the duration (in seconds) of the leakage. The finish time for the leakage is the start timeplus the duration. Hence, the user should set TMAX accordingly. The user should also note thatthe build-up time of the leak is 1 % of leak duration (maximum 1 second).

3.7.10.10 The Outlet menu

Users specify the leak conditions in the outlet menu, e.g.

AREA 0.1 (m2)MASS_FLOW 0.0 (kg/s)VELOCITY 10.0 (m/s)RELATIVE_TURBULENCE_INTENSITY 0.15TURBULENCE_LENGTH_SCALE 0.01 (m)TEMPERATURE 6.0 (°C)DIRECTION_COSINES 0.0, 0.0, 0.0

3.7.10.11 Outlet: area

The effective cross-section area A of the leak outlet must be specified for any type of leak. For aspecified mass flow m (or volume flow V and density ρ) the velocity u at the outlet should not beabove the speed of sound. The following relations apply for the conditions at the outlet:

m = ρV = ρuA (3.10)

Assuming ideal gas properties:

pV = n<T =mM<T = mRT (3.11)

where < is the ideal gas constant (8314.3 [J/kmole ·K]), T [K] the and the specific gas constant Ris equal to the ideal gas constant divided by the molecular weight M, the gas density at the outletis:

ρ =mV

=p

RT(3.12)

m = ρuA Mass flow at the leak outlet (kg/m3/s)



ρ = p/(RT) Gas density at the outlet (kg/m3)

p = pamb Pressure at the outlet is the ambient pressure (Pa)

R = </W Specific gas constant (J/kg/K)

W Mole weight for the given gas composition (kg/kmole)

T Temperature at the outlet (K) (0.0 °C = 273.15 K

u Velocity at the outlet (m/s)

The relations shown above may be used to ensure that the outlet conditions are set in a consistentway.

3.7.10.12 Outlet: mass flow

The user can specify the mass flow rate at the leak outlet here. Note that mass flow takes priorityif both mass flow and velocity are specified (see below). Make sure that the generated velocity isnot larger than the speed of sound (approximately 340 m/s for air). Setting the outlet area largeenough for a given mass flow rate causes the outlet velocity to be as low as desired. FLACS willreport the generated velocity on the log file (rt-file). The mass flow rate must have a positivevalue (zero if not specified).

3.7.10.13 Outlet: velocity

Instead of specifying a mass flow rate at the outlet (see above) the user may specify the velocity.FLACS will report the generated mass flow rate on the log file (rt-file). If you specify both MASS_-FLOW and VELOCITY then the mass flow rate takes precedence. The velocity must have a positivevalue (zero if not specified). The specified velocity should be lower than the speed of sound.

3.7.10.14 Outlet: relative turbulence intensity

This parameter must always be specified for a leakage. Normally a value in the range 0.01 to0.10 will be appropriate. If the relative turbulence intensity cannot be obtained from any source(literature or experiments) for the given leakage, the following rough classification may be used:

0.01-0.03 Low turbulence intensity

0.03-0.06 Medium turbulence intensity

0.06-0.10 High turbulence intensity

It is generally believed that the turbulence generated in the jet zone is less important than theturbulence generated by the induced downstream flow field, in which case the k − ε model inFLACS represents the actual modelling. It should be noted here that on the coarse grids nor-mally used for dispersion calculations in large geometries the generated turbulence due to fluidstresses may be underestimated, and since numerical diffusion may be high on such coarse gridsone might expect that the turbulent (or effective) mixing process is not represented with highaccuracy.

3.7.10.15 Outlet: turbulent length scale

This parameter must always be specified for a leakage. It may be difficult to obtain the value forthe turbulence length scale. It is generally related to the size of the nozzle or the feeding pipe


90 CASD

for the leakage. A rough estimate is that the turbulence length scale is in the range of 10% to20% of the nozzle diameter. When the turbulence length scale is small the dissipation rate of theturbulence kinetic energy is large, thus the turbulence will dissipate quickly for small diameternozzles. In such cases, the turbulence generated by the downstream flow field is the importantfactor, and the k− ε model in FLACS takes care of that.

3.7.10.16 Outlet: temperature

This parameter must always be specified for a leakage.

3.7.10.17 Outlet: direction cosines

This is a vector which may be used to define the direction of an oblique jet.

3.7.10.18 The Vessel menu

This menu was developed to calculate the outlet conditions for a leak from a pressurized vessel.Currently, it is not recommended to use the vessel menu. The Jet utility program should be usedinstead.

3.7.11 Ignition

In case of a gas explosion simulation the user must specify the location and size of the ignitionsource and also a time for the ignition to occur. The available parameters here are as follows:

POSITION_OF_IGNITION_REGION 0.0, 0.0, 0.0 (m)DIMENSION_OF_IGNITION_REGION 0.0, 0.0, 0.0 (m)TIME_OF_IGNITION 0.0 (s)RADMAX 99999.0 (m)

The ignition region can be a point, a line, a plane or a volume. Normally the user would choosean ignition point (zero dimension), with ignition effected at time zero. The RADMAX parameterwill be ignored by all FLACS versions after 1998, but it is only present for backward compatibility.

If the ignition point is inside a partially blocked control volume, the flame might quench. It istherefore not recommended to ignite inside a partially blocked control volume. But if the user stillchooses to do so, and if there are any problems to obtain a proper ignition and flame propagation,the user might try to increase the DIMENSION_OF_IGNITION_REGION side lengths up to about0.05 to 0.10 m.

Ignition in FLACS is usually set to occur in just one control volume. It is possible to define a largerregion for the ignition, but this is not advised. The time of the ignition may also be specified, thisis useful for igniting gas clouds which are generated during gas dispersion simulations. In anormal gas explosion simulation the time of ignition should be set to zero.

Attention:

The user should avoid specifying the ignition point exactly on a grid line, and should ratherspecify it at the centre of a control volume. Usually, it is best to specify the ignition pointaccording to the grid, and not according to the geometry. For instance, if the ignition point isplaced on one side close to a solid wall,it may not necessarily ignite at that side of the wall.During the porosity calculation, the wall may be adjusted to the nearest control volume face



(grid line) and might therefore move to the wrong side of an ignition point! It may also be agood idea to ensure that the ignition point is not inside a fully blocked control volume.

Figure 3.13: How to position the ignition

3.7.12 Water spray

Water deluge systems can be an effective way to mitigate the consequences of gas explosionsin several situations. The mitigating effect has been seen in many experiments. However, thephenomena related to the interaction between water sprays and an accelerated flow field and theflame are quite complex. The explosion accelerates the flow as well as the water droplets. Theacceleration depends on the local flow conditions, and the droplet size. When hydrodynamicforces become large enough to overcome surface tension, the droplets break up (again dependenton the droplet size). When the droplets are small enough (either because the droplets producedby the nozzle are small, or because large droplets break up into small droplets when the flow isaccelerated), they tend to reduce the burning rate due to cooling of the flame, and dilution of thegas mixture by evaporation. However, large droplets in the flame region (before break-up of thelarge droplets) tend to increase the turbulence level and thereby increase the burning rate. Sothere are competing effects when using a water spray system. In some cases water spray mayeven increase the maximum overpressure of an explosion.

The problem when modelling water sprays is how to represent all this on a coarse simulationgrid. It was necessary to do simplifications, and also to use experiments performed for tuning ofthe models. Two distinct effects were identified.

1. The acceleration of the flames due to turbulence from the sprays and the presence of thedroplets.


92 CASD

2. The reduction of burning rate experienced in certain situations because of the water sprays.

For each of the nozzles an acceleration factor, denoted F1, is determined. F1 is used to increase theburning rate if any watersprays are present. A quenching factor, denoted F2, is also determined.F2 is used to reduce the burning rate if the conditions for droplet break-up are present.

Remarks:

Due to model simplifications, there is no need for a very accurate positioning of the sprays.In regions where sprays of the same type will overlap, the user should define one waterspray region for the whole system. If more than one waterspray region is to be defined, theuser should make sure that they do not overlap, as FLACS will then stop. In FLACS, thedroplets are assigned a velocity but the transportation of droplets is not modelled. Regionswhere it is obvious that a lot of droplets will be transported ahead of the flames (for instancedirectly outside vent openings), should be included in the spray region (i.e. larger waterspray regions should be defined).

The models are validated in stoichiometric gas concentrations in a 180 m3 vented box, with vari-ous obstruction levels, in a 50 m3 model of an offshore module, and also in full- scale experiments.It is reasonable to believe that the models represent mechanisms in connection to water mitiga-tion well.

The water spray model implemented in the FLACS code is relatively simple. One or more non-overlapping water spray regions are defined. In each region there is assumed to be droplets ofa given diameter (before break up), and a given water volume-fraction. If the relative velocitybetween the droplets and the gas flow exceeds a so-called critical break-up velocity (dependingon the diameter of the droplet), it is assumed that the droplets break up.

Two non-dimensional factors are employed in the numerical model. When there is water in thereactive mixture (i.e. inside a water spray region), this is assumed to enhance the burning rate.Before break-up of the droplets, the burning enhancement-factor denoted F1 [-] (positive number)is multiplied by the laminar burning velocity and added to the ordinary burning velocity (i.e. theburning velocity employed in the FLACS code without any water spray, this burning velocity isin general in the turbulent regime) to give the effective burning velocity with water spray. Whenthe droplet break-up criterion is fulfilled, the burning velocity (without break-up) is multipliedby the burning reduction-factor F2 [-] (positive number less than 1), to give the effective burningvelocity

Swater = (Sturb + F1 · Slam)F2 (3.13)

The water spray model is validated for explosions with stoichiometric fuel-air mixtures. For non-stoichiometric fuel-air mixtures it is assumed that the pressure reduction caused by water spraymight be slightly under predicted.

Below is shown a list of parameters which may be visible on the scenario file:

INSERT 1POSITION 0.0, 0.0, 0.0 [m]SIZE 1.0, 1.0, 1.0 [m]VOLUME_FRACTION 0.2 [per thousand]MEAN_DROPLET_DIAMETER 300.0 [mm]NOZZLE_TYPE "FACTORS: 10.0 0.3" [-]



3.7.13 Insert

Integer identifying the water spray region considered. A maximum number of 25 regions maybe defined in the current version of the FLACS code.

3.7.14 Position

Cartesian coordinates [m] of the corner of the box-shaped water spray region (the corner withlowest value of the coordinate in each axis direction). Make sure not to define any overlappingregions. The FLACS code will assign the nearest grid line to your given position as the actualposition of the water spray region. In other words, FLACS will make the water spray region stickto the grid lines.

3.7.15 Size

The dimension [m] in each of the axis directions is given for the box-shaped water spray region.All three dimensions should be positive. Make sure not to define any overlapping regions, theFLACS code will assign the nearest grid line to your given (position + size) as the actual (position+ size) of the water spray region. In other words, FLACS will make the water spray region stickto the grid lines. The FLACS code will issue an error message and stop if you have defined anyoverlapping water spray regions. You should check the position and size parameters if this erroroccurs.

3.7.16 Volume fraction

Volume fraction [per thousand] of water, defined by volume of liquid water in litre divided bytotal volume in cubic meter (one cubic meter equals 1000 litre), inside the water spray region.Thus, if the volume fraction is 0 there is no liquid water, and if the volume fraction is 1000 thereis only liquid water. If this parameter is less than 0.01 (corresponds to a mass fraction less than1%), the FLACS code will give you a warning that the water spray is assumed not effective inthe numerical model. As long as the volume fraction of water is larger than this small minimumvalue, the water spray model in FLACS is effective (in the current version of FLACS the valueof the parameter VOLUME_FRACTION does not affect how the water spray model works, so anyvalue larger than the minimum value will give the same numerical results in the simulation).

3.7.17 Mean droplet diameter

The mean diameter [mm] of the water droplets before break-up due to acceleration of the gasflow, is given by the user. In the water spray model it is assumed that all droplets have the samesize, and that the droplets are uniformly distributed in space inside the water spray region. Thisis an approximation. In most real situations there will be a droplet size distribution, which inmost cases is non-uniform in space, i.e. the distribution may change from one region to anotherregion.

In the water spray model, the mean droplet diameter is defined to be the so-called Sauter diam-eter. This diameter is defined by the volume-based mean diameter cubed divided by the area-based mean diameter squared. The Sauter diameter depends on the operating water-pressureforcing water out the nozzle. The empirical relation

D = P−0.333 (3.14)


94 CASD

( D in [mm] and P in [barg]) seems to give good estimates for the Sauter diameter over a range ofvalues for the water pressure, and for various nozzles. However, for some special nozzle typesgiving very large (e.g. nozzle type LDN) or very small (e.g. nozzle type P120) droplet diameters,this empirical relation may be less accurate. One may then try to estimate otherwise the Sauterdiameter for the water-pressure considered. You may use either your estimate or data given bythe vendor of the nozzle.

3.7.18 Nozzle type

The parameter NOZZLE_TYPE should be set equal to the text string "FACTORS: F1 F2" (notethat the double quote character " is included twice in the text string), where F1 and F2 are thenumerical values of the two factors. The two non-dimensional factors F1 and F2 are modelledby

F1 = 14Uzβwater (3.15)

and

F2 =0.03

Dβwater(3.16)

where Uz [m/s] is the average droplet velocity vertically downward (absolute value),βwater[perthousand] is the water volume-fraction, and D [mm] is the Sauter diameter. How to obtain theSauter diameter (mean droplet diameter) is described above. If the nozzle spreads the waterhorizontally, the droplets will soon fall down with a constant velocity (gravity forces are balancedby drag forces). This constant velocity depends on the droplet diameter. It can be estimated fromthe empirical relation

Uz = 2.5D0.94 (3.17)

where the units of Uz is [m/s] and the units of D is [mm]. For some nozzle types the dropletsleave the nozzle with a significant velocity component vertically downward. In this case theaverage downward velocity should be estimated otherwise (giving a larger value than from theexpression above). The water volume-fraction is estimated by

βwater =n(Q/60)

XlengthYlengthZlength(3.18)

where Xlength [m] is the length in x-direction of the assumed rectangular waterspray region,Ylength[m] is the length in y-direction (it is assumed that the xy-plane is the horizontal plane), nis the number of nozzles (it is assumed that all of the nozzles within the same waterspray regionis of the same type and with the same water flow-rate), and Q is the water flow-rate [litre/min]for a single nozzle (thus Q/60 is the water flow-rate in units of [litre/s]). The water flow-ratedepends on the operating water-pressure. It is assumed that the flow-rate is related to the waterpressure P [barg] by

Q = k√

P (3.19)

where the so-called k-value of the nozzle depends on the type of nozzle considered.



3.7.19 Louvre panels

Warning:

The louvre panel model has not been thoroughly validated, and some limitations have beenidentified. Especially the thickness of the louvre slats (section Area porosity) does not seemto have a large enough effect on the results. As the louvre panel model is relatively compli-cated to set up, it is recommended to use a porous plate in most situations.

Louvre panels are common in offshore installations and also in land-based process industry. Thelouvre panels affect the flow field both due to drag forces, and deflection of the flow downstreamof the panel, determined by the geometry of the louvre slats. A new subgrid model for flowthrough louvres is described in detail in [Salvesen, October 1996]. Some validation exercisesusing the new subgrid model are documented in [Salvesen, November 1996].

A louvre panel is assumed to consist of slats mounted on a frame. The louvre slats may have anarbitrary form depending on the type of louvre panel considered, but it is assumed that the slatsof a louvre panel are more or less equal in shape, uniformly oriented, and uniformly distributed.It is assumed that the velocity vector downstream of the louvre panel is forced to be within afixed plane (a mathematical plane of zero thickness) determined by the louvre slats, when theflow exits the louvre, independently of the upstream velocity.

To be specific, let us consider an example: A louvre panel is oriented with normal in x- direc-tion. The louvre slats define a plane with unit normal vector in the xy-plane, (− sin(q), cos(q), 0),where q is an angle with absolute value less than 90°. The angle q is defined as the angle betweenthe slats of the louvre (assuming that the slats are more or less plane, or if they are curved thatthey define a tangent at the exit on the downstream side) and the normal of the louvre (in ourexample the x-axis). The assumption is that the projection of the downstream velocity vector forflow in positive x-direction, on to the xy-plane, is directed a fixed angle q relative the positivex-axis (corresponding to the tangent vector (cos(q), sin(q), 0)), independently of the upstreamvelocity. How good this assumption is in practice depends on several factors: the form (planeor curved) and the thickness of the slats, and how wide the slats are compared to the distancebetween two neighbouring slats. The component of the velocity vector in z-direction, is assumedto be unaffected by the presence of the louvre panel.

In the example above, a louvre panel is oriented in x-direction. It may also be oriented in y- orz-direction. In general the louvre slats define two distinct planes, one plane for flow exiting inpositive direction and one plane for flow exiting in negative direction. Often these two planes areidentical. This is the case e.g. if the slats are of rectangular shape. But if the slats are e.g. V-formed,the two planes are distinct. The two planes may be arbitrary oriented as long as two conditionsare satisfied: Firstly, none of the planes should coincide with the louvre plane itself (this wouldimply that the louvre panel is completely blocked). In the numerical code it is checked that theangle between the unit normal of the louvre plane and the unit normal of the plane determinedby the slats (for flow in positive or negative direction), is not too small, that is less than onedegree. Secondly, if the plane determined by the slats for flow in positive direction is distinctfrom the plane for flow in negative direction, the vector cross product of the normal vectors ofthese two planes should lie in the louvre plane (in other words, these two normal vectors and thenormal of the louvre plane all lie in the same plane).

Note that the subgrid model affects how the momentum flux is calculated for the face of thestaggered control volume adjacent to the louvre plane on the downstream side. The velocity of acontrol volume is determined by a balance of momentum fluxes over all the faces of the controlvolume. The subgrid model gives the momentum flux at the CV face adjacent to the louvrepanel, the momentum fluxes at the other CV faces are determined by the flow field otherwise.A consequence of this is that the first velocity vector downstream of the louvre plane need notbe directed exactly according to the plane determined by the louvre slats, since the flow field


96 CASD

otherwise also affects the velocity vector.

In the subgrid model it is assumed that the total drag force can be represented as the sum ofthree terms; drag due to acceleration of the magnitude of the velocity of the flow, drag due tobending of the flow, and drag due to friction. In the description of these three terms below, it isassumed that the flow is in positive direction (either x-, y- or z-direction). The subscript e (east)corresponds to downstream values, the subscript w (west) corresponds to upstream values. Asimilar description is valid for flow in negative direction (not given here). Drag (absolute value)due to acceleration of the flow is represented by:

FD,acc = Caccmax{

0.05ρe[Ue]2 − 0.5ρw[Uw]2}

(3.20)

in the case that the plane determined by the louvre slats for flow in positive direction is the sameas the one for flow in negative direction. Here ρ is the density, and U is the velocity vector. Thedrag coefficient Cacc is typically set to 1. Note that no contribution from drag due to accelerationis included if the norm of the velocity vector upstream is larger than the one downstream (thiswould correspond to a pressure increase instead of a pressure drop over the louvre).

In the case that the plane determined by the louvre slats for flow in positive direction is distinctfrom the one for flow in negative direction (e.g. if the slats are V-formed), it is assumed that thedrag is represented by:

FD,acc = Cacc

[max

{00.5(ρint[Uint]2 − ρw[Uw]2

) }+ max{

00.5(ρe[Ue]2 − ρint[Uint]2

) }](3.21)

Here Uint is the so-called intermediate velocity vector in the plane determined by the louvre slatsfor flow in negative direction. It is assumed that the velocity vector first is forced to be within thisplane, and then is forced to be within the plane determined by the louvre slats for flow exiting inthe positive direction.

Drag due to bending of the flow is modelled as:

FD,bend = 0.5ρw[Uw]2Cbend(α1 + α2) (3.22)

where α1 is the angle between the upstream and the intermediate velocity vector, α2 is the anglebetween the downstream and the intermediate velocity vector, 0 ≤ αi < π, i = 1, 2. Note thatthe intermediate velocity vector and the downstream velocity vector are identical when the twoplanes determined by the louvre slats (for flow exiting in positive and in negative direction)are identical (and then α2 = 0). The coefficient Cbend (units of 1/radian) could be estimated fromexperiments or fine grid simulations. In the validation simulations a value of Cbend = 0.11/(π/4)is used. This corresponds to a result found in the literature; a resistance coefficient for flow inpipes of 0.11 for a bending of π/4radian. Confer [Salvesen, November 1996] for more details.

If the flow is neither bended nor accelerated, there is still a drag due to skin friction assumed tobe represented by

FD, f ric = 0.5ρw[Uw]2C f ric (3.23)

where C f ric is a drag coefficient.

The coefficients Cacc, Cbend, and C f ric are expected to depend on the specific type of louvre con-sidered, and on both the Reynolds number and the Mach number of the flow. The direction ofthe upstream velocity vector may also be of importance. To investigate how the coefficients de-pend on the various parameters mentioned above, seems to be a challenging task. Performing



fine-grid simulations or experiments are ways of approaching the problem. In the present modelit is assumed that the coefficients depend on only the geometry of the louvre (neglecting thedependence on the other parameters mentioned above).

If experimental values of the coefficients Cacc, Cbend, and C f ric are known for the specific type oflouvre panel considered, these values should be used in the numerical calculations. In manycases only the so-called pressure-loss coefficient (sometimes in the literature it is also calledpressure-drop or resistance coefficient) is known from experiment. This coefficient is definedby

pw − pe = 0.5ρw[Uw]2Cpressure (3.24)

for upstream flow in positive direction with velocity vector pointing normally on the louvreplane. In the case of flow with low Mach number (incompressible or nearly incompressible fluid),the static pressure drop over the louvre panel is essentially balanced by the drag force (i.e. thecomponent of the force from the louvre on the fluid along the normal of the louvre plane) per unitarea (see [Salvesen, October 1996] for details). Thus for incompressible flow, when Uw = Ue, withupstream velocity vector pointing normally on the louvre plane, the pressure-loss coefficient canbe related to the coefficients Cacc, Cbend, and C f ric by

Cpressure = Cacc(1/(cos2 θ)− 1) + Cbendα1 + C f ric (3.25)

where the relation |Ue| cos q = Ue is utilized (confer the example above). Here it is assumed thatthe plane determined by the louvre slats for flow in positive direction is the same as the planefor flow in negative direction. One approach is to set Cacc equal 1 (this value is supported bytheoretical considerations, cf. [Salvesen, October 1996] ), set Cbend equal zero or a guessed valuebased on experimental results found in the literature (e.g. set Cbend = 0.11/(π/4) as describedabove), and set C f ric so that the relation above is satisfied for given values of Cpressure, Cacc, andCbend. Another approach is to set both Cacc and Cbend equal zero, and set C f ric equal Cpressure. Boththese approaches are expected to give good results when performing simulations. If not even thepressure-loss coefficient Cpressure is known, one may use a guessed value for this coefficient

Values of the pressure-loss coefficient are reported in the range from 9.8 to 13.9 for commerciallyavailable louvre panels for industrial use. So a value in the range, say between 10 and 11, is inmany cases expected to be a reasonable estimate for the pressure-loss coefficient.

An example of a setup for a louvre panel is shown below:

NAME "NoName"POSITION 0.0, 0.0, 0.0 (m)SIZE 0.0, 1.0, 1.0 (m)MATERIAL BLUENORMAL_VECTOR_SLATS_POSITIVE 1.0, 0.0, 1.0 (-)NORMAL_VECTOR_SLATS_NEGATIVE 1.0, 0.0, 1.0 (-)DRAG_COEFFICIENT_ACCELERATION 1.0 (-)DRAG_COEFFICIENT_BENDING 0.14 (1/radian)DRAG_COEFFICIENT_FRICTION 9.0 (-)AREA_POROSITY 0.7 (-)

3.7.19.1 NAME

Character string identifying the louvre panel considered (not used by FLACS).


98 CASD

3.7.19.2 POSITION

Cartesian coordinates of the corner of the louvre panel (the corner with lowest value of the coor-dinate in each axis direction).

3.7.19.3 SIZE

The louvre panel is assumed to be of rectangular shape. The dimension in each of the axis direc-tions is given. One dimension should be zero, this shows how the louvre panel is oriented, anddefines the louvre plane. If e.g. the dimension in x-direction is zero, the normal of the louvreplane points in x-direction. The other two dimensions should be positive.

3.7.19.4 MATERIAL

Colour used when visualizing the louvre panel as part of the geometry considered.

3.7.19.5 Normal vector slats positive

Cartesian coordinates of normal vector of plane determined by the louvre slats (ribs) for flowin positive direction. This vector need not be specified as a unit vector. But the normal vectorshould not be parallel to the normal vector of the louvre plane. This would mean that the louvrepanel is completely blocked by the louvre slats (if e.g. the louvre panel is oriented in x-direction,to specify the normal vector determined by the louvre slats as (1.0, 0.0, 0.0) would be an illegalchoice). Note that the normal vector need not lie in the xy-, xz-, or yz-plane. To specify the vectoras for example (1.0, 1.0, 1.0) would be a valid choice.

If for example the louvre panel is oriented in x-direction, and the normal vector determined bythe louvre slats is given by (− sin(q), 0, cos(q)) (here q is an angle with absolute value less than90°), the interpretation is the following: The angle q is the angle between the slats of the louvre(assuming that the slats are more or less plane, e.g. of rectangular shape, or if they are curved thatthey define a tangent at the exit on the downstream side) and the normal of the louvre plane (inthis example the positive x-axis). The tangent vector in the xz-plane of the plane determined bythe louvre slats is given by (cos(q), 0, sin(q)). A typical value of the angle q would be -45°. Thiswould correspond to a case where the louvre panel is a shield for rain (assuming that verticallyupwards is in positive z-direction).

3.7.19.6 Normal vector slats negative

Cartesian coordinates of normal vector of plane determined by the louvre slats for flow in neg-ative direction. Similar comments apply here as those given above for NORMAL_VECTOR_-SLATS_POSITIVE.

In many cases this vector equals the normal vector for flow in positive direction. This is the caseif e.g. the louvre slats have a rectangular shape. But if the louvre slats are e.g. V-formed the planedetermined by the louvre slats for flow in negative direction is different from the plane for flowin positive direction.

Note that the normal vector determined by the louvre slats for flow in negative direction shouldlie in the plane defined by the normal vector determined by the louvre slats for flow in positivedirection and the normal vector of the louvre plane. If for example the louvre panel is oriented inx-direction and the normal vector of the plane determined by the louvre slats for flow in positive



direction lies in the xz-plane, then the normal vector for flow in negative direction should also liein the xz-plane.

3.7.19.7 Drag coefficient acceleration

Drag coefficient Cacc of drag due to acceleration of the magnitude of the velocity of the flow. Thiscoefficient should be zero or positive. A typical value is 1 (this value is supported by theoreticalconsiderations). Further guidance is given above.

3.7.19.8 Drag coefficient bending

Drag coefficient Cbend of drag due to bending of the flow. This coefficient should be zero orpositive. The units of this coefficient is (1/radian). Further guidance is given above.

3.7.19.9 Drag coefficient friction

Drag coefficient C f ric of drag due to friction. This coefficient should be zero or positive. Furtherguidance is given above.

3.7.19.10 Area porosity

An area porosity of the louvre panel is specified. This is not a projected area porosity on to thelouvre plane (in many cases it is not possible to see through the louvre watching normally on it,i.e. the projected area porosity is zero), but is viewed as the ratio of open area to total area ina representative cut plane parallel to the louvre plane. If e.g. the louvre consists of rectangularslats of thickness D uniformly inclined relative to the louvre plane and uniformly spaced withcentre/centre distance 4D, the area porosity is (4D − D)/4D = 0.75 (this value is independentof the angle of inclination). If the frame of the louvre blocks the panel considerably, this may betaken into account. If the blockage of the frame corresponds to an area porosity β f rame, the totaleffective area porosity in our example will be 0.75β f rame

Note that in the subgrid model for louvre panels, the value of the area porosity affects the dragforce only indirectly. The drag force is determined by the drag coefficients, the configuration ofthe louvre slats (determining the deflection of the flow downstream of the louvre), and the flowfield. The value of the area porosity affects how the fluxes (of mass, momentum, etc.,) are setwhen solving the conservation equations at the louvre plane, and this will affect the flow field.

If the profile of the louvre slats has a complicated form, it may not be obvious how to estimatethe effective area porosity. If the area porosity varies in different cut-planes parallel to the louvreplane, it is expected that a cut-plane with a minimum area porosity is the most representative.The user may perform several simulations varying the value of the area porosity, if one wants toinvestigate the sensitivity of the value of the area porosity on the numerical results. The limitedtesting in the validation simulations reported in [Salvesen, November 1996], seems to indicatethat the maximum overpressure of an explosion simulation is in general not very sensitive tochanges in the value of the area porosity. In one scenario the maximum overpressure droppedabout 12% when the value of the area porosity was increased from 0.6685 to 1.0.


100 CASD

3.7.20 Grating

Warning:

The grating model should only be used in dispersion and ventilation simulations. The cur-rent model is designed for flow through the grating, but the important effect on flame accel-eration along the grating is not handled correctly. It is recommended to use a porous plate inexplosion simulations.

The static pressure-loss coefficient Cspl is defined by

∆p = 0.5ρ[U]2Cspl (3.26)

where ∆p is the pressure loss across the grating, ρ is the density, and U is the upstream velocitycomponent normal to the grating. The component normal to the grating of the force per unit areafrom the grating on the fluid flowing through it, i.e. the drag force per unit area, is assumed to beequal to the static pressure drop over the grating. The static pressure-loss coefficient is modelledby:

Cspl = f1(Re) f2(β) f3(M) (3.27)

where the factor depending on the geometry, characterized by the area porosity β , is given by:

f2(β) =(1− β2)

β2 (3.28)

and the factor depending on the upstream Mach number M, and the upstream choking Machnumber M∗ (upstream Mach no. corresponding to choking at the grating) is given by:

f3(M) =(

M∗

M∗ − M

)1/7(3.29)

The choking Mach number is related to the area porosity, M∗ = M∗(β). This relation is modelledby:

M∗(β) = 0.675β + 0.325β4 (3.30)

being a curve fit to experimental data.

The factor depending on the Reynolds number, Re, is modelled as a curve fit to experimentalresults. The characteristic length used in the Reynolds no. is the diameter of the wire (or ofthe rod if the grating is constructed by rods rather than wires). This curve fit and experimentaldata are shown in the figure below. Further details about modelling drag forces for flow throughgrating, are found in [Salvesen & Storvik, 1994].



Figure 3.14: Factor f1 related to Reynolds number

Factor f1(Re) in pressure-loss coefficient as function of Reynolds number (logarithmic scale).Both experimental data and a curve-fit to these data are shown.

An example of a setup for a louvre panel is shown below:

NAME "NoName"POSITION 0.0, 0.0, 0.0 (m)SIZE 0.0, 1.0, 1.0 (m)MATERIAL REDAREA_POROSITY 0.7 (-)CHARACTERISTIC_LENGTH 0.01 (m)

3.7.20.1 Name

Character string identifying the grating considered.

3.7.20.2 Position

Cartesian coordinates of the corner of the grating (the corner with lowest value of the coordinatein each axis direction).


102 CASD

3.7.20.3 Size

The grating is assumed to be of rectangular shape. The dimension in each of the axis directionsis given. One dimension should be zero, this shows how the grating is oriented. If e.g. thedimension in x-direction is zero, the normal of the grating points in x-direction. The other twodimensions should be positive.

3.7.20.4 Material

Colour used when visualizing the grating as part of the geometry considered.

3.7.20.5 Area porosity

Ratio of projected open area of the grating divided by total area. Value between 0 and 1.

3.7.20.6 Characteristic length

Characteristic length used in the Reynolds number, defined as the wire diameter (or similar di-mension if it is not a wire, but a rod). A typical value of the characteristic length is 0.01m.

3.7.21 Gas monitor region

The GAS_MONITOR_REGION function is used for monitoring amount of fuel inside a moduleor another user defined volume during a dispersion simulation. FLACS will write a text filecalled rt010100.FUEL which contains a number of columns showing the amount of fuel insidethe defined volume. The columns Q8, Q9, [LFL:UFL], Q6/Q7 are used for risk assessments inconnection to dispersion studies and estimate of explosion severity and ignition probabilities.

If more than one region is needed the user must create a cs010100.MON file. Please see sectionMonitor file for details.

3.7.22 Species

Gases not defined in FLACS can be defined manually. When defining a gas, a long list of vari-ables must be defined. GexCon plan to provide templates for some of the typically used gaseslater (e.g. ammonia and chlorine), however, in the meantime the user must define these gases. Toquickly increase the molecular weight of a gas, for a simulation in which pressure and tempera-ture does not change too much, simply copy the properties of another defined gas when definingthe USERSPEC_1 gas.

These properties can be found by running a utility program called listspecie.

Linux:

> run9 listspecie1.0 BUTANE

Windows:

>listspecie1.0 BUTANE


3.8 Block menu 103

To define chlorine gas for a dispersion calculation (ignoring any kind of aerosol and condensa-tion) simply modify the WFUEL from 0.581e2 to 0.709e2 when specifying the USERSPEC_1 fuelin SPECIES menu of CASD. If proper compression and change of temperature is needed, mod-ification of the HFUEL, AENT, BENT and DENT values is also required (this is more complicated).Several of the parameters have to do with liquid properties of the fuel, this functionality is notyet available and these are thus irrelevant.

3.8 Block menu

Analyzing the effects of far-field pressure waves can be of interest in many applications. Themulti-blocks option in FLACS allows performing far-field pressure waves, i.e. blast waves, stud-ies. A description of the settings and running commands for the multi-blocks simulation inFLACS is given here.

3.8.1 Defining a multiblock grid

Assuming a given geometry in the pre-processor CASD, a multiblock grid can be defined. Bydefault, in the sub-menu Select of the menu Block in the CASD tool bar, a block called super anda block called FLACS are defined. The block called super needs to be selected to further definethe multi-block grid. The block named super will be referred to as SuperGrid in the following.The extents of the SuperGrid domain (menu Grid and Simulation_Volume) defines the total size ofthe simulation domain and the number of control volumes in the SuperGrid gives the number ofblocks.

The block FLACS can be selected in the sub-menu Select of the menu Block. The position of thisblock into the SuperGrid can be defined in the sub-menu Properties of the menu Block by specifyingthe appropriate indices i,j,k of a SuperGrid cell. A grid relative to the block FLACS selected can bedefined in the usual way going in the sub-menu Region of the menu Grid. The following pictureshows a grid with four different blocks:


104 CASD

Figure 3.15: A multi-block grid

Two different types of blocks exist:

• FLACS• BLAST

Setup parameters for the simulation (ignition position, initial conditions,...) can only be definedfor a FLACS block. The ignition region must be contained in a single FLACS block and must notbe divided into several blocks.

The BLAST blocks are used to study the development of the pressure front only. The computa-tions in a BLAST block are faster and require less memory than in a FLACS block. Combustionprocesses should not occur in a BLAST block. Therefore, extra FLACS blocks might need to bedefined to contain the whole combustion area.

A block can be added through the sub-menu Add of the menu Block. The position in the SuperGridneed to be defined. BLAST blocks are added by default. The type of block for the selected block,can be changed in the sub-menu Properties of the menu Block. Finally, selected blocks can also bedeleted through the sub-menu Delete of the menu Block.

3.8.2 Running a multiblock simulation

Saving the multiblock setup generates a SuperJob number and one job number for each of theblocks that have been defined. Once the multiblock setup has been saved, CASD can be exitedand the porosities computed for each of the job numbers:

Linux:

> run9 porcalc 50000i


3.9 View menu 105

Windows:

> porcalc.exe 50000i

where 50000i is the job number relative to the block i.

Assuming 500000 is the SuperJob number corresponding to the considered multiblock simulation,the following command starts the simulation

Linux:

> run9 flacs setup-500000

Windows:

> flacs setup-500000

The file setup-500000 contains a list of job numbers and corresponding block type relative to theSuperJob number 500000:

VERSION 1.1$JOBSPEC

BLOCKS ="500001" "FLACS""500002" "BLAST""500003" "BLAST""500004" "BLAST""500005" "BLAST"

$END

3.8.3 General guidelines for multi-block simulations

The grid resolution in the explosion block should be reasonably fine, but it might be necessary touse a coarser grid than the one generally used in explosion simulations. The block faces must fitperfectly to their neighbors

The geometry should be as simple as possible at the block boundaries, if not the porosity patternmight become different on the two sides of the boundary.

In BLAST blocks the porosity is either 0 or 1. For stability, the CFLC number should be 0.5 in theBLAST blocks as well as in FLACS blocks due to the explicit coupling between the blocks.

In BLAST blocks the PLANE_WAVE boundary condition should be used (if not SYMMETRY). ThisPLANE_WAVE condition should also be used in the FLACS blocks to minimize the influence ofthe boundaries.

Both monitor point output and field output (r1- and r3-files) can be generated from a multiblocksimulation. For 2D field plots and line plots several blocks can be shown in one plot in Flowvis.However, for volume plots only one block can be shown for each plot.

With BLAST blocks only a selection of the variables of the FLACS blocks exists. Variables likePROD and FUEL should not be specified for output in BLAST blocks.

3.9 View menu

The View menu in CASD contains commands for manipulating the view.


106 CASD

3.9.1 Print

The Print menu allows exporting a screenshot of the CASD window into different formats:

• Postscript• RGB• IV

3.9.2 Examiner viewer and Fly viewer

The default and most widely used viewer is the Examiner viewer. The Fly viewer can be used tofly through the geometry.

3.9.3 XY View, XZ View, and the YZ Views

The option XY View and XZ View display a projection of the geometry in the XY and XZ planesrespectively. The options YZ East View and YZ West View display a projection of the geometryin the YZ plane along the positive and negative Y-axis respectively.

3.9.4 3D View

The 3D View option displays a default 3D view of the geometry.

3.9.5 Axis

The Axis option turns axis display on and off.

3.9.6 Maximise

The option Maximize maximizes the visible window to display the entire geometry and grid.

3.9.7 Grid display

Three different options are available in the Grid Display menu:

• Off: The grid is not displayed. Only the geometry would be displayed.• Working Direction: The grid would be displayed in the working direction only.• All Directions: The grid would be displayed in the three directions.

3.9.8 Annotation

The options in this menu are currently not used.


3.10 Options menu 107

3.9.9 Draw style

Different options are available in this menu:

• Off: The geometry will not be displayed.• Wireframe: Only the edges of the objects that compose the geometry would be displayed.• Filled: Surfaces of the objects that compose the geometry would be displayed.• Scenario Wireframe: Only the edges of scenario objects (for example, a fuel region) would

be displayed.• Scenario Filled: Surfaces of scenario objects would be displayed.

3.9.10 LOD and Properties

The LOD (Level Of Details) and properties menus control the details of the geometry displayed.

3.10 Options menu

The user may select certain options under the Options menu in CASD.

3.10.1 Units

The user may choose between the following units for the spatial dimensions: millimetres (mm),centimetres (cm), decimetres (dm), meters (m), and inches (in); the default option is meters.

3.10.2 Preferences

The users may set preferences for:

• General features: scenario template and version of Porcalc• Colours: background colours for examiner viewer and fly viewer• Performance: redraw options

3.10.2.1 General

The user may chose between the available scenario templates in a drop down menu. It is essentialthat the selected scenario template match the FLACS version to be used in the simulations.

Warning:

New scenario templates will not necessarily work with old scenario files.

By ticking off the option ’Write polygons (cm file) when saving’, the polygons file read by Flowviswill not be written. In case of a very complex geometry, ticking off this option allows Flowvisloading the result files faster than if the cm-file were existing. The option ’Write polygons (cmfile) when saving’ should therefore be ticked off if the geometry is very complex. See also sectionPolygon file.


108 CASD

3.10.2.2 Colours and Performance

Through these two menus the CASD window can be customized.

3.11 Macro menu

The Macro menu contains commands for running and recording macros.

• Run: This command processes all the commands on a specified file before turning controlover to the user again. If an error occurs, the processing is interrupted.

• Record: This is a toggle button for turning command recording on/off. When turningrecording on, the command requires a macro file name. All subsequent commands arerecorded on the specified file, until the recording is turned off.

• Write Geometry: This command writes the macro files needed to define the open geometry,including global objects and materials. CASD asks for the path to a directory where thefiles are placed. See section and for more information.

3.11.1 Run

To create a geometry from a set of macro files, use the Run command in the Macro menu. Alter-natively use the command input to read the macro file geometry_name.mcr: ∗ macro rungeometry_name

If the project or geometry already exists in the database, an error message is displayed and CASDexits from the macro.

Macro Descriptiongeometry_name.mcr Creates a new geometrygeometry_name_materials.mcr Creates all materials used in the geometrymaterial_name.mcr Creates material (one file for each material)geometry_name_objects.mcr Creates all objects used in the geometryobject_name.mcr Creates object (one file for each object)geometry_name_instances.mcr Creates all assemblies/instance

Table 3.7: Macro files

If some objects or materials on the macro files already exist in the database, an error message isdisplayed, and the object/material is not overwritten.

3.11.2 Record

The option Record is used to save a macro in a 000000.caj.mcr file, where 000000 is a given jobnumber. The macro file is written simultaneously as the user use CASD, thus this function actslike a log of the performed actions.

This file can be read as executing a macro with the option Run.


3.12 Help menu 109

3.11.3 Write geometry

The Write Object command in the Macro menu writes a macro file that defines the open object.CASD asks for the path to a directory where the macro file is to be placed.

Note that this macro file must be started in the CASD main window.

The Write Geometry command in the Macro menu causes CASD to write a complete set of macrofiles for the open geometry. The files include macro files which create the project, geometry, allmaterials needed, all objects needed in addition to the assemblies/instances. The files are listedin table Macro files created by the Write Macro command.

Figure 3.16: The macro file hierarchy

Note that the macro file format is not intended as a backup format. Future versions of CASD maynot be backwards compatible with the menu structure and commands in the current version.

The Copy command in the Database menu can be used to make a copy of a geometry within thesame project. The macro files created by the Write Geometry command in the Macro menu can beused for copying the geometry from one database to another. They can also be used for copyingone geometry into another (existing) geometry.

To copy one geometry (geo1) into another (geo2) in the same database, open geo1 and executethe Write Geometry command in the Macro menu. Exit from geo1 and open geo2. Create a newassembly and execute the macro:

* macro run geometry_name_instances

Problems occur if geo2 contains local object(s) with the same name(s) as in geo1.

3.12 Help menu

The purpose of the Help menu is to provide the user with relevant information concerning thegeneral use of FLACS and the active FLACS licence.


110 CASD

3.12.1 Online help

This menu opens this User’s manual in the default HTML browser on the computer (see theOptions menu for information on how to change the default HTML browser).

3.12.2 Quick reference

This menu opens a window summarizing various controls

Figure 3.17: Quick References

3.12.3 Licence terms

This menu opens a window that contains the FLACS licence terms.

3.12.4 About CASD

This menu displays the FLACS splash screen with information about the version of FLACS andthe version of CASD.

3.13 Potential bugs or problems with CASD

This chapter contains a list of potential bugs or problems with CASD, and some possibleworkarounds.


3.13 Potential bugs or problems with CASD 111

3.13.1 Problem with macro files and local objects

Using the WRITE GEOMETRY option in CASD does not correctly write local objects. The trans-formation matrix is missing, resulting in a faulty geometry when reading macro files back intoCASD.

3.13.2 CAD import

Geometry can be imported from CAD programs. Please refer to section geo2flacs.

3.13.3 Heavy hydrocarbons C5H12 and upwards

Several vapours from hydrocarbons that are heavier than butane can be modelled, but combus-tion properties have been copied from butane. Enthalpies for Dodecane have been copied fromDecane, and this may lead to strange results if burning Dodecane.

3.13.4 Reading old dump file

In previous FLACS versions the freezing point of water (0 °C) was defined to be 273 K, whereasin the newest version this has been changed to 273.15 K. For wall temperatures it has always beenassumed that 0 °C translates to 273.15 K. The gas constant, <, has similarly been changed from8314 J/(kmol K) to 8314.472J/(kmol K).

If an old dump-file is read in the new version of FLACS, small inconsistencies/disturbances maybe seen in the temperature field of this reason.


112 CASD


Chapter 4

Flacs simulator

114 Flacs simulator

This chapter describes various aspects the CFD simulator Flacs:

• how to start and stop a simulation

• how to monitor the progress of a running simulation

• a list of input files

• a list of output files

• a list of output variables

4.1 Overview

See Getting started for a detailed description of how to install FLACS and the basic steps to getstarted using it.

On Linux it is recommended that the user defines an alias for running FLACS v9.0 programs:

> alias run9 /usr/local/GexCon/FLACS_v9.0/bin/run

On Windows a desktop icon for the Run Manager is created during the installation.

4.2 The Run Manager

Linux:

> run9 runmanager

Windows:

> "C:Program Files\GexCon\FLACS_v9.0\bin\runmanager"


4.2 The Run Manager 115

Figure 4.1: The FLACS Runmanager

4.2.1 Starting simulations

See Running FLACS for the basic introduction on how to get started using the FLACS package.

The FLACS simulator (Flacs) can be started from the command line or from the Runmanager.

The command to start a FLACS simulation from the command line:

> run9 runflacs 010101

4.2.2 Monitoring simulations

On Linux the user can use the tail command to monitor the progress of a FLACS simulation.

> tail -f tt010101


116 Flacs simulator

4.2.3 Check list for running simulations

The sequence of tasks involved in a general FLACS simulation includes:

1. Stating your problem2. Defining sensitivities, or parameter variation3. Defining and verify the geometry4. Defining and verify the grid5. Calculating and verify the porosities6. Defining and verify the scenario7. Running the simulations8. Checking the simulation log files for errors9. Presenting the results

10. Storing all data for later use

Remarks:

It is important to check the correctness of all input parameters.

Below is a recommended check list for basic QA of the simulation set-up:

1. Avoid large Courant numbers (CFLV and CFLC)2. Locate ignition in an unblocked control volume3. Locate monitors in unblocked control volumes4. Define realistic discharge parameters for leaks5. Verify vent areas6. Verify gas composition7. Avoid strong transient wind build-up8. Check disk space and access rights9. The required files:

• Grid file• Obstruction file• Porosity file• Scenario file

4.3 Running several simulations in series

An efficient way to handle many simulations is to use run scripts (text file with commands).Running many simulations in parallel may exhaust the computer memory and actually increasethe total computation time for the simulations. On Linux if the user wants to stop a runningprocess the kill command can be used, the ps command will report the process id of all runningprocesses. On Windows the Task Manager can be used.

Remarks:

See section Linux Quick Reference for useful Linux commands and examples on how to runand monitor FLACS commands effectively on a Linux system.


4.3 Running several simulations in series 117

4.3.1 Create and use run scripts

Flacs simulations are easy to run in batch, in parallel or series, using the Run mananger, howeverif the user wants more control a run script can be created using a text editor, the script file my_-runfile could look like this, on Linux:

#!/bin/csh -f

# Set up an alias for running the FLACS simulator:alias my_runflacs /usr/local/GexCon/FLACS_v9.0/bin/run_runflacs

# Run the simulations in series:my_runflacs 010101my_runflacs 010102my_runflacs 010103

Make the script file my_runfile executable:

> chmod u+x my_runfile

Run the script in the background, messages are sent to the file my_listfile:

> ./my_runfile >& my_listfile &

Similarly a bat script, my_runfile.bat can be created on Windows:

@echo offsetlocal

rem Set up an alias for running the FLACS simulator:set my_runflacs="c:\program files\gexcon\flacs_v9.0\bin\runflacs.exe"

rem Run the simulations in series:%my_runflacs% 010100%my_runflacs% 010101%my_runflacs% 010102

Run the script, messages are sent to the file my_listfile:

> my_runfile.bat > my_listfile

4.3.2 Optimizing computer loads

In general Flacs simulations will use 100% of a single CPU core if available, thus optimal use of acomputer for runninig simulations is to start as many simulations in parallel equal to the numberof CPU cores available. It is however necessary to keep the total memory consumption withinphysical computer memory. If this is not done the computer will start to use virtual disk memory,which is significanly slower than physical memory. This will result in longer simulation time.

As a rule of thumb a simulation computer should have 2GB of memory per CPU core.

4.3.3 Stopping simulations

Flacs simulations can be stopped prematurely either by using the Task Manager (Windows) orusing a combination of the command line utilities ps and kill on Linux.

To find the Flacs simulation process ID (PID) run the following command on Linux:


118 Flacs simulator

> ps -edlaf|grep flacs

This will list all FLACS related programs running on the computer. The PID is the number foundin the 4th column. Also note the user name of the process, which is found in the 3rd column. Theprocess can then be stopped using the following command:

> kill 1234

where 1234 is the PID.

Alternatively the simulation can be stopped by using the Runtime simulation control file (cc file).Adding the following line to the cc file will stop the simulation at the time 123.4 sec:

TSTOP 123.4

4.4 Output variables in FLACS

This chapter describes the output variables that can be selected for output in selected monitorpoints or over selected monitor panels (under ’Single Field Scalar Time Output’ on the Scenariomenu in CASD), or throughout the entire calculation domain (under ’Single Field 3D Output’ onthe Scenario menu in CASD).

The normal output variables in FLACS are:

Name Dim Units DescriptionH 1 (J/kg) EnthalpyFUEL 1 (-) Fuel mass fractionFMIX 1 (-) Mixture fractionFVAR 1 (-) Mixture varianceK 1 (m2/s2) Turbulent kinetic

energyEPK 1 (1/s) Turbulence ratioEPS 1 (m2/s3) Dissipation rate of

turbulent kineticenergy

GAMMA 1 (-) Isentropic gasconstant

LT 1 (m) Turbulent lengthscale

MU 1 (kg/(m∗s)) Effective dynamicviscosity

OX 1 (-) Oxygen massfraction

P 1 (barg) PressurePMAX 1 (barg) Maximum pressureDPDT 1 (bar/s) Rate of pressure risePIMP 1 (Pa∗s) Pressure impulsePIMPMAX 1 (Pa∗s) Maximum pressure

impulsePROD 1 (-) Combustion product

mass fractionRFU 1 (kg/(m3∗s)) Combustion Rate


4.4 Output variables in FLACS 119

RET 1 (-) Turbulent Reynoldsnumber

FMOLE 1 (m3/m3) Fuel mole fractionFDOSE 1 (m3/m3∗s) Fuel mole fraction

DOSERHO 1 (kg/m3) DensityT 1 (K) TemperatureTURB 1 (m/s) Turbulence velocityTURBI 1 (-) Relative turbulence

intensityVVEC 3 (m/s) Velocity vectorU 0 (m/s) Velocity component

x-directionV 0 (m/s) Velocity component

y-directionW 0 (m/s) Velocity component

z-directionUVW 1 (m/s) Velocity valueUDRAG 1 (Pa) Drag component

x-directionVDRAG 1 (Pa) Drag component

y-directionWDRAG 1 (Pa) Drag component

z-directionDRAG 1 (Pa) Drag valueDRAGMAX 1 (Pa) Maximum drag

valueUDIMP 1 (Pa∗s) Drag-impulse

componentx-direction

VDIMP 1 (Pa∗s) Drag-impulsecomponenty-direction

WDIMP 1 (Pa∗s) Drag-impulsecomponentz-direction

DIMP 1 (Pa∗s) Drag-impulse valueDIMPMAX 1 (Pa∗s) Maximum

drag-impulse valueUFLUX 1 (kg/(m2∗s)) Flux component

x-directionVFLUX 1 (kg/(m2∗s)) Flux component

y-directionWFLUX 1 (kg/(m2∗s)) Flux component

z-directionFLUX 1 (kg/(m2∗s)) Flux valueUMACH 1 (-) Mach number

componentx-direction

VMACH 1 (-) Mach numbercomponenty-direction


120 Flacs simulator

WMACH 1 (-) Mach numbercomponentz-direction

MACH 1 (-) Mach number valueCS 1 (m/s) Sound velocityTAUWX 1 (-) Wall shear force

tauwxTAUWY 1 (-) Wall shear force

tauwyTAUWZ 1 (-) Wall shear force

tauwzNUSSN 1 (-) Nusselt numberRESID 1 (-) Mass residual in

continuity equationER 1 (-) Equivalence ratioERLFL 1 (-) Equivalence ratio,

LFLERNFL 1 (-) Equivalence ratio,

normalizedflammable range

EQ 1 (-) Equivalence ratio,finite bounded

EQLFL 1 (-) Equivalence ratio,LFL

EQNFL 1 (-) Equivalence ratio,normalizedflammable range

TMOLE 1 (m3/m3) Toxic mole fractionTCONS 1 (mg/m3) Toxic concentrationTDOSE 1 (mg/m3∗minute) Toxic dosePROBIT 1 (-) Toxic probitPDEATH 1 (-) Probability of death

as function of toxicprobit

Table 4.1: Output variables in FLACS

The panel output variables in FLACS are:

Name Dim Units DescriptionPPOR 1 (-) Panel average area

porosityPP 1 (Pa) Panel average

pressurePPIMP 1 (Pa∗s) Panel average

pressure impulsePDRAG 1 (Pa) Panel average dragPDIMP 1 (Pa∗s) Panel average drag

impulseTable 4.2: Panel output variables in FLACS

The following sections gives a description of the most commonly used output variables.



4.4.1 Mass fraction of fuel: FUEL

This is the mass fraction of fuel in the mixture of fuel, air, and combustion products. The fuelmay be a mixture of several components, such as hydrocarbons and hydrogen. Plots of FUEL areuseful for displaying the fuel cloud.

4.4.2 Pressure: P

This is the static overpressure (bar g). In a flowing fluid, the total pressure is the sum of thestatic pressure and the dynamic pressure. The static pressure is isotropic, whereas the dynamicpressure, caused by the relative motion of the fluid, is anisotropic. A pressure transducer placedin a flow field will in general measure the static pressure and a certain portion of the dynamicpressure, depending on the orientation of the face of the pressure transducer relative to the flowdirection. ’Head on’ measurements give the total pressure, whereas ’side on’ measurements givethe static pressure.

4.4.3 Pressure impulse: PIMP

PIMP is the time integral of the pressure:

Ip =∫ t2

t1pdt (4.1)

The pressure impulse is simply the area below the pressure-time curve, and since it is the prod-uct of pressure and time it holds information about both the amplitude and the duration of thepressure-time curve.

4.4.4 Mass fraction of combustion products: PROD

This is the ratio of mass (kg) of combustion products per unit mass (1 kg) of the total mixtureof fuel, air and combustion products for each control volume. The combustion products consistof carbon-dioxide and water vapour. Plots of PROD are useful for displaying the flame (or morecorrectly the burnt volume).

See sections Definitions and gas thermodynamics, Stoichiometric reaction and Gas compositionand volume for more information on the reactions that convert fuel onto combustion products.

4.4.5 Gas density: RHO

The gas density is:

p = ρRT (4.2)

This is the fluid mass (kg) per unit volume (1 m3). The equation of state gives the relation betweenpressure density and temperature.


122 Flacs simulator

4.4.6 Gas temperature: T

This is the absolute temperature (K) of the fluid. See RHO above for a description of the relationbetween pressure density and temperature. The temperature may be increased by compressionwhich converts mechanical energy into thermal energy, and by combustion which converts chem-ical energy into thermal energy.

4.4.7 Velocity vector: VVEC

This is the entity which gives the three velocity components of the time averaged fluid flow. Theenergy contained in the temporal fluctuations of the flow which are not captured using a givenspatial and temporal resolution is handled by a turbulence model. VVEC consists of the threecomponents U, V, and W. If you are editing the cs-file manually, always remember to include thecomponents if you have specified VVEC for output (CASD includes them automatically).

4.4.8 Drag value: DRAG

The drag value (in FLACS defined as drag force per unit area) is proportional to the dynamicpressure for the fluid flow. The expression for the dynamic pressure is:

pdyn = ρu2/2 (4.3)

An obstacle submerged in a fluid flow will interact with the fluid, thereby a drag force results.The drag force may be measured in experiments and if the Reynolds number is high, the ratio’drag force / dynamic pressure’ is constant:

Drag coefficient CD = (FD/A)/(ρu2/2)

Drag force FD = CD(ρu2/2)A

The drag value is calculated by assuming the drag coefficient CD = 1 and the cross-section areaA = 1 , with this definition the drag value is the same as the dynamic pressure.

4.4.9 Drag-impulse value: DIMP

Drag-impulse value is the time integral of the dynamic pressure:

IPdyn =∫ t2

t1pdyndt (4.4)

The drag-impulse value is equivalent to the pressure impulse, with the difference that the dy-namic pressure is being integrated instead of the static pressure.

4.4.10 Equivalence ratio: ER

The equivalence ratio, ER, is a measure of the concentration of fuel compared to the stoichiomet-ric concentration, i.e. ER equals unity at stoichiometric concentration. If (F/O) is the ratio of fuelto oxygen, the equivalence ratio is defined as follows:

ER = (F/O)/(F/O)stoichiometric (4.5)



For zero fuel, ER equals zero and for pure fuel ER goes to infinity.

4.4.11 Equivalence ratio %LFL: ERLFL

This is a measure for concentration of fuel compared to the LFL concentration, where LFL is thelower flammable limit. The LFL value normally varies with gas type and oxygen concentration(again depending on the amount of inert gases) in the mixture. In FLACS the fuel is alwaysmixed with air which has a preset oxygen concentration, so only the variation of LFL with gastype remains. The definition of ERLFL is as follows:

ERLFL = 100 ∗ ER/ERLFL% (4.6)

4.4.12 Equivalence ratio, normalized DFL: ERNFL

The flammable range is defined to be from LFL to UFL, where LFL is the lower flammable limitand UFL is the upper flammable limit. ERNFL is defined as follows:

ERNFL = (ER− ERLFL)/(ERUFL− ERLFL) (4.7)

ERNFL is zero at LFL and one at UFL.

4.4.13 Equivalence ratio, finite bounded: EQ

This is a measure for concentration of fuel similar to the equivalence ratio (see ER above). Saythat (F/O) is the ratio of fuel to oxygen, then the finite bounded equivalence ratio is defined asfollows:

EQ = (F/O)/[(F/O) + (F/O)stoichiometric] (4.8)

At stoichiometric concentration EQ equals 1/2. For zero fuel EQ equals zero and for pure fuel EQequals one.

4.4.14 Equivalence ratio, %LFL: EQLFL

This is a measure for concentration of fuel compared to the LFL concentration, where LFL is thelower flammable limit. The LFL value normally varies with gas type and oxygen concentration(again depending on the amount of inert gases) in the mixture. In FLACS the fuel is alwaysmixed with air which has a preset oxygen concentration, so only the variation of LFL with gastype remains. The definition of EQLFL is as follows:

EQLFL = 100 ∗ EQ/EQLFL% (4.9)


124 Flacs simulator

4.4.15 Equivalence ratio, normalized DFL: EQNFL

Equivalence ratio, normalized flammable range. The flammable range is defined to be from LFLto UFL, where LFL is the lower flammable limit and UFL is the upper flammable limit. EQNFL isdefined as follows:

EQNFL = (EQ− EQLFL)/(EQUFL− EQLFL) (4.10)

EQNFL is zero at LFL and one at UFL.

4.4.16 Panel average pressure: PP

This is the average pressure (Pa) acting on the panel surface in the perpendicular direction. It isthe sum of the directional pressure forces acting on the panel divided by the net surface area ofthe panel (also accounting for the area porosity for each control volume). The sign of PP indicatesthe direction of the total force, +/- along the positive/negative direction respectively.

The panel average pressure PP is calculated in the following way: For each control volume face(CV face) covering the area of the panel considered, the net static pressure is calculated (staticpressure on the negative side relative the coordinate axis minus the static pressure on the positiveside). This net static pressure is then integrated over the blocked area of the panel and the integralis then divided by the total blocked area of the panel, to give the panel average pressure PP. Ife.g. the area porosity of the panel is zero, the panel is totally blocked and the integral is dividedby the total area of the panel. And if e.g. the panel is totally open (area porosity one, no blockedarea), panel average pressure PP vanishes (is zero), since there is no blocked area to integrateover (integral with respect to net static pressure over blocked area is zero). In general the panelcan be porous (partially blocked and partially open).

4.4.17 Panel average porosity: PPOR

This is the average pressure porosity, it is the amount of open surface on the panel divided by thetotal panel area. Output of PPOR may be used to verify when the panel yields.

4.4.18 FMOLE and FDOSE

FMOLE is the mole, or volume, fraction of the gas in the gas/air mixture, and FDOSE is the inte-grated (accumulated) FMOLE. For 60s dose for monitor points, you can simply export FMOLE toASCII-format using r1-file, import to excel, and subtract the FDOSE value of time-60s. The FDOSEvariable can also be selected for 3D output. If the user wants a contour plot of the 60s exposureone or more plots must be selected for each 60s period. Using the r3file utility program one canthen generate new r3-files with FDOSE(time)-FDOSE(time-60s).

The utility programme r3file can generate the so-called dose (i.e. exposure) output:

FDOSE(t) =∫ t

0 FMOLE(t)dt

dose(t2) = FDOSE(t2)− FDOSE(t1)

(dose/time)(t2) = dose(t2)/(t2− t1)

The times (t2 and t1) are taken from the output times with a certain integer interval given by theoption ’dose=’ or ’dose/time=’:



dose=2 means that t2-t1 = 2∗DTPLOT, output is then dose(t)

dose/time=2 is similar, but you get (dose/time)(t) output

The following starting point is assumed:

• FLACS result files in the current directory, and• FDOSE output at regular time intervals (e.g. DTPLOT = 60)

To generate the dose (i.e. exposure) output in a separate directory:

1. create the directory ’work’ and enter into it2. run the r3file utility (assuming the job number is 010100)

> mkdir work> cd work> run9 r3file1.3 ../r3010100.dat3 format=r3file dose=1 name=NFDOSE force

Now the following files in the work directory can be found:

a3010100.NFDOSEcgNFDOSE.dat3 -> ../cg010100.dat3coNFDOSE.dat3 -> ../co010100.dat3cpNFDOSE.dat3 -> ../cp010100.dat3csNFDOSE.dat3r3NFDOSE.dat3 -> a3010100.NFDOSE

The results can be viewed using Flowvis.

Warning:

’region=’ cannot yet be used with the dose output.

4.4.19 Variables for toxic substances

It is possible to model the effect of toxic substances with FLACS. The toxic component is specifiedin the section GAS_COMPOSITION_AND_VOLUME on the scenario menu.

Remarks:

The toxic substance variables are only available using the default+1 scenario template.

The parameter TOXIC_SPECIFICATION can be specified in one of the following ways:

• Selecting a predifined substance• Creating a user specified toxic data file, and specifying the substance name• Specifying the substance properties directly

The following predefined substances are available:


126 Flacs simulator

Substance a b n Formula Molarmass(g/mol)

Boilingpoint (°C)

Acrolein -4.1 1 1 C3H4O 56.06 53

Acrylonitrile-8.6 1 1.3 C3H3N 53.06 77

Allylalcohol

-11.7 1 2 C3H6O 58.08 97

Ammonia -15.6 1 2 NH3 17.0306 -33.34

Azinphos-methyl-4.8 1 2

C10H12N3O3PS2317.32 200

Bromine -12.4 1 2 Br2 159.808 58.85Carbonmonoxide

-7.4 1 1 CO 28.010 -192

Chlorine -15.6 1 2 Cl2 70.906 -34.4Ethyleneoxide

-6.8 1 1 C2H4O 44.05 10.7

Hydrogenchloride

-37.3 3.69 1 HCl 36.46 -85.1

Hydrogencyanide

-9.8 1 2.4 HCN 27.03 26.0

Hydrogenfluoride

-8.4 1 1.5 HF 20.01 19.54

Hydrogensulfide

-11.5 1 1.9 H2S 34.082 -60.28

Methylbromide

-7.3 1 1.1 CH3Br 94.94 3.56

Methylisocyanate

-1.2 1 0.7 C2H3NO 57.1 39.1

Nitrogendioxide

-18.6 1 3.7 NO2 46.01 21.1

Parathion-6.6 1 2

C10H14NO5PS291.3 375

Phosgene -10.6 2 1 CCl2O 98.92 8

Phosphamidon-2.8 1 0.7

C10H19ClNO5P299.70 162

Phosphine-6.8 1 2 PH3 34.00 -87.8

Sulphurdioxide

-19.2 1 2.4 SO2 64.07 -10

Tetraethyllead

-9.8 1 2 C8H20Pb 323.44 84

Table 4.3: Predefined toxic substances

The user can specify one of these substances as the toxic component in scenario section GAS_-COMPOSITION_AND_VOLUME:

GAS_COMPOSITION_AND_VOLUMEPOSITION_OF_FUEL_REGION 0 0 0DIMENSION_OF_FUEL_REGION 0 0 0TOXIC_SPECIFICATION "Ammonia"VOLUME_FRACTIONS



TOXIC 1EXIT VOLUME_FRACTIONSEQUIVALENCE_RATIOS_(ER0_ER9) 1E+30 0

EXIT GAS_COMPOSITION_AND_VOLUME

For a pure toxic release, specify ER0 = 1E+30, although the toxic substance might not be com-bustible. This is not formally correct, but is a workaround and gives a mass fraction of one.

The alternative is to write a user defined toxic data file (./my_toxic_data.dat):

: substance , a , b , n , formula , MolarMass(g/mol) , BoilingPoint(°C)"Acrolein" -4.1 1. 1. "C3H4O" 56.06 53.="Acrylaldehyde"

Attention:

The first sign in the toxic data file must not be space or tab. Comments starts with #, ! or ’ ’(space).

The scenario specifiaction should then look like this (in this case there is also a fraction of methanin the gas mixture):

GAS_COMPOSITION_AND_VOLUMEPOSITION_OF_FUEL_REGION 0 0 0DIMENSION_OF_FUEL_REGION 0 0 0TOXIC_SPECIFICATION "Acrolein, data_file=./my_toxic_data.dat"VOLUME_FRACTIONS

METHANE 1TOXIC 9

EXIT VOLUME_FRACTIONSEQUIVALENCE_RATIOS_(ER0_ER9) 1E+30 0

EXIT GAS_COMPOSITION_AND_VOLUME

The substance properties can be specified directly in the GAS_COMPOSITION_AND_VOLUME sec-tion:

TOXIC_SPECIFICATION "probit_constants=-4.1,1,1,molar_mass=56.06"

Descrition of the variables used to monitor toxic substances can be found in the section Outputvariables in FLACS.

The keywords for TOXIC_SPECIFICATION are:

substance =name of toxic substance (or formula)

probit_constants =a,b,n

molar_mass =M

data_file =name of the datafile (default toxic_data.dat)

The TOXIC_SPECIFICATION string is stripped of spaces and converted to lowercase beforeparsing. The name of the datafile and formula are case sensitive.

Warning:

The implemented models for toxic components are limited to substances with purly gaseousbehaviour. Toxic substances with a boiling point above ambient temperature will typicallyspread as a mist and their toxic effect could for instance require direct skin contact. Sucheffects are not currently handled by FLACS.


128 Flacs simulator

Attention:

A gas cloud with toxic components should not be specified in the GAS_COMPOSITION_-AND_VOLUME scenario section. To specify a given mass fraction in a predefined gas clouduse the cloud interface.

4.4.19.1 Toxic dose TDOSE

Toxic dose TDOSE is defined as:

TDOSE(mg/m3 ·minute) =∫ t

0Cndt

For constant C:TDOSE(mg/m3 ·minute) = Cn · t

4.4.19.2 Toxic probit function PROBIT

Toxic probit function PROBIT is defined as:

PROBIT(−) = a + b · ln(TDOSE)

4.4.19.3 Probabilty of death as function of toxic probit PDEATH

Probabilty of death as function of toxic probit PDEATH is defined as:

PDEATH(−) = 0.5 ·[

1 + er f(

PROBIT − 5√2

)]where:

er f (x) =2√π

∫ x

0e−t2

dt

4.4.20 Modifying names and units for output variables

It is possible to use alternative names for some output variables in FLACS. For example theold DRAG name may be substituted by PDYN (dynamic pressure), and the units for the pressurevariables may be set in the VARIABLE_DEFINITION section (Pa, hPa, kPa, barg or mbarg):

New Old Units DescriptionP P (Pa) Static pressurePIMP PIMP (Pa∗s) Static pressure

impulseUPDYN UDRAG (Pa) Dynamic pressure

x-componentVPDYN VDRAG (Pa) Dynamic pressure

y-componentWPDYN WDRAG (Pa) Dynamic pressure

z-component


4.5 Files in FLACS 129

PDYN DRAG (Pa) Dynamic pressureUDIMP UDIMP (Pa∗s) Dynamic pressure

impulsex-component

VDIMP VDIMP (Pa∗s) Dynamic pressureimpulsey-component

WDIMP WDIMP (Pa∗s) Dynamic pressureimpulsez-component

DIMP DIMP (Pa∗s) Dynamic pressureimpulse

PTOT N/A∗ (Pa) Total pressurePTOT=P+PDYN

PRIMP N/A∗ (Pa∗s) Total pressureimpulse

Table 4.4: Modified variables names.

Remarks:

N/A = not available

4.5 Files in FLACS

Input and output data for FLACS are stored in files. The name of each file consists of two partsseparated by a dot (.). The first part of the file name contains a two-letter prefix followed by the6-digit job number. The second part of the file name, called the file type, contains a prefix of oneor more letters followed by one or more digits (dat3 for most of the files, n001 etc. for the leakdata files).

Summary of the files used by FLACS:

File name Contents of filecs000000.dat3 Scenariocg000000.dat3 Gridcp000000.dat3 Porositiesco000000.dat3 Obstructionscc000000.dat3 Runtime simulation controlcn000000.dat3 Time dependent CFL-numberscl000000.n000 Time dependent leak datar1000000.dat3 Scalar-time outputr3000000.dat3 Field outputrt000000.dat3 Simulation logtt000000 Simulation log, terminal printoutrd000000.n000 Simulation dumprx000000.n000 Simulation save, temporary file created by

FLACSTable 4.5: Summary of files in FLACS.

Summary of the identification numbers used in the files:


130 Flacs simulator

File name Meaning of file type digits??000000.dat3 Number of dimensions (1, 2, or 3)cl000000.n000 Number identifying a leakrd000000.n000 Number identifying a Dump or Loadrx000000.n000 Number identifying a variable

Table 4.6: Identification numbers

4.6 Input files to FLACS simulations

This section summarizes the various input files to FLACS.

4.6.1 Obstruction file

Defines the geometry, contains a list of primitives (boxes and cylinders) from a CASD database,generated by CASD, required by Porcalc and Flowvis.

File name template: co000000.dat3

For briefness this file may be called the co-file or obstruction file hereafter. It is a binary fileand will be generated when starting the Porcalc programme from CASD. It contains a list ofgeometrical primitives (boxes and cylinders) which are extracted from the CASD database. Theco-file is not accessed by FLACS. The Porcalc programme will require read access for the co-file,and it is also used by Flowvis when geometry is specified for a plot.

4.6.2 Grid file

Defines the computational mesh, generated by CASD, required by Porcalc.

File name template: cg000000.dat3

For briefness this file may be called the cg-file or grid file hereafter. It is a binary file and willbe created when using the grid definition menu in CASD. The grid file stores the computationalgrid, i.e. the discrete representation of the simulation volume. The simplest form of a grid is auniform grid, where all control volumes have the same size and shape.

The figure below shows a two-dimensional section of a uniform grid, and illustrates the positionof the internal nodes and the boundary nodes relative to the grid lines.


4.6 Input files to FLACS simulations 131

Figure 4.2: Section of a grid showing internal and boundary nodes

The numbering scheme for the nodes and grid lines may need some attention. In FLACS andCASD one type of numbering is used, whereas in Flowvis a different type of numbering hasbeen adapted. The explanation to this is that FLACS and CASD have kept the original numberingscheme which was developed initially, and at a later time when Flowvis was developed a newand more intuitive numbering scheme was selected. Below is a table which shows the details ofthe two numbering schemes:

Numbering schemes for nodes and grid lines:

Sizes Internal nodes Boundarynodes

Grid lines Used in

N = NX, NY, NZ 2 to N-1 1 and N 2 to N Flacs and CASDM = N-1 1 to M 0 and M+1 1 to M+1 Flowvis

Table 4.7: Numbering schemes for nodes and grid lines.

The Flowvis number is one less than the FLACS and CASD number. This should be kept in mindwhen setting up plot domains in Flowvis. (see section Flowvis for details).

The total number of nodes and control volumes will be the same for both numbering schemes.Below is a figure showing a three-dimensional grid on a typical offshore module geometry.

In the figure above the grid is extended outside the module walls and also above the moduleroof, this is because it was used for a multi-block simulation.


132 Flacs simulator

In typical semi-confined geometries like an offshore module one may create a grid only coveringthe interior including the outer walls of the module, but it is recommended to extend the gridto improve the boundary conditions. In the cases of very open geometries and in multi-blocksimulations one should always extend the grid well outside the main explosion area.

More information can be found in sections Porcalc and Calculate porosities.

4.6.3 Polygon file

Defines the polygon model used by CASD, used by Flowvis if present.

4.6.4 Header file

Defines the co, cg and cm files to be used by CASD.

4.6.5 Porosity file

Defines the porosities for each grid cell, calculated by the program Porcalc from the co and cg-files, required by Flacs.

File name template: cp000000.dat3

For briefness this file may be called the cp-file or porosity file hereafter. It is a binary file and willbe generated when using the Porcalc programme.

It contains data which are calculated based on the geometry and the grid. FLACS will stop if thecp-file is not accessible for reading. The cp-file is quite large, the size (in bytes) may be calculatedas follows:

SIZE = 10 · 4 · NX · NY · NZ

In order to save space the file may be deleted and regenerated when needed using Porcalc, or thesize may be reduced using the unix command compress, remember to uncompress the file beforeusing Flacs, CASD or Flowvis.

4.6.6 Scenario file

Defines the general scenario (monitor points, output variables, fuel region, vents, ignition posi-tion, etc.), required by Flacs.

File name template: cs000000.dat3

For briefness this file may be called the cs-file or scenario file hereafter. It is a text file and will becreated when using the scenario definition menu in CASD.

The term scenario was defined in the introduction to this User’s Guide, briefly as being the set ofparameters which may be used to control the behaviour of a given FLACS simulation.

The first line of the scenario file identifies the file format, for FLACS v9.0 this is set to the followingtext string: "VERSION 0.5". This must not be changed manually since it will be used by FLACSto determine how to read and interpret the file.

Only the most recent format is described here.

The scenario file contains several sections, which are structured as follows:



SECTION_NAME

A section may contain several lines ...

EXIT SECTION_NAME

Sections in the scenario file:

SINGLE_FIELD_VARIABLESMONITOR_POINTSPRESSURE_RELIEF_PANELSSINGLE_FIELD_SCALAR_TIME_OUTPUTSINGLE_FIELD_3D_OUTPUTSIMULATION_AND_OUTPUT_CONTROLBOUNDARY_CONDITIONSINITIAL_CONDITIONSGAS_COMPOSITION_AND_VOLUMEVOLUME_FRACTIONSIGNITIONLEAKSOUTLET / VESSELWATERSPRAYSLOUVRE_PANELSGRATINGMONITOR_VOLUMES

4.6.7 Setup file

This is an optional file used to set certain user-defined variables, such as constants in the com-bustion model. FLACS Run manager detects the file automatically if the file exists in the samedirectory as the regular job files. When using the run9 runflacs command the setup file mustbe supplied as argument #2.

> run9 runflacs 010101 cs010100.SETUP

The setup file may contain the following so-called namelists: JOBSPEC, SETUP andPARAMETERS. Note that the $ in the name list must be positioned in column 2 (only on cer-tain machine types). The namelists may be entered in any order or may alternatively be left outentirely.

VERSION 1.1$JOBSPEC

...$END$SETUP

...$END$PARAMETERS

...$END

4.6.7.1 The JOBSPEC namelist

The available keywords in JOBSPEC and their default values are summarized below:

VERSION 1.1$JOBSPEC

BLOCKS = " ", " "


134 Flacs simulator

IGNITION = " "SYNC_OUTPUT = .TRUE.KEEP_OUTPUR = .FALSE.RESET_LOAD = .TRUE.

$END

The meaning of the keywords is as follows:

Keyword DescriptionBLOCKS List of job numbers and block type for each

block:

• there should normally be only oneFLACS block, but it is possible to usemore than one

• there may be zero or more BLASTblocks

• a maximum of 10 blocks are allowed

IGNITION Job number for the block where ignition shalloccur:

• this must be a FLACS block

SYNC_OUTPUT Synchronize output (r3-file) so that all blockswrite at the same time:

• plots at same time can thereby beshown properly in Flowvis

KEEP_OUTPUT Keep old results on existing r1-file and r3-file(append new results):

• .TRUE. if you want to run acontinuation run

• .FALSE. if you restart a new scenario

RESET_LOAD Set equal to .TRUE. if you want to resetinitial condition at LOAD time:

• useful for starting an explosion after adispersion

4.6.7.2 The SETUP namelist

The available keywords in SETUP and their default values are summarized below:

VERSION 1.1$SETUP

TIME_STEPPING = "STRICT"EQUATION_SOLVER = "BI_CGSTAB"MASS_CONSERVATION = "BEST"COMBUSTION_MODEL = "BETA3"TURBULENCE_MODEL = "K-EPS"DIFFUSION_MODEL = "LINEAR"



GASDATA_MODEL = "DEFAULT"AIR = "NORMAL"AMBIENT_PRESSURE = "1.0E5"

$END

These controls will assume default values if not specified by the user.

The meaning of the keywords is as follows:

Keyword DescriptionTIME_STEPPING Selection of time stepping method:

• STRICT• STRICT:V_MEAN=<real_number>• NORMAL

EQUATION_SOLVER Selection of linear equation solver:

• BI_CGSTAB• TDMA

MASS_CONSERVATION Selection of mass conservation ’quality’:

• GOOD• BETTER• BEST

COMBUSTION_MODEL Selection of combustion model:

• BETA3• SIF (deprecated)• BETA2 (deprecated)• BETA1 (deprecated)

TURBULENCE_MODEL Selection of turbulence model:

• K-EPS

DIFFUSION_MODEL Selection of turbulence model:

• HARMONIC• LINEAR

GASDATA_MODEL Selection of gasdata model, laminar burningvelocity as function of ER for the fuels:

• DEFAULT• name of directory containing gasdata

files• BUILT-IN


136 Flacs simulator

AIR Specification of O2 fraction in air (<percent>is the fraction multiplied by 100):

• NORMAL• <percent>VOLUME• <percent>MOLE• <percent>MASS

AMBIENT_PRESSURE Specification of ambient pressure, the defaultvalue is:

• 1.0E5

Time stepping for ventilation simulations It is possible to choose a time-stepping algorithmwhich only includes the convective speed, by specifying TIME_STEPPING as ’STRICT:V_-MIN=<real number>’. The CFL-number based on convective speed (CFLV) is given as usual(in the cs-file). CFL-number based on speed of sound (CFLC) is not used (the value of CFLC givenin the cs-file is not employed). Acoustical waves are not sought resolved using this approach.

This criterion is intended as an alternative to the default criterion when the flow is station-ary/slowly varying or nearly incompressible. It has been tested in wind/ventilation simulations.It should not be used for an explosion simulation. A speed-up factor of ca. 8 is seen from testsimulations of wind/ventilation using this criterion compared to the default setup in FLACS.However, the speed-up factor depends on the scenario.

When high-momentum leaks are modelled, the convective speed is relatively large, and thespeed-up effect when using this criterion may be limited in this case. To employ this criterion,the user must specify a velocity V_MIN [m/s], for example as in ’STRICT:V_MIN=1.0’. Thisvelocity V_MIN is used by the time-stepping algorithm as a minimum speed when determiningthe time-step, its value should be positive and not too small.

If a wind field is specified, a natural choice would be to set V_MIN equal the value of WIND_-SPEEED, if no wind field is specified a value of 1.0 m/s for V_MIN would be a typical value.Note that the value of V_MIN is in general only used in the initial phase of the simulation (whennormally a flow field is started from a condition at rest to ensure that the time-step is not too largeeven though the velocity of the flow is zero or very small (when the default time-step criterion isused the value of CFLC limits the time-step even when the convective speed is zero).

Attention:

Note that the intention by using this time-step criterion is to speed up the calculation byusing a coarser resolution in time (longer time-steps), and this may change the simulationresults compared to a finer resolution in time (smaller time-steps) when the flow field istransient. A typical choice of CFLV would be 1.0, a larger value of CFLV may lead to unstableresults (depending on the scenario).

4.6.7.3 The PARAMETERS namelist

The available keywords in PARAMETERS and their default values are shown below:

VERSION 1.1$PARAMETERS

ZERO_APOR = 0.0ZERO_VPOR = 0.0



ER_LOW = 0.0ER_HIGH = 0.0FLUX_CONTROL = 1MAX_ITERATIONS = 100ERROR_LIMIT = 1.0E-6ITERATE = 1TIMEORDER = 0

$END

Note that the values in the PARAMETERS namelist are numerical, not text strings as in the SETUPnamelist.

The meaning of the keywords in the PARAMETERS namelist is as follows:

Keyword DescriptionZERO_APOR Lower limit for area porosities, may be used

to avoid problems associated with small areaporosities:

• 0.0 not in effect (default)• 0.05-0.1 to avoid MASS_RESIDUAL

ZERO_VPOR Lower limit for volume porosities, may beused to avoid problems associated withsmall volume porosities:

• 0.0 not in effect (default)• 0.05-0.1 to avoid MASS_RESIDUAL

ER_LOW Lower bound for ER range inGAS_MONITOR_REGION:

• 0.0 not in effect (using ER_LFL asdefault)

ER_HIGH Upper bound for ER range inGAS_MONITOR_REGION:

• 0.0 not in effect (using ER_UFL asdefault)

FLUX_CONTROL Controlling oscillating velocities on stretchedgrids:

• 1 = normal (default)• 2 = reduced to avoid oscillations

MAX_ITERATIONS Maximum number of iterations in the linearsolver, this may be used to speed up thecomputation:

• 100 = high effort (default)• 50 = medium effort• 5 = low effort


138 Flacs simulator

ERROR_LIMIT Error residual limit in the linear solver, thismay be used to speed up the computation:

• 1E-6 = high accuracy (default)• 1E-4 = medium accuracy• 1E-2 = low accuracy

ITERATE Residual limit in the linear solver, this maybe used to speed up the computation:

• 1 = normal (default)• n = iterate n times (n>1)

TIMEORDER Discretization order of time differentials:

• 0 = 1. order (default), enforcesITERATE = 1

• 1 = 1. order, may use ITERATE > 1• 2 = 2. order, may use ITERATE > 1

Changing the values in the PARAMETERS namelist will affect the accuracy and stability of thecode. In cases where the simulation gives MASS_RESIDUAL problems it may be beneficial to setthe values of ZERO_APOR=0.1 and ZERO_VPOR=0.1 (or similar values in the order of 0.01 - 0.1).

In cases with stretched grids one may see oscillating flow in where the ratio between smallestand largest side length of the control volume is large, try to set FLUX_CONTROL=2 to avoid theproblem.

A speed-up of 10-20% may be achieved by changing the accuracy and effort level from high tolow (MAX_ITERATIONS=5 and ERROR_LIMIT=1E-2).

Increasing the ITERATE value will increase the calculation time drastically. The memory usagewill also increase when TIMEORDER is increased. In cases where a converged solution is notachieved otherwise one may try to set TIMEORDER=1 and ITERATE=3. Note that this option isstill in the phase of testing and should be used with caution (it does not seem to help very muchat present state).

4.6.8 Example: using a setup file for vessel burst calculations

A region with given pressure and/or temperature has been available in FLACS for some time,now this possibility has been further enhanced:

1. The region can also contain flammables (used for special shape clouds)2. The calculation can be carried out using BLAST block only

The following setup-file will define a 12m diameter spherical high-pressure region at 300 bargand 2500ºC to be calculated using the BLAST solver in FLACS. This is 4-5 times faster thanthe FLACS solver, requires much less memory (i.e. larger jobs can be simulated), but does nothave panel and porosity functionality. If such functionality is required, one should instead use aFLACS block (remove jobspec-section below).




PFAC = 1HPPOS = 44, 44, 0HPSIZ = 12, 12, 12HPEXP = 2.0, 2.0, 2.0HPTYP = 1, 1, 1

$END

$SETUPKEYS="PS1=01,P_SET=Y:301,T_SET=Y:2500"

$END

$JOBSPECBLOCKS = "900000" "BLAST"

$END

Use short time step (of the order CFLC=0.05) and start FLACS with

> run9 flacs setup-file (the job number is given in the setup-file)

If the user wants to simulate the rupture of a pressurized vessel filled with evaporating liquid(BLEVE), one should consider the following approach (not validated):

1. Use a vessel of larger dimensions with correct pressure and boiling point temperature.2. Transfer all liquid into pressurized gas.

To create a cylindrical region instead of a spherical, HPTYP can be changed from 1,1,1 to 1,1,0(vertical cylinder) or 0,1,1, (x-direction cylinder) etc.

If only a z>0 hemisphere is wanted, one can change the KEY string:

KEYS="PS1=01,P_SET=Y:301,T_SET=Y:2500,HPCON===+1"

Explanation: HPCON=XYZF

X/Y/Z can have the following values (one for each direction XYZ):

’-’ (negative half),’+’ (positive half) or’=’ (both halves)

F can have the following values (FUEL lean or FUEL rich):

’0’ (lean, i.e. ER9) or’1’ (rich, i.e. ER0) concentration of fuel

This is thus an alternative to the cloud interface described below. A hemispherical cloud atambient T and P with diameter 20m in origin can be defined like this:


PFAC = 1HPPOS = -10, -10, 0HPSIZ = 20, 20, 10HPEXP = 2.0, 2.0, 2.0HPTYP = 1, 1, 1

$END

$SETUPKEYS="HPCON===+1"

$END

(HPTYP = 1,1,0 gives a vertical cylinder etc.)


140 Flacs simulator

4.6.9 Cloud file

Optional file used to define fuel clouds of arbitrary shape.

File name template: cs000000.CLOUD

The cloud interface module in FLACS can be used to specify rectangular or other than rect-angular shapes of clouds with uniform or non-uniform concentration of fuel. The cloudinterface module is automatically invoked if FLACS finds the file cs000000.CLOUD in theworking directory, where 000000 is a given 6-digit job number. The hull software estab-lishes nearest neighbor connectivity for the scattered data points in the CLOUD file (seehttp://cm.bell-labs.com/who/clarkson).

A simple example of a CLOUD file defining a 1.5 m x 1.5 m x 0.34 m rectangular cloud of homo-geneous concentration, is shown below:

VERSION 1.0

POINT:

9.2500 -0.8500 0.0240 1.000010.7500 -0.8500 0.0240 1.000010.7500 0.6500 0.0240 1.00009.2500 0.6500 0.0240 1.00009.2500 -0.8500 0.3640 1.000010.7500 -0.8500 0.3640 1.000010.7500 0.6500 0.3640 1.00009.2500 0.6500 0.3640 1.0000

In this file, the first line:

VERSION 1.0

is a version identification for future compatibility checking and the value of 1.0 should be enteredto be consistent with future interpretation.

The third line defines the type of data object used:

POINT

The following data objects can be used with the cloud interface:

• POINT: the next lines contain points :N∗(x,y,z,f)• TETRAHEDRON: the next lines contain points :4∗(x,y,z,f)• HEXAHEDRON: the next lines contain points :8∗(x,y,z,f)• ARRAY: the next lines contain points :NI∗NJ∗NK∗(x,y,z,f)• FLUENT-UNS: the next lines contain FLUENT-UNS profile data• (( : the start of FLUENT-UNS profile data

The sequences of coordinates follow the type of data object. The coordinate of the first point is:

9.2500 -0.8500 0.0240 1.0000

where the definition of each of the four numbers is:

• x = x-coordinate


http://cm.bell-labs.com/who/clarkson


• y = y-coordinate• z = z-coordinate• f = fuel mass fraction

Several point sets or more generally several data object sets, can be specified by initiating thesequence of coordinates by a

:

as on the fourth line of the example.

Additional options are available with the cloud interface, such as filters:

• ER_FLAT a b

This option filters the f-field with a cutoff value (a) and an insert value (b). If the option ’ER_-FLAT’ is not specified the program does:

if (ERf<ER9) ERf=ER9; else if (ERf>ER0) ERf=ER0

and if the option ’ER_FLAT’ is given the program does

if (ERf<a) then ERf=ER9; else ERf=b, ER9<b<ER0

The cloud interface module can recognize ASCII data files of certain format, in particular theFLUENT-UNS profile file format. Geometrical transformations to align the FLUENT coordinatesystem with the one used in FLACS are available. These are:

• INIT: intialise the transformation matrix• TRANSLATE tx,ty,tz• SCALE tx,ty,tz,scale• ROTATE tx,ty,tz,ax,ay,az,revolution,angle

4.6.10 Layer file

Optional file used to define dust layers and dust lifting parameters.

File name template: cs000000.LAYER

4.6.11 Events file

Optional file used to specify triggers and events.

File name template: cs000000.EVENTS

4.6.12 Heat file

Optional file used to specify heat transfer by convection.

File name template: cs000000.HEAT


142 Flacs simulator

4.6.13 Radiation file

Optional file used to specify heat loss by radiation.

File name template: cs000000.RAD

4.6.14 Pool setup file

This is an optional file used to specify a liquid spill on the ground. The pool model is enabled bysetting : Run Manager → Parameters → FLACS version = pool.

File name template: cs000000.POOL

It is possible to define both a static pool and a spill that moves on the ground. For a movingspill, the shallow-water equations are solved in two dimensions on the ground. Heat from theground, the flow above the pool/spill, and the radiation from the sun determine the evaporationrate. The composition of the liquid equals that for the gas composition, see Gas composition andvolume. Remember to set

EQUIVALENCE_RATIOS_(ER0_ER9) 1E+30, 0.0

Input variables are given as follows:

VERSION 0.1

$POOL_SETUPPOOL_MODEL = 3POOL_GROUND = "CONCRETE"START_POOL = 5.0MASS_POOL_0 = 0.0DMDT = 400.0RAD_I = 0.0RAD_O = 5.0XT = 0.0YT = 0.0ZT = 5.0T_SOIL = 283.0HEAT_SUN = 100.0ROUGH_L = 0.001

$END

• POOL_MODEL tells which pool model that should be used. POOL_MODEL=1 (PM1) is thestatic circular pool with macroscopic correlations for the heat and mass transfer. POOL_-MODEL=3 (PM3) refers to the spill model with a moving spill, where heat and mass trans-fer is calculated locally in each control volume.

• POOL_GROUND describes the ground. Possible grounds are "CONCRETE", "AVERAGE","WATER", "SOIL", "PLATE", "INSULATED", and "USER". Only "CONCRETE" and "AV-ERAGE" (soil) work for PM1. If "USER" is specified, values for ground conductivity"CONDUC_S" and ground diffusivity "DIFFUS_S" must be added to the setup file. It isalso possible to specify more than one ground property by doing as follows:

1. Give the general ground at the beginning of the text string: "GROUND1"2. Specify the other ground and a box where this ground properties should be

used: "GROUND1,GROUND2 [x_low,x_high,y_low,y_high,z_low,z_high]". Example:"SOIL,WATER[:: -1,0.1]". ":" means the entire range and can only be used for x and ydirections.



Conductivities and diffusivities for predefined ground are given in Ground properties ta-ble.

• START_POOL is the point in time when pool leakage starts and the pool starts to evaporate[s].

• MASS_POOL_0 is the pool mass at time START_POOL [kg].• DMDT is the mass rate inserted uniformly over the initial pool area [kgs−1].• RAD_I defines the inner radius of the pool donut. (PM1) or leak donut shaped area. Use

RAD_I=0.0 is default and gives a circular pool (PM1) or leakage area (PM3) [m].• RAD_O is the outer radius of the circular pool (PM1) or leak area (PM3) [m].• XT is the x coordinate of the centre of the pool (PM1) or leakage area (PM3) [m].• YT is the y coordinate of the centre of the pool (PM1) or leakage area (PM3) [m].• ZT defines the altitude from where Flacs searches for the ground in the negative z-direction.

If ZT is inside an object, there are two possible outcomes:

1. If there is no gap below ZT, then ZT = ZMAX.2. If there is a gap (free space) below ZT, then ZT is the altitude of the surface below the

gap.

[m].• T_SOIL is the ground temperature [K].• HEAT_SUN is radiative heat from the sun [Wm−2].• ROUGH_L is the surface roughness. Default value is 0.0005 m.• PLATE_L is thickness of the steel plate and it must be given an value when "PLATE" is in

POOL_GROUND [m].• WATER_VEL = vel_x,vel_y is a two-dimensional vector describing the velocity of wa-

ter. Default value is 0.0,0.0 [ms−1].• T_POOL should be given in the pool or leak temperature differs from the boiling point

temperature [K].• CONDUC_S is ground conductivity when "USER" is defined in POOL_GROUND

[Wm−1K−1].• DIFFUS_S is ground thermal diffusivity when "USER" is defined in POOL_GROUND

[m2s−1].

Ground Conductivity (Wm−1K−1) Thermal diffusivity (m2s−1)Concrete 1.1 10−6

Average/Soil 0.9 4.3 · 10−7

Plate 15 3.9 · 10−6

Insulated 0.0 1030

User Set CONDUC_S Set DIFFUS_SWater Heat transfer coefficient Heat transfer coefficient

Table 4.11: Ground properties

See also:

Example file included with the FLACS installation. Linux:

> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/pool/cs000006.POOL .

Windows:Copy files from C:\Program Files\GexCon\FLACS_-v9.0\doc\examples\pool\cs000006.POOL


144 Flacs simulator

4.6.15 Pool leakage file

This is an optional file and it requires that cs000000.POOL exist and that Run Manager→ Param-eters → FLACS version = pool.

File name template: cl000000.POOL

The cl-file contains:

• Line 1: ’POOL LEAK FILE’

• Line 2: Number of variables• Line 3: Name of columns. ’TIME’ should always be in the first column.• Line 4 - : Values at different points in time.

Example:

’POOL LEAK FILE’2’TIME [s]’ ’DMDT [kg/s]’0.0 0.01.0 1000.0100.0 1000.0101.0 0.01200 0.0

Attention:

Time in the cl000000.POOL is relative to START_POOL in cs000000.POOL.

See also:

Example file included with the FLACS installation. Linux:

> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/pool/cl000006.POOL .

Windows:Copy files from C:\Program Files\GexCon\FLACS_-v9.0\doc\examples\pool\cl000006.POOL

4.6.16 Leak file

Defines time dependent leaks of either fuel or suppressant – the number i = 1, 2, 3, ... identifiesthe leak; generated by the Jet program.

File name template: cl000000.n001

For briefness this file may be called the cl-file or leak file hereafter. It is a text file and may begenerated using a text editor, it may also be generated using CASD which is the preferred wayto do it.

It is possible to simulate realistic leaks in FLACS. This is done by proper subgrid modelling ofthe conditions at the leak outlet and by allowing time varying leak data and multiple leaks. Theposition and direction vector for each leak must be specified on the scenario file. For each leakthere must exist one file cl-file containing the time dependent leak data. The leak number is usedas identification number in the file type.

The format of the cl-file is as follows:



LEAK_CONTROL_STRING7TIME AREA RATE VEL RTI TLS Tdata (1-7)......

The seven parameters that constitute the time dependent leak data are chosen so that it is possibleto specify rate, velocity, turbulence, and temperature for the leak. (at the nozzle throat) The namesin the file format described above have the following meaning:

Parameter Units DescriptionTIME (s) TimeAREA (m2) AreaRATE (kg/s) Mass flowVEL (m/s) VelocityRTI (-) Relative turbulence intensityTLS (m) Turbulence length scaleT (K) Temperature

Table 4.12: Leak parameters

Details of the calculation of turbulent kinetic energy and its rate of dissipation from the set ofparameters listed above (VEL, RTI and TLS) are given for the WIND condition.

Line number one of the leak file contains the LEAK_CONTROL_STRING which is used to specifythe type of leak and the leak outlet. The following letters are recognized:

Letter DescriptionJ Jet leakD Diffuse leak+ Direction is now positive- Direction is now negativeX Leak through control volume face

perpendicular to the x-axisY Leak through control volume face

perpendicular to the y-axisZ Leak through control volume face

perpendicular to the z-axisTable 4.13: Leak control string letters

Diffuse leaks are handled by setting fixed values for the scalar quantities and introducing a masssource equal to the specified leak rate at the leak node. The area porosity at the leak outlet is setso that the specified area is obtained. A leak velocity that satisfies mass conservation is obtainedautomatically through the flow field solution procedure.

Jet leaks are handled by setting fixed values for the scalar quantities and one or more of themomentum components at the leak node. The area porosity is calculated so that the specifiedmass flux and momentum flux is obtained. The mass conservation is not solved for at a jet leaknode during the flow field calculation.

The default leak type is diffuse (D) and the default direction is positive (+). Take care not tospecify a control volume face more than once. The following examples deal with the LEAK_-CONTROL_STRING.


146 Flacs simulator

4.6.16.1 Example 1: Diffuse leak in positive x-direction

’X’

4.6.16.2 Example 2: Diffuse leak in all directions:

’D+XYZ-XYZ’

4.6.16.3 Example 3: Erroneous specification:

’DJ+X+XY’

The following error messages would be issued on the log file(s) for example 3:

*** (LKINIT) ALREADY SET DIFFUSE

*** (LKINIT) ALREADY SET +X

Note that the simulation does not stop unless severe errors occur. In example 3 the leak specifi-cation would be taken as:

’D+XY’

Note that the LEAK_CONTROL_STRING must be enclosed in apostrophes (’...’) and may containblanks. The maximum length is 16 characters.

Line number two contains the number of data values to be found in each of the next lines.

Line number three contains the names of the variables. The sequence of the variables is prede-fined and may not be changed. The time will always be the first value on a line containing datavalues.

Line number four and onward contain values for each variable. As the simulation proceedsthe file will be scanned for new data values when old values expire. Intermediate values arecalculated by linear interpolation or time integral (as appropriate) between old time and newtime.

A leak is started when the first time on its leak file is reached and is stopped when the last timeon the file is reached.

In the case of a jet leak the nozzle density, DENS, and pressure, PRES, are calculated as follows:

DENS = RATE/(VEL ∗ AREA)

PRES = DENS ∗ R ∗ T

Where R is the specific ideal gas constant.

If the nozzle pressure is higher than the ambient pressure an analytic jet expansion will be calcu-lated.

Note that the gas composition for each leak is taken to be the one specified by the equivalenceratio for rich gas, ER0, and volume fractions given on the scenario file.

The leak position and direction vector is given on the scenario file. Note that the direction vectordoes not need to be of unit length. FLACS will calculate the direction cosines by dividing each



component in the direction vector by the length of the vector. The direction cosines are used tomodify the flow velocity and the area for the X-, Y- and Z- direction independently. This may beused to simulate jet leaks that are not parallel to the main axes. It is noted that this option shouldbe used with care in cases where the grid resolution is coarse due to the fact that numericaldiffusion tends to smear out the jet.

4.6.16.4 Example 4: Jet at 45.0 degrees:

’J+XY’

Direction vector given as 1.0 1.0 0.0

4.6.16.5 Example 5: Jet at 0.0 degrees:

’J+X’

Direction vector given as 1.0 0.0 0.0

4.6.17 Runtime simulation control file

Optional file, can be used to define output, generate or load dump files, terminate the simulation,etc. (further details can be found in the FLACS manual).

File name template: cc000000.dat3

For briefness this file may be called the cc-file or control file hereafter. It is a text file and may begenerated using a text editor. It is a facility for giving instructions to FLACS during a simulation.Keywords and values may be given on the cc-file which will be read and interpreted each timestep. It may contain any number of lines. Each line shall contain a keyword followed by a valueseparated by a comma (,) or a space ( ). Float values may be entered with or without a decimalpoint, integer values are taken to be the integer part of a float value. The following keywords arecurrently recognized:

Keyword MeaningNLOAD Set load identification and load data at

simulation startNDUMP Set dump identificationTDUMP Dump data at given timeIDUMP Dump data at given iteration numberFDUMP Dump data at given fuel levelTOUTF Write results at given timeIOUTF Write results at given iteration numberFOUTF Write results at given fuel levelTSTOP Stop simulation at given timeISTOP Stop simulation at given iteration numberFSTOP Stop simulation at given fuel level

Table 4.14: Valid cc file keywords

Fuel level is defined as the current total mass of fuel divided by the initial total mass of fuel. Forgas dispersion simulations, where the initial mass of fuel is zero, the fuel level cannot be used.


148 Flacs simulator

The control file is useful when you want to produce output that is not possible to specify on thescenario file and when you want to monitor a simulation while it is running. The dump/loadoption is one example of output and input that is not possible to specify on the scenario file. Thefollowing examples shows how to use the cc file.

4.6.17.1 Example 1

Contents of cc-file (including comments):

"NDUMP" 1 dump file name is "rd000000.n001""TDUMP" 0.20 dump when time is 0.2 s"NDUMP" 2 dump file name is "rd000000.n002""TDUMP" 0.50 dump when time is 0.5 s"TDUMP" 0.70 dump when time is 0.7 s

What happened:

1. The file "rd000000.n001" was written at time 0.20 s.2. The file "rd000000.n002" was written at time 0.50 s.3. The file "rd000000.n002" was written at time 0.70 s.

4.6.17.2 Example 2

Contents of cc-file (including comments):

"NLOAD" 2 load file name is "rd000000.n002"

What happened:

The file "rd000000.n002" was read at simulation startup restoring the exact conditions at the timewhen the file was written. Simulation output was appended to the existing result files. An errormessage was given if the file did not exist.

4.6.18 Fuel file

Defines that the combustible dust.

4.6.19 Time dependent CFL file

Optional file used to specify time dependent CFL numbers (Please see sections CFLV and CFLVfor details).

File name template: cn000000.dat3

For briefness this file may be called the cn-filecn-file or CFL-numbers fileCFL-numbers file here-after. It is a text file and may be generated using a text editor. It is a facility for giving instructionsto FLACS during a simulation.

High flow velocities and change rates and small control volumes put severe constraints on thesize of the time step in computational fluid dynamics. Using large time steps may cause thesolution to become unstable. However, to be able to simulate over a long period with todayscomputer resources, it is necessary to maintain as large a time step as possible. During a gas


4.7 Output files from FLACS simulations 149

dispersion simulation the stability requirements vary, characterized by stronger requirements inperiods of highly unsteady flows and weaker requirements in steady flow periods.

To be able to effectively simulate gas dispersion scenarios including high momentum jets a facil-ity for specifying time dependent Courant numbers is available. FLACS will read time varyingCourant numbers from the cn-file if it exists. Courant numbers specified on the scenario file arethen overridden.

An example showing the format of the cn-file is given below:

HEADLINE3TIME CFLC CFLVdata (1-3)......

Line number one is intended to be just for the user’s information and contains no information tothe system other than being the first line.

Line number two contains the number of data values to be found in each of the next lines, in thiscase the number of data values is 3.

Line number three contains the names of the variables. The sequence of the variables is prede-fined and may not be changed. The time will always be the first value on a line containing datavalues.

Line number four and onward contain values for each variable. As the simulation proceedsthe file will be scanned for new data values when old values expire. Intermediate values arecalculated by linear interpolation or time integral (as appropriate) between old time and newtime.

The cn-file must be generated manually using a text editor. It is advised that only experts makeuse of the cn-file to change the default time setting in FLACS.

4.7 Output files from FLACS simulations

This section summarizes the various output files from FLACS simulations.

4.7.1 Simulation log file

Output log containing the information on the rt-file (and output messages from the operatingsystem); it can be useful to monitor the writing of this file during simulations:

Linux:

> tail --f tt010100

Remarks:

This file is not created on Windows.

4.7.2 Scalar-time output file

Binary output file containing the value of parameters defined in the cs-file in the specified moni-tor points, required by Flowvis for scalar-time plots (further details can be found in section Plot


150 Flacs simulator

type).

File name template: r1000000.dat3

For briefness this file may be called the r1-filer1-file or scalar-time output filescalar-time outputfile hereafter. The values of selected variables for scalar-time output at selected locations (monitorpoints) are written to the r1-file at a given iteration interval.

In the case of a simulation restart the r1-file will by default be deleted (i.e. old results are lost)when using the 2.X file format (but it is kept when using the 1.2 file format). If you want to keepthe old results at restart you must set KEEP_OUTPUT=.TRUE. using a job file. During a LOAD anexisting r1-file will be scanned up to the time recorded on the specified dump file before any newresults are written. It is usually wise to make a copy of the original r1-file before a load/rerun isperformed.

The r1-files generally require modest amounts of disk space. The number of bytes in one datarecord can be calculated from the following expression (the 1.2 and 2.X file formats have approx-imately the same record size):

RecordSize4 ∗ (2 + SUM)

Where SUM is the total number of references made to monitor points and panels.

4.7.3 Field output file

Binary output file containing the value of parameters defined in the cs-file in all grid cells, re-quired by Flowvis for 2D and 3D plots (further details can be found in section Plot type).

File name template: r3000000.dat3

For briefness this file may be called the r3-file or field output filefield output file hereafter. Thewhole matrix of each selected variable for field output is written to the r3-file, when triggered byone or more events (or signals). Simulation start and stop will trigger output as will the eventof passing certain fuel levels and time values. Runtime specified events may also trigger output.A message is issued on the log file(s) for each event that triggers output. Several events mayhappen to trigger output simultaneously, this will generate just one instance of output.

In the case of a simulation restart the r3-file will by default be deleted (i.e. old results are lost)when using the 2.X file format (but it is kept when using the 1.2 file format). If you want to keepthe old results at restart you must set KEEP_OUTPUT=.TRUE. using a job file. During a LOAD anexisting r3-file will be scanned up to the time recorded on the specified dump file before any newresults are written. It is usually wise to make a copy of the original r3-file before a load/rerun isperformed.

The r3-files generally require large amounts of disk space. The number of bytes in one data recordcan be calculated from the following expression (the 1.2 and 2.X file formats have approximatelythe same record size):

RecordSize4 ∗ (2 + NX ∗ NY ∗ NZ) ∗ NVAR

NVAR number of variables specified

NX,NY,NZ matrix dimensions

4.7.4 Simulation log files

Binary output log of general messages


4.7 Output files from FLACS simulations 151

File name template: rt000000.dat3

For briefness this file may be called the rt-file or log file hereafter. It contains a log of the messagesissued during a simulation. It is wise to check if any error messages have been issued. Errormessages are identified by "∗∗∗" at the beginning of the message, plain messages are introducedby a space ( ) or a number sign (#).

In the case of a simulation restart the rt-file will be deleted. It is usually wise to make a copy ofthe original rt-file before a load/rerun is performed.

The information on the log file is also written to a connected terminal (standard output) duringthe simulation.

File name template: rt000000.∗ (new specific log files, see below).

Several log files will be generated when using FLACS, they are all text files and may be printedas time series using common graphical tools (such as gnuplot on Linux or Excel on Windows).The contents of each file is explained at the start of the file, below you will only find a briefexplanation of the new log files:

File name Contents of filert000000.CFL CFL-numbers and time step log filert000000.EQSOL Equation solver (residual) log filert000000.FUEL Amount of fuel log filert000000.MASS Total mass log filert000000.P Pressure log filert000000.dat3 The usual simulation log file

Table 4.15: Flacs log files

Note that some of the files listed above may not appear for a given installation, this is becausethey are intended for debugging and testing purposes only. The most relevant file for a regularFLACS user is the "rt000000.FUEL" file, where the amount of fuel in the simulation volume (totalamount and combustible amount) is logged.

4.7.5 General simulation log file

4.7.6 CFL log file

Output log of CFL numbers.

4.7.7 Fuel log file

Output log of the amount of fuel in the simulation volume.

4.7.8 Mass log file

Output log of the mass in the simulation volume.

4.7.9 Pressure log file

Output log of pressure.


152 Flacs simulator

4.7.10 Simulation dump file

Creates a ’snapshot’ of the simulation at a selected time defined in the cc-file – the number i (i =1, 2, 3, ... , 999) identifies the dump (further details can be found in the FLACS manual).

For briefness this file may be called the rd-file or dump file hereafter. It contains the informationnecessary to restart a simulation from a given time required that the initial input data have notbeen changed drastically. There may be more than one rd-file.

Attention:

If you want to append new data to existing result files, do not change the output specificationor the grid or the geometry when using a dump file!

4.7.11 Monitor file

4.7.12 The frozen cloud concept

The frozen cloud principle assumes that for a given dispersion/ventilation scenario the variationin fuel concentration is approximately proportional to the ratio [leak rate/ventilation rate]. Thisconcept may be used in the following way:

4.7.13 Example: dispersion and ventilation study

Prepare a FLACS dispersion/ventilation simulation in the usual way, say for run no. 123456as in the following example. The simulation files for run no. 123456 should include the filecs123456.MON (in addition to the other needed files, for example cg123456.dat3). The content ofthe text file (could be edited with any text editor) cs123456.MON could be something like as inthe following example.

Attention:

Note that you may have comment lines starting with ’#’

VERSION 1.0# name=<suffix>; output file will be named rt123456.MON.<suffix># start=x,y,z; start position of region in space considered, unit [m]# end=x,y,z; end position of region in space considered, unit [m]# see=1.0; in general this parameter is set to a fixed value 1.0# mix=<fraction>; when the output in the file rt123456.MON.<suffix> is calculated,# the extrapolated fuel concentration in every control volume is calculated# as the fuel concentration of standard FLACS multiplied with the factor <fraction>volume(name="123455",start=138,-21,32,end=198,21,48,output="FUEL(see=1.0,mix=0.8)")volume(name="123456",start=138,-21,32,end=198,21,48,output="FUEL(see=1.0,mix=1)")volume(name="123457",start=138,-21,32,end=198,21,48,output="FUEL(see=1.0,mix=1.2)")

The file cs123456.MON will be automatically read by FLACS when the simulation for run no.123456 is started. Three output files (text files) rt123456.MON.123455, rt123456.MON.123456,and rt123456.MON.123457 will be generated. Output from these files can be used during e.g. aprobabilistic analysis. Using the frozen cloud concept, it could for example be assumed that theoutput in rt123456.MON.123455 are approximately equal similar output that could be obtainedby a new full FLACS simulation with scenario equal run no. 123456 except that the leak rate isreduced by multiplying with the factor 0.8 (ventilation and other parameters are kept fixed).


4.8 Potential bugs or problems with Flacs 153

4.7.14 Simulation save file

Temporary file generated by DESC/FLACS in some situations (further details can be found inthe FLACS manual).

For briefness this file may be called the rx-file or save file hereafter. It is a temporary file thatFLACS uses to store intermediate data for output. There may be more than one rx-file. Thesefiles are needed when time integrals of whole matrices are to be calculated. It avoids using largeamounts of core memory but accessing large files is slow compared to memory access. Note thatthe simulation save files are only written if you have specified field output of variables that needtemporary file storage. These files are deleted when the simulation terminates normally. In caseswhen the simulation is not allowed to terminate normally these files will exist after programmestop, they must then be deleted manually. Existing temporary files may confuse the programmeat start-up. Below is presented a list of the variables that require temporary file storage whenspecified for field output:

Name DescriptionNPIMP Pressure impulseUDIMP Drag-impulse component x-directionNVDIMP Drag-impulse component y-directionNWDIMP Drag-impulse component z-directionNDIMP Drag-impulse valueNRESID Mass residual in continuity equation

It is advised that the above listed variables are not specified for field output. The file handlingwill increase the execution time for the programme drastically!

4.8 Potential bugs or problems with Flacs

This chapter contains a list of potential bugs or problems associated with the CFD simulatorFlacs, and some possible workarounds.

4.9 Warning and error messages

Flacs will write any error or warning messages to the Simulation log file. The following sec-tions give a description of the most common warnings and errors, and possible causes andworkarounds.

4.9.1 Mass residual

The message "∗∗∗MASS RESIDUAL =..." indicates that there are problems to achieve a convergedsolution. If the mass residual is not too large (should be smaller than order of 1/10) or it doesnot last for many iterations it may be ignored. The built in limit for the mass residual is ratherstrict to enable an accurate solution over long time, a short lasting violation of the limit does notnecessarily lead to a suspect result.

There are several possible solutions to this problem

• Lower the CFLV and CFLC numbers. This will result in longer simulation time. It is rec-ommended to try to divide both numbers by 2 in case of mass residual


154 Flacs simulator

• Change the values of ZERO_APOR and ZERO_VPOR. (details)• In cases with stretched grids set FLUX_CONTROL=2. (details)

4.9.2 Leak excess area

If a leak is defined with a larger area than the control volume face where the leak is located, thewarning "∗∗∗ LEAK EXCESS AREA ..." will be issued.

The solution is to:

• Increase the grid cell size where the leak is defined• or redefine the leak so that the leak area is decrease


Chapter 5

Flowvis

156 Flowvis

Flowvis is the postprocessor for the CFD-code FLACS, and is a program for visualizing resultsfrom computer aided simulations of gas explosions, gas dispersion and multi phase flow.

Flowvis was originally developed for presenting output from the multi-phase simulation pro-gram GALIFA (GAs LIquid Flow Analyser). This work was financed by Christan Michelsen Re-search (formerly Christian Michelsen Institute). The basic concepts were based on experiencegained from developing and using the VIEW program which was the result visualization tool forFLACS86 and FLACS89.

The development of Flowvis has been financed through several different projects (see Acknowl-edgements), starting with the Gas Safety Programme 1990-1996.

This chapter describes the menus in Flowvis.

5.1 Overview

The user starts Flowvis either by clicking the Flowvis icon in the run manager window,:

Figure 5.1: The Flowvis desktop icon

or by executing the following command (in Linux):

> run9 flowvis

A Flowvis save file is called a presentation file, and has the suffix .pres, eg. "2D-plots.pres".

The command for starting Flowvis is usually

> run9 flowvis options

The only option available is -verify_por.

The purpose of Flowvis is to visualize simulation results from FLACS. The results from a simulationare called job data in Flowvis and in this guide. The job data, which are stored on a set of files,include grid, geometry ( obstructions), bulk data ( porosities), scenario and results.

The user visualizes simulation results by creating a presentation. A presentation contains one ormore pages, each containing one or more plots. Presentations can be saved on files. Available plottypes include curve plots, 2 dimensional contour and vector plots, and 3-dimensional volumeplots.

One page is displayed at a time in the graphical area. 220 range colours are available for producingfilled contours and shaded plots. In addition, a set of single colours are used for plotting curves,text etc. Flowvis supplies functions for colour table manipulation. Different colour tables can beused for each page within a presentation. In addition, the value to colour or phase to colour mappingcan be set individually for each plot.

Flowvis offers a high degree of flexibility with regards to combining different types of data (grid,geometry, porosities, scalar variable, vector variable etc.) in one plot.


5.1 Overview 157

Cut, copy and paste commands are available for both pages and plots. Geometric primitives canbe blanked to make a better view of data inside the geometry. The contents of the plots can berotated and mirrored.

Flowvis can export to several different formats, including image and video formats. Flowvis canshow animations directly on the screen, or write them to files.

Experimental data on ASCII files can be imported into curve plots. This makes it possible to com-bine experimental and simulated data in the same plots.

If one wants to start Flowvis for porosity verification, a start-up option is available which instructsFlowvis to automatically create a plot for this purpose.

5.1.1 The Flowvis window

The main window is displayed all the time while Flowvis is running. The window contains a menubar, a graphical area, a time slider, a page slider and a button row.

Figure 5.2: Flowvis elements

The user interface consists of a main window, and a set of pop-up dialog boxes. The main window isdisplayed all the time while Flowvis is running. It contains a menu bar which lists the availablemenus. One page at a time is displayed in the graphical area. In addition, there are sliders forselecting which timestep and page to display, and several push buttons for eg. redrawing theselected plot and for re-scanning the input files.

Many of the Flowvis commands have a pop-up dialog box associated with them. The dialog boxpops up when the command is selected. The main window does not accept input while a pop-updialog box is open.

Usually, only one dialog box is open at a time. But under certain conditions, for example upon anerror, a message dialog box may pop up in addition to the one already open. The message dialogbox contains a text and a push button labelled Ok. No other Flowvis dialog boxes takes input


158 Flowvis

while this dialog box is open. The user must verify that the information is perceived by clickingthe Ok button. The message dialog box then pops down.

The Ok and Cancel push buttons are present in most Flowvis pop-up dialog boxes. Clicking theOk button applies all changes specified in the dialog box. The graphics are redrawn if applicable,and the dialog box is closed. The Cancel button dismisses all the changes, and the dialog box isclosed. Many dialog boxes contain an Apply button in addition. Clicking this button causes theselected plot or page to be redrawn with the current changes applied. The dialog box remainsopen.

5.1.2 The menu bar

The menu bar lists the available menus.

1. The File menu contains commands for opening, saving, exporting and printing the presen-tation.

2. The Edit menu contains commands which manipulates both pages and plots.3. The Page menu contains commands for manipulating the displayed page.4. The Plot menu contains commands for creating and editing plots.5. The Verify menu contains a command for verifying porosities, and other control volume

attributes.6. The Options menu contains toggle buttons for turning on and off the display of page head-

ers, implicit redraw and choosing between white and black background.7. The Help menu contains on-line help and about Flowvis.

Not all menu items are available all the time while running Flowvis. This depends on the con-tents of the current presentation and the selected page or plot. Unavailable menu item texts aredimmed.

5.1.3 The graphical area

One page at a time is displayed in the graphical area. Either a page or a plot is selected at everystage while running Flowvis, except when the current presentation is empty. If a plot is selected,this plot is indicated by a frame surrounding the plot. If no frame is displayed, the displayedpage is selected.

A plot is selected by clicking MOUSE+LEFT inside the plot. The displayed page is selected bydouble clicking MOUSE+LEFT somewhere in the graphical area.

5.1.4 The Page slider

The Page slider is for selecting which page to display in the graphical area. The scale is onlyvisible if the current presentation contains more than one page. The pages are numbered from 1.

5.1.5 The Time slider

The Time slider is for selecting which time step to display. The scale is only visible if the displayedpage contains plots with data for more than one time step. The time steps are numbered from 0(zero).


5.1 Overview 159

5.1.6 The Play / Pause button

The Play / Pause button starts an animation sequence of the current plot(s). Clicking the buttonwhile the animation is ongoing, will pause the animation.

5.1.7 Add Page button

Shortcut: CTRL+A

This command adds a new page after the displayed page. The new page contains one plot.

5.1.8 The Previous Plot push button

Shortcut: CTRL+LEFT

Clicking the Previous Plot push button causes the selection of the previous plot. If the previouslyselected plot was the first plot in the page, the page is selected. If the page was selected, the lastplot in the page is selected.

5.1.9 The Next Plot push button

Shortcut: CTRL+RIGHT

Clicking the Next Plot push button causes the selection of the next plot. If the previously selectedplot was the last plot in the page, the page is selected. If the page was selected, the first plot inthe page is selected.

5.1.10 The Rescan Data push button

Shortcut: CTRL+R

Clicking the Rescan Data push button causes Flowvis to rescan all data files referenced in thecurrent presentation. This makes it possible to view the output from a simulation while thesimulation is running.

5.1.11 The Redraw Plot push button

Shortcut: F5

Clicking the Redraw Plot push button causes Flowvis to redraw the selected plot.

5.1.12 The Cancel Redraw Button

Shortcut: ESCAPE

A time consuming plot operation can be interrupted by pressing the cancel redraw button.


160 Flowvis

5.1.13 The pop-up menu

Some commonly used choices from the Edit and Plot menus are collected in a pop-up menuavailable anywhere in the main window (press MOUSE+RIGHT).

5.1.14 Keyboard short-cuts

The following shortcut key combinations are available.

CTRL+N Create new presentation

CTRL+SHIFT+N Create new presentation in directory

CTRL+O Open existing presentation

CTRL+S Save presentation

CTRL+SHIFT+S Save presentation with file new name

CTRL+I Import file

CTRL+E Export presentation to image or movie format

CTRL+P Print presentation

CTRL+Q Exit Flowvis

CTRL+X Cut plot

CTRL+C Copy plot

CTRL+V Paste plot

CTRL+A Add a page to the presentation

CTRL+M Modify the page

Page Up, MOUSE+WHEEL Display previous page

Page Down, MOUSE+WHEEL Display next page

LEFT RIGHT, ALT+WHEEL Select previous/next time step

CTRL+1 Create new scalar-time plot

CTRL+2 Create new 2D cut plane plot

CTRL+3 Create new 3D cut plane plot

CTRL+4 Create new volume plot

CTRL+5 Create new scalar line plot

CTRL+6 Create new grid plot

CTRL+7 Create new monitor points plot

CTRL+8 Create new panels plot

CTRL+9 Create new particle traces plot

CTRL+0 Create new scalar time annotation

CTRL+LEFT, CTRL+RIGHT Select previous/next plot

F1 Online help


5.2 Creating a new presentation 161

5.2 Creating a new presentation

This section introduces the structure of a Flowvis presentation. A Flowvis presentation is a col-lection of plots showing the results from one or several simulations.

5.2.1 The structure of a Flowvis presentation

Flowvis can plot results from simulations located in the directory selected by the user. It is notpossible to include results from more than one directory in a single presentation. To work aroundthis limitation FLACS files can either be copied to a single directory, or on Linux, softlinks can becreated using the following command:

> ln -s ../simulations2/cs010100.dat3

Note that the command must be repeated for all neccesary file (cs, co, cp, cg, r3, r1 files).

To start a new presentation either choose File→New or File→New At. This will create an emptypresentation. The New At command opens a directory selection dialog enabling the user to selectthe working directory. The New command will look for simulation results in the current directory,that is the directory where Flowvis was started.

Remarks:

Only variables saved during a Flacs simulation can be plotted in Flowvis. If a variable ismissing, the simulation scenario must be edited using CASD and simulation must be runover again.

• Pages are added by clicking the Add Page button, or CTRL+A• Subdivide a page to have more than one plot by typing CTRL+M• Add a plot by going MOUSE+RIGHT in the plot area, and select a plot type, or type for

instance CTRL+1 to have a scalar time plot

– A dialog where the user can select simulation number and variable to plot is show

• Modify the plot by MOUSE+RIGHT, selecting Plot Specifiaction, Plot Domain and Variable Ap-pearance

• The presentation can be saved on file using the Save or Save As commands

HINT! It is not necessary to create a new presentation from scratch for each new simulation.An existing presentation can easily be adapted to new simulations by using the Substitute Jobcommand.

5.3 File menu

The File menu supplies commands for creating, opening, saving, exporting and printing presen-tations.

The default file type for presentation files is .pres.


162 Flowvis

5.3.1 New

Shortcut: CTRL+N

This command creates a new empty presentation. If there were unsaved changes to the currentpresentation, a dialog box is displayed, asking the user about saving changes.

5.3.2 New at

Shortcut: CTRL+SHIFT+N

This command supplies the possibility to read data from a new directory. A dialog box for spec-ifying the directory pops up. When a directory has been specified, Flowvis proceeds as for theNew command.

The New At dialog box is a File Selection Box. It lets the user traverse through directories, view thesubdirectories in them and select a directory.

A directory is selected by clicking in the list, or by typing the directory name in the Directory field.To open the selected directory, click the Choose button.

5.3.3 Open

Shortcut: CTRL+O

This command opens an existing presentation file. The dialog box allows the user to specifythe path and file name for the presentation file. If there were unsaved changes to the currentpresentation, a dialog box is displayed, asking the user about saving changes.

5.3.4 Save

Shortcut: CTRL+S

This command saves the current presentation on a file. If no file name has been specified, theSave As dialog box is displayed. This command is available only when the current presentationis not empty.

5.3.5 Save as

Shortcut: CTRL+SHIFT+S

This command saves the current presentation on a new file. The Save Asdialog box is displayed,allowing the user to specify the path and file name for the presentation file. If an existing filename is specified, Flowvis displays a dialog box asking if it is OK to overwrite the file. Thiscommand is available only when the current presentation is not empty.

The file selection box lets the user traverse through directories, view the files and subdirectoriesin them and select files. A presentation file is selected by clicking in the lists, or by typing thefile name in the File name field. The selected file is opened by clicking the Open button or doubleclicking the selected file name in the Files list.The Cancel button cancels the operation.


5.3 File menu 163

5.3.6 Import

Shortcut: CTRL+I

This command imports data from an ASCII file into a Scalar Line plot, and thus makes it possibleto include experimental data in this plot type. This command is available only when the selectedplot is a Scalar Line plot.

5.3.7 Export

Shortcut: CTRL+E

This command exports the displayed page or all pages to various image and movie formats. Thiscommand is only available when the current presentation is not empty.

The Export dialog box allows choosing between exporting the Displayed Page or All Pages, andthe Displayed Timestep or All Timesteps. The output formats are as follows:

Format Type DescriptionPNG Portable Network GraphicsGIF CompuServe graphics interchange formatJPEG Joint Photographic Experts Group JFIF

formatBMP Microsoft Windows bitmap image fileHTML Web PresentationMPEG Motion Picture Experts Group file

interchange format (Movie format)MPEG4 Microsoft MPEG-4v2 video (Movie format)DIV3 DivX ;-) (Movie format)WMV7 Microsoft WMV7 (Movie format)OGG Xiph.org Foundation video format (Movie

format)GIFAnimated CompuServe graphics interchange format

(Movie format)Table 5.1: Flowvis export formats

Some formats might not be available on all systems.

The file name is derived from the presentation name. The page number and timestep number areadded to the name, and the file type depends on the selected device, see above. When exportingall pages and/or timesteps, the file name contains formats for inserting the actual page and/ortimestep number in each file name (one file per. page/timestep). Note that the file suffix mustnot be changed as this is what determines which image format to use for export.

During export the desktop must stay open, ie. you can not switch desktops. You can normallycover the Flowvis window without creating problems for the export.

Note that the HTML export facility has a few limitations.

5.3.8 Print

Shortcut: CTRL+P

This command prints the displayed page or all pages, or creates a pdf (Portable Document For-


164 Flowvis

mat) file. This command is only available when the current presentation is not empty.

The Printdialog box allows choosing between printing the Displayed Page or All Pages, and theDisplayed Timestep or All Timesteps.

To create a pdf file choose Print to file, and give the file a name with suffix ".pdf".

5.3.9 Exit

Shortcut: CTRL+Q

This command ends the Flowvis application. If there were unsaved changes to the current pre-sentation, a dialog box is displayed, asking the user about saving changes.

5.4 Edit menu

The Edit menu supplies commands for editing the selected plot or page.

The Cut, Copy and Paste commands applies to the selected plot or page. If a plot is selected, thisplot is indicated by a white frame surrounding the plot. If no frame is displayed, the displayedpage is selected. A plot is selected by clicking MOUSE+LEFT inside the plot, or by clicking theSelect Previous or Select Next Plot push buttons. The displayed page is selected by double clickingMOUSE+LEFT somewhere in the graphical area.

5.4.1 Cut

Shortcut: CTRL+X

This command copies the selected page or plot to the clipboard, and deletes it from the presen-tation. The command is not available if the current presentation is empty, or if an empty plot isselected.

5.4.2 Copy

Shortcut: CTRL+C

This command copies the selected page or plot to the clipboard. The command is not available ifthe current presentation is empty, or if an empty plot is selected.

5.4.3 Paste

Shortcut: CTRL+V

If the current presentation is empty, or if the displayed page is selected, this command is availableonly if the clipboard contains a page. The command then inserts a copy of the clipboard pageafter the selected page. If a plot is selected, the Paste command is available only if the clipboardcontains a plot, and the selected plot is empty. The command then copies the clipboard plot tothe selected plot.


5.5 Page menu 165

5.4.4 Paste plot format

This command is available only if the clipboard contains a plot. The command copies the format,or parts of the format, of the clipboard plot to the selected plot, or all the plots of the same typein the selected page or the presentation.

The Paste Plot Format dialog contains two sets of buttons. The first are three radio buttons forselecting the target for the paste operation. The format of the clipboard plot can be pasted eitherto the selected plot only (if same plot type), to all plots of same the type in the displayed page, or to allplots of same the type in the entire presentation.

In addition, there are toggle buttons for deciding which parts of the plot format to paste. Theoptions are found in the table below.

Name DescriptionRelative Timestep This is the timestep for the plot relative to

the timestep for the pageRotation Matrix This implies the settings in the Rotate dialog

boxVariable Appearance This implies the settings in the Variable

Appearance dialog boxUtilities This implies the Utility settings in the Plot

Specification dialog boxPlot Domain This implies the settings in the Plot

Domaindialog boxTable 5.2: Paste plot format options

5.4.5 Substitute job

This command is not available for empty presentations. The command substitutes a job numberwith another in all plots in the selected page or in the entire presentation.

The Substitute Job dialog box contains two text fields. One is for entering the job number to besubstituted, and the other is for entering the new job number. In addition, the dialog box containstwo radio buttons for selecting whether to apply the substitution only to the displayed page orto all pages in the presentation.

Note that when substituting a job with one with greater grid resolution, the minimum and max-imum grid indices in the plot specification is unchanged.

5.5 Page menu

The Page menu supplies commands for adding new pages, and editing the displayed page. Thenumber of plots in a page is defined by the subdivision in the horizontal and vertical directions.A page includes a time specification which decides the time units used for the page, in addition tominimum, maximum and delta timestep used for displaying the page.

5.5.1 Add

Shortcut: CTRL+A

This command adds a new page after the displayed page. The new page contains one plot.


166 Flowvis

5.5.2 Modify

Shortcut: CTRL+M

The Modify command makes it possible to change parameters for the displayed page, such assubdividing the page into several plots.

The upper frame is for Time Specification. It contains an options menu for selecting Time Units.Available units are Seconds and Milliseconds.

The rest of the Time Specification area is included only if at least one plot in the displayed pagecan be plotted for several timesteps. The next field is for editing the Minimum Timestep for thedisplayed page. Another field shows the Maximum Timestep for which there are datas for thispage. The latter field is not editable. To modify the maximum timestep, press the Fixed togglebutton and modify the Fixedfield. In addition, the Delta Timestepfield decides the delta used whenprinting all timesteps (using the Print command in the File menu).

The second area in the dialog box is the Subdivision of Page. This area contains two fields forsetting number of plots in the horizontal (x) and vertical (y) direction for the displayed page.When a new page is created, it contains only one plot. Note that if the number of plots is reduced,the remaining plots are deleted without any warning.

5.5.3 Single colours

This command modifies the single colours used in the displayed page.

It contains a set of sliders, a push button and a colour display in addition to the Ok and Cancelpush buttons. The first scale is for selecting a single colour index. The corresponding colour isindicated by an arrow in the single colour display. The single colour display contains a horizontalrow of colour rectangles, one for each single colour. In addition to using the index slider, a colourindex can be selected by clicking the mouse in the colour display.

There are three scales for setting Hue, Light and Saturation for the selected colour. Pressing the In-terpolate push button causes interpolation of the hue, light and saturation values from the secondto the last colour.

5.5.4 Range colours

This command modifies the range colours used in the displayed page.

At the top, there is an area called Range Colours Setup, which contains radio buttons for selectingthe number of different colours and the number of shades (light/saturation values) per colour.There are a total of 220 range colours available. These can be organized as 220, 18, 12, or 6different colours (hues) with 1, 10, 14 or 22 shades for each colour. The light and saturation rangeis used for shading 3 dimensional plots.

The next area is labelled Hue for Range Colour, and contains two scales, one for selecting a rangecolour index, and one for setting the hue for the selected colour.

The range colour display contains two horizontal rows of colour rectangles. The upper row con-tains one rectangle for each range colour, while the lower row shows the light and saturationrange for the selected colour. The selected colour is indicated by an arrow. Other colour(s) areselected by clicking MOUSE+LEFT in or below the desired rectangle.

The next area is called Light and Saturation Range for each Colour, and contains scales for settingthe minimum and maximum light and saturation for each range colour.


5.6 Plot menu 167

5.5.5 Header

This command turns page header on/off, and modifies the contents of the header.

At the top of the Page header dialog box is a toggle button for turning the dispay of page headerson/off. This can also be done using the Header On command in the Options menu.

Below the Header On/Off toggle button is an area for deciding the contents of the page. Thesesettings can be applied to all the pages in the current presentation, or to the displayed page only.

The items that can be included in the header are the Flowvis version string, the current date, thepage number and a user defined string.

5.5.6 Next page

Shortcut: Page Down

Shows the next presentation page.

5.5.7 Previous page

Shortcut: Page Up

Shows the previous presentation page.

5.5.8 Next time step

Shortcut: RIGHT

Shows the next time step.

5.5.9 Previous time step (left arrow)

Shortcut: LEFT

Shows the previous time step.

5.6 Plot menu

The Plot menu contains commands for creating and editing plots.

5.6.1 Plot type

This command is available only if the selected plot is empty. The Data Selection dialog box popsup after a plot type has been selected (and the selected plot type is not Annotation or Scalar TimeAnnotation).

The following plot types are available in Flowvis:

• Scalar Line plot. This is a curve plot where the value of one variable is plotted along one gridcurve. Output from several simulations and for several phases can be combined.


168 Flowvis

• Scalar Time plot. This is a curve plot where the value of one variable is plotted along the timeaxis. Output from several simulations and for several monitor points can be combined inone plot.

• Grid plot. This is a 3 dimensional plot of the grid.• Monitor plot. This is a 3 dimensional plot where the monitor points are displayed together

with the geometry.• 2D Cut Plane plot. This is a 2 dimensional plot of a grid plane, or part of a grid plane. The

plot can include output for none, one or two variables. If two variables shall be combined,one must be a scalar variable and the other must be a vector variable. Scalar variables aredisplayed using filled contours, and the vector variables are displayed as arrows. The plotcan include output from several simulations. Grid, geometry and porosities can be includedin the plot.

• Monitor Points, Ignition Region and Panels. The colours for the monitor points define thepoint classifications as described above. This is also true for the ignition region if it is apoint and not a region. Only the points/planes that intersect with the displayed cut planeare drawn. In addition, the geometry is now drawn in colours if no data is displayed.

• 3D Cut Plane plot. This is a 3 dimensional plot of a set of grid planes. The type of output isidentical to that of 2D Cut Plane plots, except that contour curves cannot be drawn, and thegeometry is drawn as a 3 dimensional shaded model. By combining many cut planes andby limiting the visible value range for the scalar variable, a 3 dimensional view of the valuerange is obtained (honeycomb plot).

• Volume plot. This is a 3 dimensional plot where the values of a variable are visualized as araytraced volume. Output for one scalar variable for one simulation can be visualized inthis plot type. A 3 dimensional shaded model of the geometry can be included.

• Particle Traces plot. This is a 3 dimensional plot where a velocity field is visualized by tracinga set of particles in the field.

• Annotation plot. This is a text plot which extracts text information from other plots in thesame page. Repetition of the same text in several plots are then avoided.

• Scalar Time Annotation plot. This is a text plot which contains a table of minimum andmaximum values for Scalar Time plots in the same page.

A plot is selected by clicking the MOUSE+LEFT inside the plot, or by clicking the Select Previous orSelect Next Plot push buttons.

A plot is selected by clicking MOUSE+LEFT inside the plot. If the selected plot is empty, it can bedefined using the Plot Type command in the Plot menu. The available plot types are describedbriefly below.

Plot type Timespecifica-tion

No. ofjobs

No. ofvariables

Geometrydisplay

Griddisplay

Porositydisplay

ScalarLine

Timestep n 1 None (Theabscissa isa grid line)

None

ScalarTime

Timerange

n 1 None None None

Grid None n 0 Volume Volume NoneMonitor None 1 0 Volume None None2D CutPlane

Timestep n 0 - 2 2D cut 2D cut 2D cut

3D CutPlane

Timestep n 0 - 2 Volume 2D cuts 2D cuts

Volume Timestep 1 0 - 1 Volume None None


5.6 Plot menu 169

Part.Traces

Timestep 1 0 - 1 Volume None None

Annota-tion

Timestep 1 0 None None None

Scal.T.Annot

Timerange

n n None None None

Table 5.3: Summary of plot types in Flowvis

5.6.2 Data selection

This command is available only if the selected plot is not empty and not an Annotation plot. Thecommand lets the user change the selection of jobs, variables, phases, monitor points and panelsfor the selected plot. Legal selections depends on the plot type.

Figure 5.3: Flowvis data selection

The box contains four lists:

Job Numbers Contains the job numbers for all scenario and grid files found in the current di-rectory

Variables All variables that can be plotted. If a variable is not listed here, it was not selected foroutput in CASD

Phases Selecting phases. This is currently not used in FLACS

Monitors Monitor points or monitor panels

Plot type No. of jobs No. ofvariables

No. of phases No. of monitorpoints

Scalar Line n 1 n 0Scalar Time n 1 1 nGrid n 0 0 0Monitor 1 0 0 n2D Cut Plane n 0 - 2 n 03D Cut Plane n 0 - 2 n 0Volume 1 0 - 1 (scalar) 1 0


170 Flowvis

Particle Traces 1 0 - 1 (vector) n 0Table 5.4: Valid data selections depending on plot type

For Grid plots, 2D Cut Plane plots and 3D Cut Plane plots, several jobs which are part of onemultiblock simulation can be combined in one plot.

5.6.3 Plot specification

This command is available for all plot types except Annotation plots. The contents of the PlotSpecification dialog box depends on the plot type. It contains controls for manipulating valuerange, the display of axis, geometry etc.

Figure 5.4: Flowvis plot specification

The contents of the Plot Specification dialog box depends on the plot type for the selected plot.

On top of the dialog box for all plot types that are drawn for one timestep at a time, there is a textfield labelled Relative Timestep. This is the timestep for the selected plot relative to the timestepfor the page.

Instead of Relative Timestep, the Plot Specification dialog box for Scalar Time plots contains an areafor setting the Time Range for the plot.

The Plot Specification dialog box for Monitor plots contains toggle buttons for turning on/offprojections of the monitor points to the xy-, yz- and xz-planes, and a toggle button for turninggeometry extent marking on or off.

The Plot Specification dialog box for all plot types has an area for selecting Utilities. The utilitiesthat are available for a plot, depends on the plot type. A short description of all the available plotutilities is given below.

Utility ScalarLineplot

ScalarTimeplot

Gridplot

Moni-torplot

2D CutPlaneplot

3D CutPlaneplot

Vol-umeplot

ParticleTraces

Axis x x x x x x x x


5.6 Plot menu 171

Legend x xScalarLegend

x x x

VectorLegend

x x x

Grid x xGeom-etry

x x x x x

AreaPorosi-ties

x x

Vol.Porosi-ties

x x

DepthShad-ing

x x x

Trilin-ear

x

Table 5.5: Plot utilities

Axis Turns on or off the display of the axis.

Legend Turns on or off the display of a legend explaining the meaning of a curve colour or font.

Scalar Legend Turns on or off the display of a legend annotating the value to colour or phase tocolour mapping.

Vector Legend Turns on or off the display of a legend annotating the value to colour and valueto vector length mapping.

Grid Turns on or off the display of the grid.

Geometry Turns on or off the display of the geometry.

Area Porosities Turns on or off the display of the area porosities.

Volume Porosities Turns on or off the display of the volume porosities.

Depth Shading Turns on or off depth shading. If depth shading is turned off, flat shading isapplied.

Trilinear Turns on or off trilinear interpolation. This type of interpolation gives a more accuratereconstruction of the data. (It takes longer time to render the plot.)

5.6.4 Plot domain

This command is available for all plot types except Scalar Time, Monitor and Annotation plots.The purpose of the Plot Domain dialog box is to define the part of the simulation volume todisplay.

On top of the Plot Domaindialog box is a list of all jobs included in the selected plot. One job isselected at a time, and the controls in the rest of the dialog box applies to this job.

The Plot Domaindialog box for Scalar Line plots contains radio buttons for selecting the directionof the grid curve to plot along. In addition, it contains scales for selecting the indices of the curve,


172 Flowvis

and minimum and maximum indices along the curve. For 2 dimensional simulations, only theminimum and maximum indices can be set.

The Plot Domain dialog box for Grid plots contains scales for selecting the minimum and maxi-mum grid indices in each direction.

The Plot Domain dialog box for 2D Cut Plane plots contains toggle buttons for selecting planeorientation in addition to scales for selecting the index of the plane, and minimum and maximumindices in the plane. For 2 dimensional simulations, only the minimum and maximum indicescan be set. A dialog box for verifying porosities pops up upon clicking MOUSE+LEFT inside theplot. The porosities in the indicated control volume is displayed in the dialog box.

For a 3D Cut Plane plot, the Plot Domain dialog box contains sliders for selecting the minimumand maximum grid indices in each direction in the simulation volume. The user can specifynumber of planes, step between each plane and start index for the planes to be drawn in eachdirection, using the spin boxes.

For a Volume and Particle Traces plots, the Plot Domain dialog box contains scales for selectingthe minimum and maximum grid indices in each direction.

5.6.5 Variable appearance

This command is available for all plot types except Grid, Monitor and Annotation plots, giventhat one (or two) variable has been selected in the selected plot. The content of the VariableAppearance dialog box depends on the plot type. For example, the curve colours and legend textscan be modified for curve plots.

Each page in a presentation has its own colour table, consisting of a set of single colours, and aset of range colours. The single colours are used for plotting curves (Scalar Line and Scalar Timeplots), while the range colours are used for filled contours and shaded plots (2D, 3D Cut Planeand Volume plots). There are a total of 220 range colours available. These can be organized as220, 18, 12, or 6 different colours (hues) with 1, 10, 14 or 22 shades (light and saturation values)for each colour. The light and saturation range is used for shading 3 dimensional plots.

5.6.5.1 Scalar Line and Scalar Time plots

The Variable Appearance dialog box for Scalar Line and Scalar Time plots contains four areas. Theuppermost area is the Entry Listcontaining a list of all entries in the plot. The settings in the restof the dialog box applies to the selected entry.

Below the entry list is an area for Value Range, controlling the plotted value range for the selectedvariable. The area contains radio buttons for selecting Linear or Logarithmic interpolation. Linearinterpolation is default. Logarithmic interpolation is not available for particle plots.

Below the interpolation buttons are radio buttons for selecting Automatic, Fixed, or Variable valuerange. Automatic means that the minimum and maximum values used in the plot equal theminimum and maximum values found on the data file (for all timesteps). Selecting Fixed valuerange, and entering the desired minimum and maximum values, makes it possible to manuallycontrol the plotted value range. For Automatic and Fixed value range, the plotted value range isconstant for all timesteps. Selecting Variable, makes Flowvis adjust the plotted value range to theminimum and maximum values for the displayed timestep. Variable value range is not availablefor Scalar Time plots.

The Single Colour Display contains one rectangle for each single colour. The single colour asso-ciated with the selected entry in the entry list is marked with an arrow. Another colour can beselected by clicking MOUSE+LEFT in or below the desired rectangle.


5.6 Plot menu 173

Below the colour display are two option menus for selecting Curve Type and Marker Type. TheIndex scale can be used for selecting the single colour index for the selected entry. The legend textfor the selected entry can be edited in the Legend Textfield .

5.6.5.2 2D Cut Plane, 3D Cut Plane, Volume and Particle Traces plots

The contents of the Variable Appearance dialog box for these plot types depends of the contents ofthe selected plot, but three areas are always present. The first is the Entry list which contains alist of all entries in the plot. The settings in the rest of the dialog box applies to the selected entry.

The second area always present is the Value Range settings. This area is described in the previoussection.

The third area always present is the Range Colour Display which contains two horizontal rowsof colour rectangles. The upper row contains one rectangle for each range colour, while thelower row contains the light and saturation range for the selected colour. If the selected entryis a variable plotted for only one phase, two arrows mark the colour range that is mapped tothe value range. If the current entry is a variable plotted for several phase, only one arrow isdisplayed. It marks the colour used for plotting the current variable/phase combination. Othercolour(s) are selected by clicking MOUSE+LEFT in or below the desired rectangle.

The contents of the rest of the dialog box depends on the entry that has been selected.

The selected entry is a variable plotted for only one phase

If the selected entry is a variable plotted for only one phase, two areas controlling the value tocolour mapping are included below the range colour display. The first one contains fields dis-playing the Minimumand Maximum Values in the plot, and Indexscales for setting the correspond-ing range colour indices.

The next area controls the Colours for Values less than Minimum or greater than Maximum. It containstwo sets of radio buttons. The first controls how to display values that are less than the minimumvalue, and the second controls how to display values that are greater than the maximum value.The alternatives are to display the value in the same colour as the minimum/maximum values, inthe previous/next colours in the colour table, or not to use any colour at all. The last alternative isuseful when making a honeycomb plot in a 3D Cut Plane plot (by specifying a fixed value rangeless than the total in the Plot Specification dialog box)

If the plot type is 2D Cut Plane plot, and the selected variable is a scalar variable, an options menufor selecting Draw mode is included in the dialog box. The available options for drawing sin-gle phase scalar variables are Filled (filled contours), Contours (iso-curves), or Filled and Contours(filled contours combined with contour curves). The Contours command controls contours andcontour annotation. The Contours dialog box is described below.

If the selected entry is a variable plotted for several phases, a scale and a text field are includedbelow the range colour display. The scale can be used for selecting the range colour index for theselected entry, while the legend text for the selected entry can be edited in the text field.

5.6.6 Contours

This command controls contours and contour annotation in 2D Cut Plane plots. The command isavailable for 2D Cut Plane plots with scalar variable output for one phase, and is only availableif the current 2D plot is drawing contours.

This dialog box controls contours and contour annotations in a 2D Cut Plane plot. To turn oncontour drawing, select Contours (iso-curves), or Filled and Contours from the Draw options menuin the Variable Appearance dialog box.


174 Flowvis

At the top of the dialog box is a scale for setting the number of contours. The next area is theContour List which contains a list of all contour levels. The settings in the rest of the dialog boxapplies to the selected entry.

The Value at Contourfield is for editing the value of the selected contour, while the Linear Distribu-tion push button is for performing linear distribution of all the contour values. Pressing the ClearAnnotationpush button clears all annotation for all timesteps.

To annotate a contour, press MOUSE+LEFT near the desired contour in the desired position Thecontour annotation is only displayed at the time step for which it has been created. If the plotis changed in a way that makes the annotation wrong, the annotation must be cleared first. Apop-up dialog box reminds the user of this.

5.6.7 Vectors

This command is available for 2D and 3D Cut plane plots with vector output. The commandcontrols the vector scaling, the vector projection mode, the vector arrow type and the vectorlegend contents.

At the top of the Vectors dialog box are two radio buttons for selecting automatic or fixed vectorscaling. Automatic, which is default, means that the vectors are scaled to fit the minimum controlvolume sizes. Fixed means that the user specifies the scaling factor.

Below the scale controls are toggle buttons for controlling the vector legend. The user can selectto include or not include arrows and colour boxes in the legend.

The next control is for selecting projection mode. This is only applicable for 3D plots. The usermay then select to draw the real 3 dimensional vectors, or the projections of the vectors onto thecut planes.

The last controls decides the arrow types and opening angles for the arrows.

5.6.8 Geometry appearance

This command is available for 3 dimensional plots containing geometry, and not containing grid,porosities or variable(s). The Geometry Appearance dialog box contains one toggle button forturning on/off geometry shading. When shading is turned off, the geometry is displayed asa wireframe model. For plots containing grid, porosities or variable(s), only shaded geometry isavailable.

5.6.9 Blank and un-blank primitives

This command is available for 3 dimensional plots containing geometry. The command lets theuser hide selected primitives in the geometry plot.

The dialog boxes for blanking and unblanking are identical. When the dialog box is shown, thegeometry is drawn as a wireframe model. If the command is Blank Primitives, all visible (notblanked) primitives are drawn. If the command is Unblank Primitives, all blanked primitives aredrawn.

Primitives are selected from the graphical area by clicking MOUSE+LEFT on any point insidethe primitive. To simplify the selection, the dialog box contains push buttons for traversing theprimitive list. Pressing the Closer push button selects the next primitive in the direction towardsthe user, while pressing the Farther push button selects the next primitive in the direction awayfrom the user.


5.6 Plot menu 175

Pressing the Blank or Unblank Primitive push button marks the primitive for blanking/un-blanking. Pressing the Apply push button redraws the wireframe model with selected primitivesremoved.

This Unblank Primitives command is available for 3 dimensional plots containing geometry, andwhere one or several primitives have been blanked. The command lets the user select primitivesfor un-blanking in the selected plot.

5.6.10 Rotate

This command rotates the contents of the selected plot. The command is not available for emptyplots or annotation plots.

5.6.10.1 2D and line plots

While the mouse pointer is placed inside the Rotate dialog box, the following key functions areavailable:

X, x Mirror about x-axis

Y, y Mirror about y-axis

Z, z Rotate about z-axis in steps of 90 degrees.

Pressing the upper case keys gives rotation in counter-clockwise (positive) direction, while lowercase keys gives rotation in the clockwise (negative) direction.

The dialog box contains a push button labelled Default View. Pressing this button reinstalls thedefault transformation for the plot.

5.6.10.2 3D plots

While the mouse pointer is placed inside the Rotate dialog box, the following key functions areavailable:

X, x Rotate about x-axis

Y, y Rotate about y-axis

Z, z Rotate about z-axis.

Upper case keys means positive rotation angle while lower case means negative.

The Rotate dialog box contains an options menu for selecting a projection. Available projectionsare 3D View, XY View, YZ View and XZ View.

While rotating the plot, Flowvis draws a wireframe box marking the simulation volume extentinstead of the plot contents. The Apply push button causes the entire plot to be redrawn.

5.6.11 Particle traces specification

This command is only available for Particle Traces plots. To create a Particle Trace plot the variableVVEC must be saved during the simulation and plotted in Flowvis.


176 Flowvis

Figure 5.5: A particle trace plot

The Particle Traces Specification dialog box contains four frames. The first from the top is forselecting Velocity Field Type, Trace Time, Number of Traces and Minimum Velocity. The next framesare the Firing Specification, the Firing Volume Specification and the Animation Specification

There are two choices for the Velocity Field Type. Steady State means that the particles are tracedin a constant velocity field given by the current timestep. Transient means that the particles aretraced in a varying field from the current timestep to the last.

The innermost loop of the algorithm for drawing the particle traces is a loop through all particlesdrawing the particle traces over a time interval given by the Trace Time. Pressing Ok or Applyfor Steady State draws only one trace. The Trace Time decides the length of the particle tracesdirectly. For Transient, the traces are drawn from the current timestep to the last. The No. of Tracesthen decides the number of traces required for drawing the plot. A short trace time increases thedrawing time, but gives more accurate traces, and longer trace lines.

If the velocity of a particle gets lower than the Minimum Velocity, the particle is removed fromthe plot and the trace of that particle is ended. In an animation, the particle disappears from thescreen.

The Firing Specification frame contains radio buttons for selecting Random or Fixed Firing Positions.Random means that the particles are fired at random positions inside the volume given by theFiring Volume Specification. The number of particles fired are given by the No of Particles field.If Fixed firing positions is selected, the particles are fired from a fixed grid given by the No ofParticles in X, Y and Z-dir. fields, and the Firing Volume Specification (frame below).

The Animation Specification frame controls particle animation. The No. of Frames Between Firing pa-rameter gives the interval at which Flowvis fires new particles. Flowvis then fires a new particlefor each particle that has been killed since the last time. The two next fields decides the minimumand maximum lifetime for a particle. The Trace Time decides the duration of each frame. A shortTrace Time gives shorter traces.

An animation is started by pressing the Draw Frames push button. If the Record toggle button is


5.6 Plot menu 177

selected, the frames are stored in memory, and can be replayed using the Play push button. Theframes may also be stored on files, by selecting the Record on files toggle button. The files arestored on the directory where Flowvis was started, and named fl_framedd.png, where dd isthe frame number.

Remarks:

In the current version of Flowvis recording of particle plot images to file is not working.

Figure 5.6: The particle trace properties window

5.6.12 Next plot

Shortcut: CTRL+RIGHT

Choosing Next Plot causes the selection of the next plot. If the previously selected plot was thelast plot in the page, the page is selected. If the page was selected, the first plot in the page isselected.


178 Flowvis

5.6.13 Previous plot

Shortcut: CTRL+LEFT

Choosing Previous Plot causes the selection of the previous plot. If the previously selected plotwas the first plot in the page, the page is selected. If the page was selected, the last plot in thepage is selected.

5.7 Verify menu

The verify menu enables the user to have a detailed look at the area and volume porosities, andcoordinates for a geometry, and pick exact values from lines curves.

5.7.1 Porosities

This command is available only for 2D Cut Plane plots. In addition to porosities, this commandmakes it possible to verify coordinates, areas and volumes for control volumes.

This command lets the user select a control volume and verify the porosities, areas and volumefor this control volume. Two dialog boxes pops up. One is the Plot Domain dialog box. It is usedfor selecting the cut plane to view. The other dialog box is the Verify Porosities dialog box. Theselected control volume is shaded in the plot.

Since a 2D Cut Plane plot may include several jobs, the Verify Porosities dialog box includesa field showing which job the selected control volume belongs to. In addition there are fieldsdisplaying the selected control volume indices, and push buttons for navigating through thecontrol volumes.

The four next fields show the area and volume porosities for the selected control volume.

At the bottom of the dialog box are two push buttons for verifying Coordinates and Areas and Vol-umes. Clicking the Coordinates push button makes the Verify Coordinates dialog box pop up. Thisdialog box contains 8 x 3 fields showing the X, Y and Z coordinates for the eight corners of theselected control volume. Clicking the Areas and Volumesbutton makes the Areas and Volumesdialogbox pop up. This dialog box contains 3 x 3 fields showing the X, Y and Zcomponents of the areasin the three grid directions ( I, J and K). The last field shows the volume.

Flowvis can be started in a special verify porosity mode, from the command line by using thefollowing command line option.

Linux:

> run9 flowvis -verify_por 010101

Windows:

> flowvis -verify_por 010101

010101 is the job number.

5.7.2 Value on curve

This command makes it possible to verify the values in Scalar Time and Scalar Line plots.


5.8 Options menu 179

The Values On Curve dialog box lets the user select a point on the abscissa (time for Scalar Timeplots and coordinate value for Scalar Line plots). The corresponding value, or values if severalcurves, are shown in the value list in the dialog.

The point on the abscissa can be selected by clicking MOUSE+LEFT on the desired position in theplot, or by typing the abscissa value in the input field in the dialog, and then pressing the Selectbutton.

5.8 Options menu

Options are saved for each user under

Linux: ∼/config/GexCon/Flowvis.conf

Windows: In Windows registry "HKEY_CURRENT_USER\Software\GexCon\Flowvis"

5.8.1 Header on

This option specifies whether a header is displayed on top of each page. The header containsFlowvis version number, plotting date and page number.

5.8.2 Background white

By default the page background is white.

5.8.3 Redraw picture on OK

By default, the selected plot is redrawn when the user selects Ok from a popup dialog box whichchanges the contents of the plot. This can be avoided by turning off this option. The time neededfor changing a plot through the use of several dialog boxes can thus be reduced.

5.8.4 Interpolate

Flowvis interpolates the results between grid cells, by default when showing 2D, 3D and Volumeplots. In some situations it can be necessary to turn this off, for instance when investigatingpossible simulation errors.

5.8.5 Fonts

Font type and size can be changed separately for each of the text areas in a Flowvis plot. The fontsetting is global for all plots and pages in the presentation.

Remarks:

Flowvis will not be able to scale all elements in the plots according to the size and type of thefonts, thus in some cases the fonts will be covered, or cover the plot other graphical elements.Clicking the Restore Defaults button will normally alleviate this problem.


180 Flowvis

5.8.6 Plots

The Plots options enables some control of the position of plot elements, such as.

• Changing plot size• Moving the plot on the page• Moving and resizing the legend

In addition the user can select to have overrun text be highlighted to be warned if some textelements are not shown on the plot.

5.9 Help

The Help menu gives access to the online help and license terms.

5.9.1 Online help

Shortcut: F1

The online help is shown in the Qt Assistant help browser.

In Qt Assistant the user has full access to a searchable index in addition to a full text search in theuser manual. It is possible to have several tabs open, create bookmarks and print selected pages.

The FLACS User’s manual is also available as a single file pdf (Portable Document Format).

5.9.2 Licence terms

Shows the FLACS license terms, including the specific license type for the specific FLACS instal-lation.

5.9.3 About Flowvis

Shows about Flowvis dialog.

5.10 Flowvis examples

This section gives two examples on Flowvis use. Results from a FLACS simulation in the M24module are used in both examples. The grid resolution is 40x13x12 m3. The simulation files canbe copied from your FLACS installation directory.

Linux:

> cp /usr/local/GexCon/FLACS_v9.0/doc/examples/ex1/*110101* .

Windows:

Copy files from C:\Program Files\GexCon\FLACS_v9.0\doc\examples\ex1\∗110101∗(∗110101∗ means all files containing the text "110101")


5.10 Flowvis examples 181

The simulation must be run before creating the presentation:

On Linux:

> cd <simulation file directory>> run9 porcalc 110101> run9 flacs 110101

On Windows start the FLACS runmanager, click the Add directory button, and select the directorywhere the simulation files are located. Select the job number in the list, and click the Simulatebutton.

When the simulation has finished continue to the next section.

5.10.1 Creating a presentation with multiple plots

This example illustrates how to create a simple presentation from scratch.

In the examples, there are several tasks which implies the use of several dialog boxes to define aplot. This can be time consuming because the plot is redrawn every time an Ok button is clicked.To avoid this, turn the Redraw Plot On Ok option off (in the Options menu). Use the Redraw Plotbutton in the main window when a redraw is wanted.

This example covers 2D and 3D Cut Plane plots, Scalar Time plots and Monitor Point plots.

The presentation shall contain two pages. The first shall contain 2 dimensional cut plane plots(the JK cut plane with grid index equal to 20) showing the combustion product, pressure andvelocity at timestep 17. There shall also be a plot showing the position and orientation of the cutplane compared to the geometry. The porosities in the cut plane shall also be visualized.

The second page shall contain a scalar time plot and a plot showing the positions of the monitorpoints. At last, it is outlined how to adapt the presentation to the results from a new simulation.

1. Start Flowvis2. Change directory to where the simulation results are

(a) Select the New At command from the File menu. The New At dialog box makes itpossible to navigate through the directory structure by double clicking on directorynames. When the desired directory is reached, press the Ok button.

3. Create a page

(a) Select the Add Page command from the Page menu. Select the Modify Page commandand change the subdivision of the page to 2 plots in X-direction and 3 in Y-direction.

4. Create a plot which shows the geometry and cut planes in 3 dimensions

(a) A 3D Cut Plane plot is used for this purpose. Click MOUSE+LEFT while pointing themouse pointer somewhere near the top left corner of the page. A frame indicates theselected plot.

(b) Select the Plot Type command from the Plot menu. Select 3D Cut Plane plot from thesubmenu. The Data Selection dialog box pops up. Select the appropriate job number.When the variable list is displayed, select PROD (combustion product). When the Okbutton is pressed, a plot with default settings is displayed. This means that three cutplanes are displayed, one perpendicular to each axis direction.

(c) Select the Plot Domain command from the Plot menu. The Plot Domain dialog box popsup. Set the number of IJ and IK planes equal to 0 (keep the number of JK planes equalto 1). The Start Index for the JK plane is set to 20. Press the Ok button.


182 Flowvis

(d) Select the geometry utility in the Plot Specification dialog box, and press the Ok button.

5. Some primitives need to be blanked to make it possible to see inside the geometry.

(a) Select the Blank Primitive command from the Plot menu. The geometry is redrawn asa wireframe model.

(b) When the dialog box pops up, click MOUSE+LEFT while pointing the mouse pointer atthe roof of the geometry. The selected primitive is highlighted.

(c) Press the Blank Primitive button to blank the roof. Do the same thing to the XZ wallclosest to the viewer. If it is difficult to hit the wanted primitive when clicking, theCloser and Farther buttons makes it possible to loop through all primitives intersectingthe mouse pointer position.

(d) The plot is redrawn with shaded geometry when pressing the Ok button.

6. Create an Annotation plot

(a) Select the plot to the right of the previous plot by clicking MOUSE+LEFT while point-ing the mouse pointer somewhere in that area. Select the Plot Type command, andAnnotation from the submenu.

(b) The Annotation plot is now displayed. It contains the job number, iteration number,time, minimum and maximum grid indices from the 3D Cut Plane plot. The corre-sponding text is removed from the 3D Cut Plane plot which is redrawn because theremoved text makes room for larger graphics.

7. Create a 2D Cut Plane plot for visualizing the porosities

(a) Select the plot below the 3D Cut Plane plot. Select the Plot Type command and 2D CutPlane from the submenu. Select the appropriate job number in the Data Selection dialogbox, and press the Ok button.

(b) Select the Plot Specification command. Select the Area Porosities and Volume Porositiesutilities and deselect the Axis and Grid utilities. Press the Ok button.

(c) Select the Plot Domain command. Select the JK plane and the appropriate grid index(20) using the I indexslider. Press the Ok button to see the result.

(d) Note that the Annotation plot changes when this 2D Cut Plane plot is created. In-stead of the minimum and maximum grids indices, the Annotation plot now containsinformation about the plane displayed in the 2D Cut Plane plot.

8. Create 2D Cut Plane plots of the combustion product and the pressure

(a) Select the Copy command from the Edit menu to copy the selected plot to the clipboard.Select the plot to the right, and select the Paste command. A copy of the previousplot is displayed. Select the Data Selection command from the Plot menu, and selectthe combustion product (PROD) from the variable list. Press Ok and select the PlotSpecification command. Deselect the Area Porosities and Volume Porosities utilities andselect the Geometry utility. Press Ok to see the result.

(b) Copy this plot to the clipboard and paste it in the next plot. Select the Data Selectioncommand and select the P variable instead of PROD.

9. Create a 2D Cut Plane plot of the velocities

(a) Select the last plot in the displayed page and select the Paste command. The contentsof the clipboard is copied to the selected plot. Select the Data Selection command andselect the VVEC variable instead of PROD.

(b) To get longer vectors, select Variable Value Range in the Variable Appearance dialog boxand select Fixed Scale in the Vectors dialog box on the Plot menu. Set the scaling factorequal to 1.5.

10. Add a new page

(a) Select the Add command from the Page menu. The new page is added after the previ-ous. Modify the page subdivision to be two plots in Y-direction.



11. Create a Scalar Time plot of the Pressure

(a) Select the top plot in the page. Select the Plot Type command and Scalar Time fromthe submenu. Select the appropriate job, the P variable (pressure) and the wantedmonitor points in the Data Selection dialog box. A sequence of adjacent (in the list)monitor points can be selected by moving the mouse pointer over the items whileholding MOUSE+LEFT down. Several monitor points can also be selected by clickingwhile holding the Ctrl key down. Upon Ok, the plot is drawn.

12. Create a Monitor Point plot showing the points used in the previous plot

(a) Select the bottom plot, and select the Plot Type command and Monitor from the sub-menu. Select the appropriate job, and the same monitor points as in the previous plot.

(b) Press the Monitor Point Listtoggle button in the Plot Specification dialog box.

13. Save the presentation

(a) Select the Save command from the File menu. Since the presentation is new, no filename is associated with it. The Save As dialog box is therefore displayed. Specify aname and press the Ok button.

14. Adapt the presentation to the results from a new simulation

(a) Start a new Flowvis instance. Select the Open command from the File menu. Selectthe file specified in the previous step. Select the Substitute Job command from the Editmenu. Specify the previous and the new job numbers. Select the All Pages option andpress the Ok button.

Figure 5.7: The first page of the Flowvis presentation in example 1 (location of plots differsslightly from text)


184 Flowvis

Figure 5.8: The second page of the Flowvis presentation in example 1 (location of plots differsslightly from text)

5.10.2 Creating a honeycomb plot

This example shows how to create a honeycomb plot by specifying a limited visual value rangein a 3D Cut Plane plot.

1. Create a 3D Cut Plane plot

(a) Open the presentation from the previous section.(b) Copy the 3D Cut Plane plot from the first page.(c) Add a new page and paste the plot from the clipboard.(d) Select the Plot Domain command. Change the number of cut planes in each direction

to be equal to the number of grid indices in that direction. The number of grid indicescan be seen from the scales in the volume specification.

(e) Turn off the grid display in the Plot Specification dialog box.

2. Modify the variable appearance

(a) Select the Variable Appearance command. Select Fixed Value Range Setting, and set theFixed minimum value to be equal to 50% of the maximum value for All Timesteps. TheFixed maximum value is set to be equal to that of All Timesteps. Set the Colour for val-ues < min. to be None. Press the Ok button. The combustion product is drawn as ahoneycomb volume covering the volume where the combustion product is above thespecified limit.

3. Change the colour table

(a) To get more flame-like colours, select the Range Colours command from the Page menu.Select the first index using the index slider. Set the hue for this index to be equal to 0.



Select the last index and set this hue to be 60. Press the Interpolate button. Press Ok tosee the result.

Figure 5.9: The honeycomb plot


186 Flowvis


Chapter 6

Utility programs in FLACS

188 Utility programs in FLACS

This chapter describes the utility programs in FLACS. The user can run the programs from thecommand line in the terminal window under Linux, or on the command line in the cmd windowunder Windows. It is possible to start the cmd command shell from the FLACS Run Manager.

The utility programs are categorized according to application:

1. Geometry, grid, and porosities: geo2flacs, gm, and Porcalc2. Release source modelling: Jet, Flash3. Modifying simulation files: rdfile, cofile, comerge4. Post-processing simulation data: r1file, r3file, a1file

6.1 Geometry, grid and porosities

6.1.1 geo2flacs

geo2flacs is a utility designed to import geometry from various 3D geometry formats to FLACS.The following input formats are supported:

• Microstation dgn version 7• AutoReaGas• AutoCAD (when contained in dgn file)• StereoLitography file (STL)

In addition to the dgn format geo2flacs can export from the prp format (Propagated steel, Frame-works format, very similar to dgn format). Often the dgn files are accompanied by a drv file(Design Review). geo2flacs can use the information in the drv file to group single primitives intolarger objects and give the objects the proper names.

The dgn file format can be produced by a number of different software packages. It is the nativeformat of Microstation, but ie. PDMS can be exported to dgn format using the PDMS moduleExPlant-I.

Microstation can open AutoCAD files and save them in dgn format. The internal structure of thegeometry is not changed, and export from these types of files sometimes result in incomplete orcompletely empty geometries.

The output format is CASD macro files.

6.1.1.1 Usage

geo2flacs is a command line program using mandatory and optional arguments.

The conversion is a three-step process. During the export process itself the geometry files areread via an input file.

Running the program with

"--help"

option prints a help message:

Linux:

> run9 geo2flacs --help


6.1 Geometry, grid and porosities 189

Windows:

> geo2flacs --help

The above command gives in the following output:

Usage: geo2flacs [OPTION...] DIRECTORYGeometry conversion tool for FLACS. Converts Microstation,AutoReaGas, STL, and other 3D geometry formats, to CASDmacro files.

-a, --noAlign Do not align objects to nearest axisand allow irregular objects (specialversion of CASD required)

-c, --create Create input file-d, --minDia=DIAMETER Do not export objects with diameter

smaller than DIAMATER (mm)-e, --export Start export of files-f, --force Force delete output directory before

exporting-g, --gexconCol Gexcon colour scheme-l, --minLength=LENGTH Do not export objects with largest

length smaller than LENGTH (mm)-n, --group=NUMBER Group NUMBER of primitives into one object-o, --outdir=DIR Output sub-directory (default = out)-r, --rotate=ANGLE,X,Y,Z Rotate geometry ANGLE degrees around (X,Y,Z)-s, --translate=X,Y,Z Translate geometry (X,Y,Z) before export-t, --terrain[=HEIGHT] Use HEIGHT as bottom value for terrain

export-v, --verbose Produce verbose output (increase

diagnostic output)-x, --addLength Add small length to cylinders with 0

length which would otherwise be skipped-z, --buildingBase=Z0 Create buildings starting at Z0 (only valid for

Shape files)-?, --help Give this help list

--usage Give a short usage message-V, --version Print program version

Mandatory or optional arguments to long options are also mandatoryor optional for any corresponding short options.

Object type specification:P: PipingE: EquipmentS: SteelR: Race WayT: TerrainB: Bounding Box

Specify option in the "geo2flacs_files.txt" file.

Report bugs to <[email protected]>.

6.1.1.2 Step-by-step procedure

1. Begin the export process by creating a new directory and copy all files to be converted tothis directory.

2. Run geo2flacs first time with argument -c: "<tt>run9 geo2flacs -c .</tt>". This will createthe input file geo2flacs_files.txt in the current directory.

3. Open the input file geo2flacs_files.txt in a text editor (eg. emacs, vi, pico, kwrite, etc and addexport arguments as described in section Input file parameters. Save the file.



4. Run geo2flacs again with argument -e to do the export: "<tt>run9 geo2flacs -e .</tt>".This will create a sub directory out if it does not exist, and start the export. The exportedfiles are created in the out directory. If the out directory contains exported files geo2flacswill exit without exporting. A log file geo2flacs.log will be created.

5. Start CASD, create a database and read the macro files created by geo2flacs. To do this goMacro → Run ... and select the file 0-RunMe.mcr. The geometry will be read into FLACS.Note that if the geometry has more than 10000 objects (2000 in FLACS versions prior to 9.0)CASD must be started with arguments -numObj XXXXXX -numAsis XXXXXX, whereXXXXXX is the number of objects. The number of objects created is reported at the endof the export.

6.1.1.3 Input file parameters

If a dgn or prp file does not have an accompanying drv file geo2flacs must have informationabout the type of geometry inside the file. This is done in the input file geo2flacs_files.txt. Whenthis file is initially created all dgn files without a drv file will have a trailing "-". Change the "-"into one of the following letters.

Parameter Geometry typeP PipingE EquipmentS SteelR Race WayT TerrainB Bounding Box

Table 6.1: Input file parameters

The parameters P, E and S are most commonly used. R is used if the file contains only RaceWay elements. T is for importing grid or triangular terrain geometry. B can be used to exportgeometry from files geo2flacs otherwise is unable to export from.

6.1.1.4 Command line options

The following command line options are recognized by geo2flacs:

Short Option Long Option Description-a –noAlign Do not align objects to

nearest axis and allowirregular objects (specialversion of CASD required)

-c –create Create input file-d –minDia=DIAMETER Do not export objects with

diameter smaller thanDIAMATER (mm)

-e –export Start export of files-f –force Force delete output directory

before exporting-g –gexconCol Gexcon colour scheme-l –minLength=LENGTH Do not export objects with

largest length smaller thanLENGTH (mm)


6.1 Geometry, grid and porosities 191

-n –group=NUMBER Group NUMBER ofprimitives into one object

-o –outdir=DIR Output sub-directory(default = out)

-r –rotate=ANGLE,X,Y,Z Rotate geometry ANGLEdegrees around (X,Y,Z)before export

-s –translate=X,Y,Z Translate geometry (X,Y,Z)before export

-t –terrain[=HEIGHT] Use HEIGHT as bottomvalue for terrain export

-v –verbose Produce verbose output(increase diagnostic output)

-x –addLength Add small length tocylinders with 0 lengthwhich would otherwise beskipped

-z –buildingBase=Z0 Create buildings starting atZ0 (only valid for Shapefiles)

-? –help Give this help list–usage Give a short usage message

-V –version Print program versionTable 6.2: geo2flacs command line options

DIRECTORY is mandatory when using the "-c" or the "-e" argument. The directory can either bethe current directory (use ".") or any other directory (use full path or relative path).

Using the "-f" argument will force overwriting of any existing files in the out directory.

"-t HEIGHT" is used when exporting terrain data. If HEIGHT is negative use the long option, eg:

--terrain=-10.0

If one or more of the coordinates are negative when using the translate option the coordinatesmust be specified like this:

-s "\-100,\-100,\-100"

Note the double quotes ("") around the coordinates.

The argument "-g" force geo2flacs to export geometry using the Gexcon colour scheme. By defaultgeo2flacs will try to preserve the colours found in the dgn file.

"-a" is a special purpose option that prevents objects from being rotated to the nearest axis. Thisrequires a special version of CASD and is not validated for general use.

"-v" will output as much information as possible during the export.

6.1.1.5 Geometry verification

As with all conversion from one format to another the process cannot be guaranteed to producecorrect results. It is therefore very important to do a thorough verification of the geometry in



CASD. Because CASD only offers a subset of the functionality in most CAD packages some in-formation will be lost and some information might be interpreted with errors.

It is important to have special focus on details having significant influence on the simulation re-sults. These includes walls, decks (with and without grating) and major obstacles. The degree ofporosity of the grating can have a significant influence on explosions, but no porosity informationis exported.

6.1.1.6 Problems and solutions

Note that geo2flacs is under continuous development. This means that the software will havebugs, although Gexcon has done significant testing of the program to ensure that the programwill operate well in most situations. Contact GexCon if any bugs are found. GexCon will considerfixing the bug if it is part of the core functionality of the program.

Increased diagnostic output geo2flacs will output information during the export. Increaseddiagnostic is written if the "-v" option is used. This information can be redirected to a file byusing the ">" sign:

Linux:

> run9 geo2flacs -v -e . > out.txt

Windows:

> geo2flacs -v -e . > out.txt

This will create a file out.txt in the current working directory.

Dgn version supported The current version of geo2flacs supports dgn file version 7. Version 8is not supported and must be converted to version 7 using Microstation.

Note about PDMS ExPlant-I There is bug in ExPlant-I that sometimes exports geometry ina format that cannot be exported with geo2flacs. If an object is represented by only "Shape"elements it will not be exported. How to fix this in ExPlant-I is not known to GexCon, but aknowledgable ExPlant-I operator will be able to work around the problem.

6.1.2 gm

The gm program allows making grids by scripting. The following command starts the gm pro-gram:

Linux:

> run9 gm

Windows:

> gm


6.2 Release source modelling 193

The following sequence of commands creates a homogeneous grid of 1 m resolution in all direc-tions over a simulation domain extending from 0 m to 10 m in all directions.

Executing "gm"gm> x add 0:10:1gm> y add 0:10:1gm> z add 0:10:1gm> save cg999999.dat3

The grid is saved in a file named cg999999.dat3. Assuming a 999999-job number this grid can beused in a FLACS simulation. The gm program can read the FLACS cg-file and print the positionof the gridlines on the screen:

Executing "gm"gm> open cg999999.dat3gm> x printgm>x 110.000 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 10.000gm> y printgm>y 110.000 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 10.000

More options are available with the gm-program and can be listed typing the command ’help’ inthe gm-interface:

gm> help

6.1.3 Porcalc

The Porcalc program starts either from the ’Calculate’ option under the ’Porosities’ menu inCASD, or from command line input.

Linux:

> run9 porcalc 010100

Windows:

> porcalc 010100

6.2 Release source modelling

6.2.1 Jet

The jet program is used to compute the area and the subsonic velocity after shock of a jet issuingfrom a high-pressure reservoir. The jet program is started with the following command:

Linux:

> run9 jet < jet-in > jet-out

Windows:



> jet < jet-in > jet-out

Below is an example of a basic jet-in file:

cl-file ! output format (-, *, cl-file, ...).’METHANE=0.8,ETHANE=0.1,PROPANE=0.04,BUTANE=0.06’ ! gas type (AIR, METHANE, ...)100 ! reservoir volume (m3)50 30 ! reservoir pressure (barg) and temperature (C)1 30 ! atmospheric pressure (bara) and temperature (C)0 0 ! heat transfer coefficients (J/s) and (J/sK)30 ! wall temperature (C)0.1 0.62 ! nozzle diameter (m) and discharge coefficient (-)0 ! start time (s)1 10 ! time step (s) and number of iterations (-)"+XJ" ! leak control string1e-5 1e5 ! shutoff pressure (barg) and release mass (kg)0.1 ! relative turbulence intensity RTI (-)0.1 *D ! turbulence length scale TLS (m) + function

Most of the parameters can be defined intuitively, a brief description of the most unconventionalparameters is given below.

The first line says that the output of the jet program will directly be written with the format of acl-file, the leak file used to define gas releases in FLACS. This is the most useful output formatand it will be described further in the following.

The heat transfer coefficients do not need to be defined, the default zero values can be conserved.

The time step and the number of iterations control the duration and the times at which the prop-erties of the jet will be written in the cl-file. In this example, the properties will be written in thecl-file each seconds over a duration of 10 s.

The leak control string will be directly added into the cl-file for the definition of the direction ofthe leak.

The output file generated by the jet program used with the previous inputs is:

Starting jetEnter output format (-, *, cl-file, ...):Enter gas type (AIR, METHANE, ...):Enter reservoir volume (m3):Enter reservoir pressure (barg) and temperature (C):Enter atmospheric pressure (bara) and temperature (C):Enter heat transfer coefficients (J/s) and (J/s/K):Enter wall temperature (C):Enter nozzle diameter (m) and discharge coefficient (-):Enter start time (s):Enter time step (s) and number of iterations (-):Enter leak control string:Enter shutoff pressure (barg) and release mass (kg):Enter relative turbulence intensity RTI (-):Enter turbulence length scale TLS (m) + function:## GENERAL:# gas type = METHANE=0.8,ETHANE=0.1,PROPANE=0.04,BUTANE=0.06# mole weight = 21.093 kg/kmol# heat ratio, kappa = Cp/Cv = 1.291 -# critical pressure ratio = 0.547 -## RESERVOIR:# critical pressure = 0.827 barg# pressure = 50.000 barg# temperature = 30.000 C# density = 42.679 kg/m3


6.2 Release source modelling 195

# volume = 100.000 m3# initial mass = 4267.876 kg## MACH:# speed of sound, M=0 = 392.713 m/s# speed of sound, M=1 = 366.957 m/s# pressure, M=1 = 26.919 barg# temperature, M=1 = -8.460 C# density, M=1 = 26.758 kg/m3# maximum velocity = 1030.249 m/s## NOZZLE:# effective diameter = 27.953 mm# effective area = 613.675 mm2# discharge coefficient = 0.620 -# sonic massflow = 6.026 kg/s# jet force = 3863.166 N## ATMOSPHERIC:# pressure = 1.000 bara# temperature = 30.000 C# density = 0.837 kg/m3’+XJ’7’TIME (s)’ ’AREA (m2)’ ’RATE (kg/s)’ ’VEL (m/s)’ ’RTI (-)’ ’TLS (m)’ ’T (K)’

0.0000E+00 3.3311E-02 6.0258E-02 2.0741E+00 1.0000E-01 2.0594E-02 2.9086E+025.0000E-01 3.3311E-02 6.0258E+00 2.0741E+02 1.0000E-01 2.0594E-02 2.9086E+021.0000E+00 3.3280E-02 6.0209E+00 2.0739E+02 1.0000E-01 2.0585E-02 2.9080E+022.0000E+00 3.3219E-02 6.0112E+00 2.0734E+02 1.0000E-01 2.0566E-02 2.9068E+023.0000E+00 3.3159E-02 6.0015E+00 2.0730E+02 1.0000E-01 2.0547E-02 2.9056E+024.0000E+00 3.3098E-02 5.9918E+00 2.0726E+02 1.0000E-01 2.0528E-02 2.9045E+025.0000E+00 3.3037E-02 5.9821E+00 2.0722E+02 1.0000E-01 2.0510E-02 2.9033E+026.0000E+00 3.2977E-02 5.9724E+00 2.0718E+02 1.0000E-01 2.0491E-02 2.9021E+027.0000E+00 3.2917E-02 5.9628E+00 2.0714E+02 1.0000E-01 2.0472E-02 2.9009E+028.0000E+00 3.2857E-02 5.9531E+00 2.0710E+02 1.0000E-01 2.0453E-02 2.8997E+029.0000E+00 3.2797E-02 5.9435E+00 2.0706E+02 1.0000E-01 2.0435E-02 2.8985E+029.5000E+00 3.2797E-02 5.9435E-02 2.0706E+00 1.0000E-01 2.0435E-02 2.8985E+02

The first 15 lines of this file concern the input parameters defined in the jet-in file. The 34 follow-ing lines summarize the properties of the jet and physical parameters used in the computations.Finally, the last 15 lines constitute the cl-file that will be used by FLACS for a dispersion simu-lation. Therefore, providing the fact that in CASD, for a job number 999999 a leak position hasbeen defined, this file (with the 49 first lines removed) can be renamed as cl999999.n001 and di-rectly used for a dispersion simulation. In the new version of the jet program, specifying theoption -silent in the command line, automatically removes the 49 first lines of the output file. Thecommand line is:

Linux:

> run9 jet -silent < jet-in > jet-out

Windows:

> jet -silent < jet-in > jet-out

6.2.2 Flash

The flash program is a utility that computes the physical properties of flashing releases of pres-surized liquefied gas. The term flashing is usually used to describe vapor formation by pressurechanges. Many materials (such as propane, ammonia or chlorine for example) commonly storedas pressurized liquids in the industry can flash as released into the atmosphere.



For such pressurized storage conditions the release has, in the free atmosphere, the appearanceof a two-phase jet composed of droplets and vapor. The image below summarizes the thermody-namic state of the material as the distance with the release location increases.

Figure 6.1: The Flash utility

The end of the near field region is defined as the position where all the liquid in the jet haschanged phase. At this position, denoted xf , the jet is in a single vapor phase and assuming thatall the required properties needed to define a gas leak in FLACS are known, the flashing releasecan be treated as a so-called jet leak in FLACS.

The following command starts the flash program.

Linux:

> run9 flash

Windows:

> flash

The flash program can handle flashing releases of 9 different species, namely:

• acetylene• ammonia• butane• chlorine• ethane• ethylene• methane• propane• propylene

The following inputs are needed:

• Area of the exit orifice


6.3 Modifying simulation files 197

• Temperature of the liquefied gas at the exit orifice• Value of the discharge coefficient (by default this value is set to 0.62)• Mass flow rate or pressure at the orifice• Temperature of ambient air

Finally, from these five inputs the flash program gives the following outputs:

• Position xf where all the fluid is in a single vapor phase• Area of the jet at the position xf• Velocity of the jet at the position xf and mass flow rate• Mass fraction of air and released material at the position xf• Mass fraction of released material that rained-out and formed a pool on the ground.

In the previous list, the first three outputs can directly be used to define a leak in FLACS, thefourth output allow deriving the value for the equivalence ratio of the released material in FLACSand if significant, the last output should be part of a pool setup (see pool model).

6.3 Modifying simulation files

6.3.1 rdfile

The rdfile program is mainly used to adapt a certain dump-file to a new grid resolution. Considera job number 999999 with a given simulation volume and a given grid resolution. Consider a jobnumber 888888 with the same simulation volume than the job number 999999 but with a differentgrid resolution. Assume that all the other parameters of the two job numbers are the same andthat a dump file rd999999.n001 exists. The following command creates a dump file rd888888.n001from the dump file rd999999.n001:

Linux:

> run9 rdfile rd999999.n001 rd888888.n001

Windows:

> rdfile rd999999.n001 rd888888.n001

The command

Linux:

> run9 rdfile

Windows:

> rdfile

lists all the options available for use with the rdfile program:



Starting rdfileFLACS rdfile (version 1.3, 2005-06-24)Copyright 2005, GexCon ASusage: rdfile in_file out_file [options]options (enter in any order after file names):

force force overwriteversion=string set file versionsetup=file_name use setup fileconvert[=MIX,H,TAU] convert fieldsblock=number set block numbertime=value set timeiter=value set iteration count

6.3.2 cofile

The cofile program extracts ASCII data from the FLACS geometry file co-file. The program readsa co-file and writes obstacle size distribution to the screen. The specification of the option list=?with the cofile program leads to the full list of primitives. The cofile program can also list the totallength of cylinders/boxes for each diameter size classes providing the option(s) classes_cyl=c1,c2„or/and classes_box=b1,b2,.,.

The following command lists all the options available for use with the cofile program.

Linux:

> run9 cofile

Windows:

> cofile

The output of this command is:

Executing "cofile"FLACS cofile (version 1.0, 2005-05-02)Copyright 2005, GexCon ASusage: cofile file_name [options ...]options (enter in any order after file_name):

region=x,X,y,Y,z,Z region of interestclasses_cyl=c1,c2,,, classes of cylindersclasses_box=b1,b2,,, classes of boxesinch=0.0254 set inch to meter scaleaccu=0.001 set size accuracyplot=0/1 plot=no/yes (using gnuplot)list=0/1 list=no/yessilent=1/2 do not write to screen/file

6.3.3 comerge

The comerge program is used to create new FLACS geometry co-files from existing co-files. Con-sider a job number 999999 with a given geometry. The following command creates a new co-fileco888888.dat3 for the job number 888888:

Linux:

> run9 comerge region=x_min,x_max,y_min,y_max,z_min,z_max co999999.dat3 co888888.dat3

Windows:


6.3 Modifying simulation files 199

> comerge region=x_min,x_max,y_min,y_max,z_min,z_max co999999.dat3 co888888.dat3

Note: To ensure that region=... is a single argument it must either be without spaces or embeddedbetween quotation marks.

> ... region=x_min,x_max,y_min,y_max,z_min,z_max

or

> ... "region= x_min , x_max , y_min , y_max , z_min , z_max"

but not

> ... "region = ..."

The geometry of the new job number 888888 is the same than the geometry of the job number999999 in the region specified in the command line.

Several co-files from existing job numbers can be used to generate a new co-file. The command:

Linux:

> run9 comerge

Windows:

> comerge

lists all the options available for use with the comerge program.

Executing "comerge"FLACS comerge (version 1.2, 2005-05-10)Copyright 2005, GexCon ASusage: comerge [transform] input_file[s] output_file [force]enter transform before each input_file

force force overwriteregion=x,X,y,Y,z,Z region of interestinit identity transformtranslate:tx,ty,tz translate in x,y,z directionsturn:axis,angle,x,y,z turn around axis at point x,y,z

axis = x/y/z, angle = +-90*N (+ is CCW)skip:box skip boxesskip:cyl skip cylindersskip:col= skip objects with given colour hueonly:col= only objects with given colour huebox=cyl convert boxesbeam=cyl convert composite beams (T/I/H/U shaped)only only keep the converted objectsshow show the objects to be convertedskip skip the converted objectsmax_W=value set maximum beam widthmin_L/W=value set minimum beam length/width ratiomin_T/W=value set minimum beam thickness/width ratio (box beams)max_T/W=value set maximum beam thickness/width ratio (composite beams)



6.4 Post-processing of simulation data

6.4.1 r1file

The r1file program extracts ASCII data from the FLACS r1-file. Considering a job number 999999the basic command is:

Linux:

> run9 r1file r1999999.dat3 name=NP force

Windows:

> r1file r1999999.dat3 name=NP force

The previous command creates an ASCII file named a1999999.NP containing the time-history ofthe variable NP measured at all the monitor points defined in the job number 999999. The optionforce overwrites the file a1999999.NP if it already exists.

The name of the output file can be set using "output=". The following example generates a fileABC.NP from a file r1999999.dat3:

Linux:

> run9 r1file r1999999.dat3 name=NP force output=ABC

Windows:

> r1file r1999999.dat3 name=NP force output=ABC

Typing the command:

Linux:

> run9 r1file

Windows:

> r1file

lists all the options available for use with the r1file program:

Executing "r1file"FLACS r1file (version 1.0, March 1998)Copyright 1998, Christian Michelsen Research ASusage: r1file file_name [options ...]options (enter in any order after file_name):

name=string variable nameoutput=string output file nameformat=ascii/binary output formatforce force overwritetime=start,finish output time rangemonitors=+-::,*,...? monitor point list:

=+- set, add or remove:: first:last:step, positive numbers

* all (same as : or ::), separator? last character, print monitor mapexample "monitors=1:20,-3,5,7,+31,33"example "monitors=2:10:2,13,19?"


6.4 Post-processing of simulation data 201

6.4.2 r3file

The r3file program extracts ASCII data from the FLACS r3-file. It may also process the data inthe r3-file. Considering a job number 999999 the basic command is:

Linux:

> run9 r3file r3999999.dat3 name=NP force

Windows:

> r3file r3999999.dat3 name=NP force

The previous command creates an ASCII file named a3999999.NP containing the values of thevariable NP over the entire simulation domain defined in the job number 999999. The optionforce overwrites the file a3999999.NP if it already exists.

In addition to the extraction functionalities, the r3file program can process the data of the NF-DOSE variable. Assuming a job number 999999 and that the file r3999999.dat3 contains outputsof the variable NFDOSE at regular time intervals (i.e. the time intervals are given by the DTPLOTvariable in the cs999999.dat3 file) the r3file program can compute an average dose. For example,considering the following command:

Linux:

> run9 r3file r3999999.dat3 dose=2 name=FDOSE force

Windows:

> r3file1.3 r3999999.dat3 dose=2 name=FDOSE force

This will produce an average dose over dose∗dtplot seconds which in our case gives a 20 s averageperiod.

Typing the command

Linux:

> run9 r3file

Windows:

> r3file r3999999.dat3

lists all the options available for use with the r3file program:

Starting r3fileFLACS r3file (version 1.3, 2006-03-15)Copyright 2005, GexCon ASusage: r3file file_name [options ...]options (enter in any order after file_name):

name=string variable nameoutput=string output file nameformat=ascii/binary output formatforce force overwritetime=value output at timeinterpolate[=0/1] time interpolation



grid[=0/1] grid outputload=n### load rd-file.n###dump=n### dump rd-file.n###region=x,X,y,Y,z,Z region of interestgridfit[=0/1] fit region to griddose=integer steps for dosedose/time=integer steps for dose/timedtplot=value time between plotsdteps=value time proximityzero_apor=value apor<value : apor=0zero_vpor=value vpor<value : vpor=0small_por=value vpor<value : ?por=0mix=value mixture mole scaleverbose[=0/1] verbose output

6.4.3 a1file

The a1file program is used to process ASCII files with multicolumn data. Consider a job number999999 and an ASCII file a1999999.NP containing the pressure measurements at three differentmonitor points. The following command writes the time-integrals of the pressure at the threemonitor points:

Linux:

> run9 a1file a1999999.NP force integrate :1 :2 :3

Windows:

> a1file a1999999.NP force integrate :1 :2 :3

The integrals are written in the file a1999999.NP.out. Here is another example with the a1fileprogram:

Linux:

> run9 a1file a1999999.NP force slope=0.25,0.75

Windows:

> a1file a1999999.NP force slope=0.25,0.75

The previous example writes the slopes on screen (slope between 0.25 and 0.75 of maximumvalue). Finally, the last example shows how to process data by combining columns of the datafile:

Linux:

> run9 a1file a1999999.NP force integrate :a=1-2

Windows:

> a1file a1999999.NP force integrate :a=1-2

This last example, writes the time integral of the expression ’data in column 1 minus data in column2’ into the a1999999.NP.out file.

The following command:

Linux:


6.4 Post-processing of simulation data 203

> run9 a1file

Windows:

> a1file

lists all the options available for use with the a1file program:

Starting a1fileFLACS a1file (version 1.0, 2005-12-31)Copyright 2005, GexCon ASusage: a1file file_name [options ...]options (enter in any order after file_name):

force force overwriteclamp=start,finish use clampingclip=start,finish clip datascale=scale scale data:1=2-3 calculate and output:1 just outputslope=low,high calculate slopeintegrate calculate integral


Chapter 7

Best practice examples

206 Best practice examples

This chapter presents examples and best practice guidelines for FLACS. The examples includesimulations that require specialized versions of FLACS, such as FLACS-Fire or DESC.

7.1 Combined dispersion and explosion simulations withFLACS

There are at least three different ways to perform a combined dispersion and explosion simula-tion in FLACS:

• Run a dispersion simulation where ignition time and position are set before you startthe simulation. If the fuel concentration at the ignition time and position is outside theflammable region there will be no explosion. With this approach it is not possible to use theWIND condition because it enforces a fixed velocity, which is not applicable in the explo-sion.

• Run a dispersion simulation, look at the results and decide where and when to have theignition, rerun the dispersion simulation with ignition time and position set. Since youhave selected a proper ignition time and position there will be an explosion, but you havespent a lot of extra time to rerun the complete dispersion simulation. With this approach itis not possible to use the WIND condition because it enforces a fixed velocity, which is notapplicable in the explosion.

• Run a dispersion simulation, create simulation dump files at selected time instants, look atthe results and restart the simulation from the dump file with time closest to the desiredtime of ignition. This gives you the flexibility to select several ignition positions withouthaving to rerun the dispersion simulations. You can monitor the progress of the dispersionand decide to create dump files also after the simulation has been started (use the cc-file).With this approach it is possible to use the WIND condition during the dispersion simula-tion and to switch it off (change to EULER) for the explosion simulation.

7.2 Simulation Example

7.2.1 Initialization

It is recommended to start out with an empty directory for storing the files. Please note that if youhave followed the previous example in the beginning of this manual, you can skip this section.

Linux:

Make a distinct directory (DIRECTORY_NAME) in which you perform the exercise:

> mkdir DIRECTORY_NAME

Move into this directory:

> cd DIRECTORY_NAME

Copy geometry files (notice the space before the ".").

> cp /usr/local/GexCon/FLACS_v9.0/examples/ex3/*00001* .

Start up the FLACS runmanager (this assumes that you have set up an alias run9 that points to :


7.2 Simulation Example 207

> run9 runmanager

Windows:

1. Make a distinct directory in which you perform the exercise: Open the file browser ("MyDocuments") and choose File → New → Folder.

2. Copy files from C:\Program Files\GexCon\FLACS_v9.0\examples\ex3\∗00001∗(∗00001∗ means all files containing the text "00001").

3. Start the FLACS runmanager by clicking the desktop icon, or go to Start Menu → AllPro-grams → GexCon → FLACS_v9.0 → FLACS Runmanager.

7.2.2 Wind and Dispersion Simulations

In the run manager, use Add Directory to find the folder that contains the geometry files. UseRun Manager→ Tools→ CASD (or click the FLACS pre-processor icon). Open the file 200001.caj(and ignore any error messages that appear). The geometry is a representation of a full-scaleprocess module. The dimensions are 28 m × 12 m × 8 m.

7.2.2.1 Scenario

Monitor Points and Output Variables Define a regular pattern of 16 monitor points insidemodule (X=3, 9, 15, 21, Y=2, 6 and Z=2, 6).

Measure FMOLE and UVW at monitor points (use SINGLE_FIELD_SCALAR_TIME_OUTPUT).Remember to use your mouse to select all 16 monitor points.

Measure FMOLE, ER and VVEC for 3D-output (use SINGLE_FIELD_3D_OUTPUT). Rememberto hold the CTRL key while selecting multiple variables for output

Figure 7.1: Specification of Monitor Points



Simulation and Output Control Choose NPLOT=0 and DTPLOT = 2.5, CFLC = 100 (increaseddue to grid refinement, see below. For more details, see Section X.X.X in the manual) CFLV = 2.Choose total simulation time TMAX=75.

Boundary Conditions Define Wind inflow from XLO and YLO, 2m/s diagonally [use WindDirection (1,1,0)], use Wind Build-up time = 0, NOZZLE at other boundaries.

Figure 7.2: Specification of Wind boundary condition

Initial Conditions Choose initial turbulence low for stability (CHARACTERISTIC_-VELOCITY=0.1, RELATIVE_TURBULENCE_INTENSITY=0.1, TURBULENCE_LENGTH_-SCALE=0.01). Use Reference Height: 10, Surface Roughness: 0.01 (logarithmic wind profile),and Pasquill class F. Leave the other parameters unchanged.



Figure 7.3: Specification of Initial Conditions

Gas composition Natural gas (91% Methane, 7% Ethane and 2% Propane) and set ER0 = 1e30(pure gas release), Zero cloud size (no gas initially)

Leak In the leak menu, specify leak position as (X=6m, Y=5.05m, Z=2.38m). Leak direction +X(use OPEN_SIDES). Start gas dispersion at T=10 seconds, with a duration of 40 seconds. The 10second start time is chosen so that the wind field can reach steady state.

Click on OUTLET to open a new Window. Use a mass release rate of 4 kg/s through a 0.02m2leak area. Set relative turbulence intensity = 0.2, turbulence length scale = 10% of leak diameter= 0.014, Temperature = 20 °C, leave direction cosines as (0, 0, 0).

Ignition Specify an ignition time of 100 seconds (a random value after the end of the simulationso that FLACS does not try and ignite the gas cloud). Leave the ignition position unspecified.

Gas Monitor Region Define gas monitor region to cover the module (X from 0-28m, Y from0-12m and Z from 0-8m).

7.2.2.2 Grid

Simulation Volume Choose a simulation domain extending from 32 to 60 (X), -28 to 40 (Y) and0 to 32 (Z).

Grid Resolution Define a 1.333 m grid resolution in all directions (i.e. 3 grid cells for every 4m). Stretch the grid from the module to the boundaries (use CTRL and arrow keys, see SectionREF for more details if needed).



Grid Refinement The grid needs to be refined across the leak in order to make sure that it doesnot get strongly diluted. In this case, the area of the leak is 0.02 m2. According to the requirementspresented in the section Grid guidelines, the area of a control volume near the leak should belarger than 0.02 m2 but not larger than 0.04 m2. Therefore, in order to optimize the simulationtime, it is decided to refine the grid to 0.2m in Y- and Z-direction around leak. It should be keptin mind that only the grid cell containing the leak and one neighbour cell on each side needto be refined and then the grid resolution should be smoothly increased to the prevailing gridresolution (i.e. 1.333 m). This is shown in the sketch below:

Figure 7.4: Schematic of grid refinement

The following steps should then be followed:

Y-direction Since the leak position is at 5.05 m, we need to manually create grid lines 3/2∗0.2= 0.3 m from the leak position in both directions, i.e. 4.75 and 5.35 m. Change the grid directionto Y and use Grid → Add to add these two grid lines. Select these two grid lines using CTRLand arrow keys (a message in CASD in the yellow box below the geometry can confirm whatgrid lines are selected). After selecting these two grid lines, use Grid → Region and enter 3 tocreate three grid cells of resolution 0.2 m. The next step is to smooth the grid between the twosignificantly different grid resolutions. For this purpose, we select the grid lines between -1.333 mand 4.95 m. In this way, one fine grid cell is selected and selecting a grid cell 4-5 away provides thepossibility to achieve a reasonably gradual transition (the goal is that the grid resolution shouldnot be changing by more than 30-40 % from one grid cell to the next). This can be confirmed byusing Grid → Information. Another limitation in this case that we need to keep the grid line at0 m (and 12 m) intact and therefore, we cannot select a grid cell beyond -1.333 m (and 13.333 m).After these two grid lines have been selected, use Grid→ Smooth. Repeat on the other side whenthe grid lines between 5.15 m and 13.333 m are selected. Check Grid → Information to ensurethat the Max. percentage difference factor is acceptable.



Figure 7.5: Grid in Y-direction

Z-direction By a similar logic as above, we need to manually create grid lines at 2.08 and 2.68m. Change the grid direction to Y and use Grid → Add to add these two grid lines. Select thesetwo grid lines using CTRL and arrow keys and use Grid → Region and enter 3 to create threegrid cells of resolution 0.2 m. For smoothing in the +Z direction, select the grid lines between2.48 and 9.333 m and use Grid → Smooth. In the Z direction, it is not possible to use the smoothcommand as there is only one large grid cell. In this case, select the grid from 0 m to 2.28 m anduse Grid → Stretch → Negative. Check Grid → Information to ensure that the Max. percentagedifference factor is acceptable.

The last step before starting the simulation is to make a cc-file. In the FLACS runmanager, clickon the dispersion job, click parameters, and edit cc-file and type the following (use capital lettersand extra line shift at the end of cc-file):

NDUMP 1TDUMP 40NDUMP 2TDUMP 55

This gives 2 dumps at 40 and 55 seconds which can be used to restart the calculations.



Figure 7.6: Defining the cc file for dispersion simulation

7.2.2.3 Simulation

Start the simulation by clicking on the job and clicking simulate in the runmanager.

7.2.2.4 Results

The most important result from this simulation is the gas cloud distribution. This can be studiedin Flowvis based on the material presented in sections Flowvis examples and Introductory exam-ple. 2D pictures of the gas cloud at times 40 s and 55 s are shown below. The concentrations areplotted in the flammable range for natural gas (i.e. between 5 % and 15 % natural gas).



Figure 7.7: Gas cloud distribution in the flammable range at time 40 seconds

Figure 7.8: Gas cloud distribution in the flammable range at time 55 seconds

7.2.3 Explosion Simulations

In order to carry out explosion simulations (ignition of a realistic gas cloud produced as a resultof the release simulated above), copy files to new job (Use CASD to save as a new job number).We have dumped the dispersion results at two discrete times: 40 s and 55 s. The gas cloud at 40s is used to start an explosion job.



7.2.3.1 Scenario and Grid

Use flowvis to find a suitable ignition position e.g. make contour (2D cut plane) of FMOLE orER in plane of leak (2.4 m) to find regions where concentration is close to stoichiometric (that isexpected to lead to Worst-case explosion pressure). In this case, based on the figure above, anignition position of (23, 4.5, 2.4) was chosen. Certain changes need to be made to the scenarioand the grid to make the job suitable for explosion simulations. The following steps should befollowed:

1. Change grid to uniform explosion grid e.g. 0.5 m (stretch from module to boundaries)2. Simulation and Output Control: Change TMAX to -1, NPLOT = 50, CFLC = 5, CFLV = 0.5.

Change DTPLOT back to -1.3. Change output variables (both scalar time and 3D): Use P and PROD (it is possible to re-

move the other variables)4. Change ignition position (use step 2 above) and ignition time (40.05)5. Save and calculate porosities

After that, the user should click on the job in the runmanager (if it is not visible, click on rescandirectory). Click Parameters and define a cc-file. The cc-file should contain only one line (theuser should remember to include an extra line in the end):

NLOAD = 1

The last step is to generate a new rd file for restarting calculation based on the explosion job. Thistransfers all required information from the dispersion grid to the explosion grid.

Linux: In terminal window, type:

> run9 rdfile rdXXXXXX.n001 rdZZZZZZ.n001

where XXXXXX is the dispersion job and ZZZZZZ is the explosion job. The user should makesure that you are in the correct directory.

Windows: In command window, type:

> rdfile rdXXXXXX.n001 rdZZZZZZ.n001

where XXXXXX is the dispersion job and ZZZZZZ is the explosion job. The user should makesure that you are in the correct directory.

7.2.4 Simulation

Start the simulation by clicking on the job and clicking simulate in the runmanager.

7.2.5 Results

It is possible to generate results using Flowvis. The scalar time plot of pressure at all monitorpoints is shown below. It can be seen that the maximum pressure of 1.3 barg occurs at t = 40.34second at monitor point 1.


7.3 Equivalent Stoichiometric Gas Cloud 215

Figure 7.9: Overpressures at monitor points as a result of the explosion of the dispersed gas cloudat 40 seconds

7.3 Equivalent Stoichiometric Gas Cloud

7.3.1 General Principles

For a dispersion study following a leak in a process area, the main evaluation parameter is thesize of the flammable gas cloud. In order to evaluate the hazard of a given gas cloud, we havedeveloped methods used for natural gas in the oil and gas industry that aim at estimating anequivalent stoichiometric gas cloud with comparable explosion consequences. These methodshave been developed in order to reduce the number of simulations that need to be carried outin order to do a thorough risk evaluation. The size of the equivalent stoichiometric cloud at thetime of ignition is calculated as the amount of gas in the flammable range, weighted by the con-centration dependency of the flame speed and expansion. For a scenario of high confinement,or a scenario where very high flame speeds (faster than speed of sound in cold air) are expected(either large clouds or very congested situations), only expansion based weighting is used (de-noted as Q8). For most situations lower flame speeds are expected and the conservatism can bereduced. Here a weighting of reactivity and expansion is used (denoted as Q9). The Q8 and Q9equivalent volumes can be defined as:

Q8 = ∑ V × E/Estoich (7.1)

andQ9 = ∑ V × BV × E/(BV × E)stoich (7.2)

Here, V is the flammable volume, BV is the laminar burning velocity (corrected for flame wrin-kling/Lewis number effects), E is volume expansion caused by burning at constant pressure inair, and the summation is over all control volumes. Thus, Q9 cloud is a scaling of the non-homogeneous gas cloud to a smaller stoichiometric gas cloud that is expected to give similarexplosion loads as the original cloud (provided conservative shape and position of cloud, andconservative ignition point). This concept is useful for QRA studies with many simulations, andhas been found to work reasonably well for safety studies involving natural gas releases (NOR-SOK, 2001). Figure 1 presents the pressure as a function of equivalent stoichiometric volume Q9for several full-scale experiments carried out by HSE as a part of the Phase 3B project. It can beseen that the pressure values correlate with the equivalent gas cloud sizes to a reasonable degree.



Figure 7.10: Overpressures predicted by FLACS as a function of equivalent cloud size comparedwith HSE large scale experiments (Phase 3B project).

This concept has also been applied to hydrogen systems for the FZK workshop experiments andhas been found to give reasonably good predictions. Figure 2 shows the explosion pressure plot-ted as a function of the size of the equivalent gas cloud, along with reference values calculatedby using a stoichiometric gas cloud of the same volume. This figure shows that this approachgives a very good indication of expected overpressures (without initial turbulence, but with con-servative cloud position and shape, and ignition location). Comparison with experimental datarevealed that in general, the simulations were able to predict very representative values of over-pressures for both geometries and all release scenarios. More details can be found in Middha, etal., 2007.


7.3 Equivalent Stoichiometric Gas Cloud 217

Figure 7.11: Overpressures predicted by FLACS in the hood (red) and plate (blue) configurationsfor “workshop” experiments carried out by FZK as function of estimated Q9 equivalent cloudsize for ignited leaks. Reference calculations with homogenous stoichiometric gas clouds areincluded.

7.3.2 Shape of Equivalent Gas Cloud

As a practical guideline, the user is recommended to choose the shape of the cloud that will givemaximum travel distance from ignition to end of cloud for smaller clouds. For larger clouds,end ignition scenarios with longer flame travel should also be investigated. The cloud shouldbe made as a cubical rectangular box with assumed “planes of symmetry” towards confinement.The aspect ratio for a free cloud should be 1:1:1, for a cloud towards the ceiling 2:2:1, towardsceiling with one sidewall 2:1:1, etc. For a free jet in a less confined situation, the jet momentumwill usually dominate the mixing of the jet until the hydrogen concentration has become leanunless the wind is very strong. The cloud should be assumed to be located a small distancedownwind of the jet, if possible conservatively towards obstructions/walls. Alternatively, fora low momentum release in a confined situation with strong stratification, the cloud should beassumed to cover full ceiling area (or area between beams). For highly buoyant gases such ashydrogen, it can always be assumed that a confined and semi-confined cloud is located near theceiling or below any other horizontal confinement (a possible deviation from this may be largeliquid hydrogen releases in hot and dry surroundings).

7.3.3 Ignition of Equivalent Gas Cloud

For smaller clouds (flame travel less than 1m to open boundary), ignition with maximum distanceto edges of gas cloud is normally a conservative choice (for a cloud located in a corner, this meanscorner ignition, not central ignition). This could be used to represent all scenarios. Alternativelya distribution of ignition points could be applied. For larger clouds, a homogenous distributionof ignition positions should be applied. It should be kept in mind that the gas clouds with thepossibility of the longest flame travel are often the most dangerous ones.

For a quantitative explosion risk assessment the explosion simulations should be performed with



various idealized clouds of variable size and typically using stoichiometric concentration. For thepurpose of QRA, the distribution of ignition locations should be chosen to represent reality. Ifthere are one or more highly likely ignition locations that dominate the ignition frequencies, thesemay be used. Otherwise, it should be assumed that a constant ignition source might lead to endignition (where concentration reaches LFL), whereas intermittent ignition sources will be morearbitrarily distributed (with higher likelihood centrally in the cloud where concentrations areabove LFL). For stratified clouds, end ignition will mean ignition in the lower end of the cloud.

Two ignition probabilities should generally be established (probability for spontaneous ignitionshould also be considered in case of hydrogen):

• PIconst = ignition probability from constant ignition sources (per m3 that is exposed toflammable gas for the first time in the last 1 second)

• PIintermittent = ignition probability for intermittent ignition sources (per m3 flammable vol-ume and second)

From the CFD calculations the volume of the gas cloud with concentration between LFL andUFL will give the volume that may be ignited if exposed to an intermittent ignition source. Thisinformation should thus be related to the intermittent ignition frequencies defined. In the caseof constant ignition source, the Q6 output from FLACS (or similar from another CFD tool) givesthe cloud volume that was exposed to flammable gas concentrations for the first time last sec-ond. This information should be combined with the probabilities for ignition by constant ignitionsources. Thus for each time step (or each 1s) for each dispersion calculation, the probability forignition from spontaneous, intermittent and constant ignition sources should be established, andthis probability should be added to a gas cloud size class (based on corresponding Q8, Q9 or acombination).

If gas cloud becomes rich (in a well mixed state, but more gas available than needed to fill roomwith stoichiometric concentration) this may be represented either as a stoichiometric cloud oras a rich cloud with a slightly higher reactivity than observed. This can be done by using theflammable volume to establish the volume of the cloud and the flammable mass to establish itsconcentration.

7.3.4 Further Guidelines

As described above, for scenarios (cloud/congestion combinations) with low or moderate con-gestion (or cloud sizes much smaller than room dimension, e.g. < 10%) where very high flamespeeds can be achieved, one should use Q8 value instead of Q9. To evaluate this, a number ofcloud sizes can be simulated to identify a critical cloud size, Qcrit, for which flame speeds ex-ceed e.g. 200 m/s. For clouds with Q8 < Qcrit, the weighting procedure above can be applied,for clouds with Q8 > Qcrit, one should apply the Q8 cloud as representative cloud size. Forvented rooms and other situations with a significant confinement, a weighting between Q8 andQ9 volumes is suggested.

If the ventilation or a high-momentum leak (jet) creates a significant turbulence in the regionwhere ignition is expected, this turbulence should be defined as an initial condition for the CFDsolver. The main parameter that should be defined should be turbulent fluctuations (productof characteristic velocity and turbulence intensity). The turbulent length scale should not beassigned a too large value (As a guideline, maximum 10% of smallest cloud dimension and max-imum 50% of grid cell size should be used).


7.4 Dispersion simulation with wind 219

7.4 Dispersion simulation with wind

FLACS–Dispersion is a CFD-code for predicting the spread of flammable or toxic gases in com-plex geometries or terrain, including density effects. GexCon released the first FLACS versionwith dispersion capabilities in 1989, and since 2001 it has been possible to purchase dispersioncapabilities. All functionality of FLACS-DISPERSION is also found in the full FLACS version.

7.5 Hydrogen explosions and DDT

FLACS-Hydrogen is a CFD-code for predicting dispersion and explosion scenarios with hydro-gen gas in complex geometries. Hydrogen has been available as a gas in FLACS since 1989,and could be purchased as a dedicated tool since 2001. The validity was strongly improved withFLACS 8.1 in 2005. All functionality of FLACS-Hydrogen can be found in the full FLACS version.

Important changes are carried out for hydrogen combustion properties:

• Modified laminar burning velocity curve, LFL has been reduced to 4%• Adjustment to include effect of Lewis number (more than normal wrinkling on lean con-

centrations, less than normal wrinkling on rich concentrations)• Stronger enhancement due to flame wrinkling (3.5 times increase in burning velocity with

distance from ignition point at stoichiometric concentration).• Normal time step guidelines should be applied, i.e. CLFC=5 and CFLV=0.5

These changes are resulting from the recent active work on validation for hydrogen safety, bothwithin the activity supported by Norsk Hydro, Statoil and IHI, but also through work withinHYSAFE network. New experiments have become available. Much of the work behind is ofconfidential nature, and it is not clear that we can share the validation data with all FLACS users.However an extensive validation matrix has been simulated and we are quite satisfied with theresults simulating:

• 20m diameter hemispherical hydrogen cloud (large-scale laminar flames)• 1.4m small-scale channel (different baffles, gas concentration, ignition location)• 3D corner small-scale pipe arrays (different congestion, gas concentration, ignition)• 30m tunnel FLAME facility (different venting, baffles, gas concentration)• And more ...

Based on the results we consider the FLACS V8.1 performance simulating hydrogen comparableto what is generally seen when simulating other gases.

The laminar diffusivity for hydrogen in FLACS may historically be somewhat too high. Thiswill only be important in situations with absolutely no turbulence, i.e. mainly in closed vesselswith no ventilation and weak temperature gradients. When discovering the possible error, theconstant was not changed in FLACS. Instead we included a possibility for the user to define theconstant for dynamic laminar viscosity manually through setup-file or KEYS-string in scenario-file.

The default constant for dynamic laminar viscosity in FLACS is 2.0e-5 (see below), a more correctvalue is probably 0.6e-5.

VERSION 1.1$SETUP

KEYS="AMUL=Y:2.0e-5"$END



7.5.1 Use normal time step for hydrogen explosion simulations

One important observation done is that you should not change CFLC-number (time step crite-ria) when simulating deflagrations (explosions) involving hydrogen. We previously (last year)recommended using a shorter time step for hydrogen simulations (CFLC=0.5), and we also havebeen recommending this for far field blast calculations (CFLC=0.5). We have for a long time beenaware that shorter time-step in some situations will give different explosion development. Nowwe have seen that results in hydrogen simulations become better keeping the normal time step.For pressure wave propagation a distance away from the calculation there is a possibility to setan option in the scenario-file that keeps a short time step when blast wave propagates away fromthe explosion (STEP=KEEP_LOW). For multi-block simulations with BLAST solver you may stillneed to use a short time step for stability reasons. Note that "STRICT:" is assumed as defaulttime stepping mode in the new version of Flacs. You may enter "NOSTRICT:" as your choice,but this is normally not recommended.

For dispersion calculations we still recommend to use long time steps (CFLC=20, CFLV=2). Withlocal grid refinement (for instance near leak), we also recommend to increase CFLC proportionalto the grid refinement factor: CFLC=20 x CVnormal / CVfine. If stability problems occur (thismay well happen), it is recommended to reduce the CFLC and CFLV numbers by a factor in therange of 2 - 4.

For explosive and vessel burst calculations it is recommended to use short time steps. The modelsfor handling explosives are not included in the standard version. It is recommended to set CLFCand CFLV=0.025 when using BLAST solver, and slightly larger values could be used with FLACSsolver. Vessel burst calculations can be done as a normal FLACS calculation, then CLFC andCFLV around 0.1 is recommended.


Chapter 8

Technical Reference

222 Technical Reference

This chapter contains an overview of the theoretical foundation for the FLACS software, includ-ing physical and chemical models.

8.1 Definitions and gas thermodynamics

This section presents definitions of gas and mixture parameters and relations for ideal gases.

8.1.1 Definitions

Number of moles of species:

ni =miMi

(8.1)

Mole fractions:Xi ≡

ni

∑Ni=1 ni

(8.2)

Mass fractions:Yi ≡

mi

∑Ni=1 mi

(8.3)

Fuel-oxidant ratio:(F/O) ≡

mfuelmox

(8.4)

Equivalence ratio:

Φ ≡ (F/O)(F/O)stoich

(8.5)

Mixtures of fuel and oxidant are characterised by the equivalence ratio as follows:

Φ > 1 : Fuel rich mixture

Φ = 1 : Stoichiometric mixture

Φ < 1 : Fuel lean mixture

The mixture fraction, ξ, describes the degree of mixing between two well-defined states (0 and 1)and is defined as follows:

ξ =φ− φ0

φ1 − φ0 (8.6)

where φ is a general variable.

The progress variable χ tells how much of the potential fuel that has burnt and is defined asfollows:

χ =Yfuel

Y0fuel + ξ

(Y1

fuel −Y0fuel

) (8.7)

8.1.2 Mixing of several gases

Mole fraction:Xi ≡

Yi/Mi

∑Ni=1 Yi/Mi

(8.8)

Mass fraction:Yi ≡

Xi Mi

∑Ni=1 Xi Mi

(8.9)


8.2 Stoichiometric reaction 223

8.1.3 Ideal gas relations

Ideal gas law for a mixture:p = ρRT (8.10)

Ideal gas law for single specie:pi = ρiRiT (8.11)

Dalton’s law for a perfect gas:

p =N

∑i=1

pi =RuT

V

N

∑i=1

ni (8.12)

Isentropic ratio:

γ =cp

cv(8.13)

Speed of sound:

c ≡√

γRT =√

γpρ

(8.14)

Mach number:Ma ≡ u

c(8.15)

Pressure-density-temperature:(pp0

)=(

ρ

ρ0

)γ

=(

YY0

)γ/(γ−1)=(

1 +γ− 1

2Ma2

)−γ/(γ−1)(8.16)

8.2 Stoichiometric reaction

Combustion is oxidation of a fuel accompanied by the production of heat and light. In mostburning processes, air is oxidant. A simple main reaction can be written as:

CncHnhOno + aO2 → ncCO2 + bH2O + Q (8.17)

This reaction is stoichiometric because there is neither fuel nor oxidant left after the reaction iscompleted. The stoichiometric amount of oxidant on mole basis can be calculated by:

a = nc +nh4− no

2(8.18)

The combustion products produced in the reaction are water vapour (H2O) and carbon dioxide(CO2).

Some relations for mixing fuel with air are listed below. The mole fraction of O2 in air is set to20.95%, which corresponds to a mass fraction of 23.2%. This is the normal air composition inFLACS, see Initial conditions .

Stoichiometric oxidant-fuel ratio on mass basis:

rox = aMO2

Mfuel(8.19)

Stoichiometric air-fuel ratio on mass basis:

rair =(

1 +YN2

YO2

)rox = 4.31rox (8.20)



Mass fraction of fuel given an equivalence ratio Φ:

Yfuel =Φ

Φ + rair(8.21)

8.3 Governing equations for fluid flow

This section describes the mathematical model for compressible fluid flow used in FLACS.

Conservation of mass:∂

∂t(βvρ) +

∂

∂xj

(β jρuj

)=

mV

(8.22)

Momentum equation:

∂

∂t(βvρui) +

∂

∂xj

(β jρuiuj

)= −βv

∂p∂xi

+∂

∂xj

(β jσij

)+ Fo,i + βvFw,i + βv (ρ− ρ0) gi, (8.23)

where Fw,i is flow resistance due to walls and Fo,i is flow resistance due to sub-grid obstructions:

Fo,i = −ρ

∣∣∣∣ ∂β

∂xi

∣∣∣∣ ui |ui| (8.24)

σij is the Stress tensor.

Transport equation for enthalpy:

∂

∂t(βvρh) +

∂

∂xj

(β jρujh

)=

∂

∂xj

(β j

µeffσh

∂h∂xj

)+ βv

DpDt

+QV

(8.25)

Transport equation for fuel mass fraction:

∂

∂t(βvρYfuel) +

∂

∂xj

(β jρujYfuel

)=

∂

∂xj

(β j

µeffσfuel

∂Yfuel∂xj

)+ Rfuel (8.26)

where Rfuel is the fuel reaction rate, which will be handled in combustion modelling.

Transport equation for the mixture fraction:

∂

∂t(βvρξ) +

∂

∂xj

(β jρujξ

)=

∂

∂xj

(β j

µeffσξ

∂ξ

∂xj

)(8.27)

Transport equation for turbulent kinetic energy:

∂

∂t(βvρk) +

∂

∂xj

(β jρujk

)=

∂

∂xj

(β j

µeffσk

∂k∂xj

)+ βvPk − βvρε (8.28)

Transport equation for the dissipation rate of turbulent kinetic energy:

∂

∂t(βvρε) +

∂

∂xj

(β jρujε

)=

∂

∂xj

(β j

µeffσε

∂ε

∂xj

)+ βvPε − C2βvρ

ε2

k(8.29)


8.3 Governing equations for fluid flow 225

The stress tensor in the above equations is given by:

σij = µeff

(∂ui∂xj

+∂uj

∂xi

)− 2

3δij

(ρk + µeff

∂uk∂xk

)(8.30)

The effective viscosity is defined as follows:

µeff = µ + ρCµk2

ε, (8.31)

where the second term is known as the turbulent viscosity or eddy viscosity.

Flow shear stresses, Gs, wall shear stresses, Gw, buoyancy, Gb, and sub-grid objects, Go contributeto the production of turbulent kinetic energy Pk:

Pk = Gs + Gw + Gb + Go (8.32)

The production rate of turbulent kinetic energy due to shear stresses appears from the derivationof the transport equation and reads:

Gs = σij∂ui∂xj

(8.33)

Production due to buoyant forces modelled by a simple gradient model:

Gb = −1ρ

µeffσb

gi∂ρ

∂xi(8.34)

The turbulence generation due to sub-grid obstructions is modelled by:

Go = Coβvρ |~u| u2i fi (8.35)

where Co is a model constant and fi is a parameter depending on sub-grid objects.

The production of dissipation, Pε, is modelled as follows:

Pε = C1εkε

Pk

(1 + C3εR f

)(8.36)

where the model for the buoyancy term follows Rodi (1980):

R f = −GbPk

|~u×~g||~u| |~g| (8.37)

In FLACS, the buoyancy terms Gb and R f are zero when products are present.

8.3.1 Turbulence model

Turbulence is modelled by a two-equation model, the k− ε model. It is an eddy viscosity modelthat solves two additional transport equations; one for turbulent kinetic energy and one for dis-sipation of turbulent kinetic energy. Following Boussinesq eddy viscosity assumption, an eddyviscosity models the Reynolds stress tensor as follows:

− ρu′′i u′′j = µeff

(∂ui∂xj

+∂uj

∂xi

)− ρ

23

kδij (8.38)



A few constants are included in the equations mentioned above. In FLACS, the following set ofconstants is used, which agree with the model of Launder and Spalding (1974):

Cµ C1ε C2ε C3ε

0.09 1.44 1.92 0.8 (8.39)

In addition, there is a set of turbulent Prandtl-Schmidt numbers, σφ. Prandtl-Schmidt numberstells about the diffusion of the variable in question compared to the dynamic viscosity. Theturbulent Prandtl-Schmidt numbers are:

σh σfuel σξ σk σε σb0.7 0.7 0.7 1.0 1.3 0.9

(8.40)

8.4 Wall functions

Boundary layers are regions in the flow field close to walls and obstructions where there aresteep gradients and peak values for turbulent kinetic energy and its dissipation rate. Very closeto the wall surface dominates viscous forces over inertial effects. The motivation for using wall-functions is to model the influence of the wall at a point a certain distance from the wall.

A dimensionless wall distance is defined by:

y+ =ρC1/4

µ k1/2yµ

, (8.41)

where y is the distance from the wall point to the wall. Wall point is defined as the point closest tothe wall where transport equations are solved. The shear stresses caused by the wall are modelledby:

τw,i =

µ ui

y if y+ < E+

ρuiκC1/4µ k1/2

κE++ln(

y+

E+

) if y+ ≥ E+ (8.42)

Then wall friction term in the momentum equation becomes:

Fw,i = −βvτw,iAw

V(8.43)

Production of turbulent kinetic energy in the wall point is modelled by:

Gw =

0 if y+ < E+

2τ2w |~u|κC1/4

µ k1/2

κE++ln(

y+

E+

) if y+ ≥ E+ (8.44)

Dissipation of turbulent kinetic energy is given a value at the wall point by solving the followingintegral:

εw =1

ycv

∫ ycv

0εdy (8.45)

The integral is estimated by:

εw =

1

ycv

(2µkρy + ε(ycv − y)

)if y+ < E+

1ycv

(2µ[

ky+

ρyE+ −(

y+

E+ − 1) (

k+1−kρ(y+1−y)

)]+ C3/4

µ k3/2

κ ln(

y+

E+

)+ ε(ycv − y)

)if y+ ≥ E+

(8.46)


8.5 Wind boundary 227

k+1 denotes the value of k and y+1 denotes the wall distance in the point beyond the wall pointin the opposite direction of the wall. ε denotes the mean value of ε between the cell point and thecontrol volume boundary in the opposite direction of the wall.

8.5 Wind boundary

Wind boundaries reproduces the properties of the atmospheric boundary layer close to Earth’ssurface. Monin and Obukhov (1954) developed a theory to explain buoyancy effects on the at-mospheric boundary layer and defined a charcteristic lenght scale:

L = −ρacpTau∗3

κgHs(8.47)

where Hs is the sensible heat flux from the surface and u∗ is the friction velocity. The Monin-Obukov length is a measure for the stability of the atmospheric boundary layer. Table Monin-Obukhov lengths and stability shows an interprentation of the Monin-Obukhov lenths with re-spect to the atmospheric stability. (Bosch and Veterings, 1996)

Monin-Obukhov length StabilitySmall negative, −100m < L0 Very unstableLarge negative, −105 < L < −100 Unstable /tr>Very large, |L| > 105 NeutralLarge positive, 10 < L < 105 StableSmall positive, 0 < L < 10 Very stable

Table 8.1: Monin-Obukhov lengths and stability.

In FLACS, the Monin-Obukhov length is estimated by using Pasquill classes, which is a methodof categorizing the amount of atmospheric turbulence present. The user have to give averagewind velocity, U0, a reference height, zre f , an atmospheric roughness length, z0 and the Pasquillclass under Initial conditions. The velocity profile is logarithmic:

U(z) =

{u∗κ ln

(z+z0

z0

)if z0 > 0

U0 if z0 = 0(8.48)

where, u∗, the friction velocity. u∗ is generally given by:

u∗ =

U0κ

ln( zre f

z0

)+ 5

L (zre f−z0)if L > 0

U0κ

ln( zre f

z0

)−ψ1−ψ2

if L < 0(8.49)

Attention:

At present two friction velocities are used in FLACS. The expression in the friction velcoc-ity equation above is used for k and ε. For velocity is the following friction velocity used:u∗ = U0κ/ ln(zre f /z0). This is due to the development history of FLACS, where a logritmicvelocity profile was implemented before the profiles for the turbulence parameters.

Table Wind profile parameters below gives an overview of parameters that are used to calculatevalues for velocity, k, and ε at wind boundaries. ψ1 and ψ2 in the equation for the friction velocity



are given by:

ψ1 = 2 ln(

1 + ψ3

2

)+

12

ln(

1 + ψ23

)− 2 arctan (ψ3) +

π

2(8.50)

ψ2 = 2 ln(

1 + ψ4

2

)+

12

ln(

1 + ψ24

)− 2 arctan (ψ4) +

π

2(8.51)

ψ3 =(

1−16zre f

L

)1/4

(8.52)

ψ4 =(

1− 16z0

L

)1/4(8.53)

Pasquill class Stability Boundary layerheight, h

Ls zs

A Unstable 1500 m 33.162 m 1117 mB Unstable 1500 m 33.258 m 11.46 mC Slightly

unstable1000 m 51.787 m 1.324 m

D Neutral 0.3u∗ Lf 1.0 m 0 m

E Slightly stable 0.4√

u∗Lf -48.33 m 0 1.262 m

F Stable 0.4√

u∗Lf -31.323 m 0 19.36 m

Table 8.2: Wind profile parameters. Values for Ls and zs are takenfrom Bosch and Wetering (1996) and orginates from the graphs ofGolder (1972). Values for h are taken from Han et al. (2000).

From the values in the wind profile parameters table, the Monin-Obukhov length can be calcu-lated as follows (Golder, 1972):

1L

=1Ls

logz0

zs(8.54)

The set of expressions for the wind boundary profiles for turbulent kinetic energy, k, and itsdissipation, ε proposed by Han et al. (2000) are implemented in FLACS. Different experssionswere proposed for for unstable and stable/neutral boundary layers. Unstable boundary layersare caused by heat from the ground that increases the temperature of the air close to the surface.Hence, the density close to the surface is less than the density of the air above, which gives anunstable situation. The mean surface heat flux, qs, is therefore an important parameter when theturbulence profiles at the inlet is estimated for unstable boundary layers. The inlet profiles forustable boundary layers (A, B, and C) are:

k(z) =

{0.36w∗2 + 0.85u∗2 (1− 3 z

L)2/3 if z ≤ 0.1h(

0.36 + 0.9( z

h)2/3 (1− 0.8 z

h)2)

w∗2 if z > 0.1h(8.55)

and

ε(z) =

u∗3

κz

(1 + 0.5

∣∣ zL

∣∣2/3)3/2

if z ≤ 0.1hw∗3

h(0.8− 0.3 z

h)

if z > 0.1h(8.56)

where the heat velocity , w∗, is given by:

w∗ =(

gqshT0ρcp

)1/3(8.57)


8.6 Combustion modelling 229

where the air properties, ρ, and cp are obtained at ambient temperature T0 and pressure p0.Profiles for neutral and stable boundary layers depend on the friction velocity and the Monin-Obukhov length as follows:

k(z) =

{6u∗2 if z ≤ 0.1h6u∗2 (1− z

h)1.75 if z > 0.1h

(8.58)

and

ε(z) =

{u∗3

κz(1.24 + 4.3 z

L)

if z ≤ 0.1hu∗3

κz(1.24 + 4.3 z

L) (

1− 0.85 zh)3/2 if z > 0.1h

(8.59)

8.6 Combustion modelling

Ignition of a premixed cloud of fuel and oxidant may escalate to an explosion. Before escalation,a steady non-turbulent premix of fuel and oxidant will burn with a laminar burning velocity:

S0L = S0

L(fuel, Φ) (8.60)

The laminar burning velocity depends on the fuel and the equivalence ratio Φ. For mixtures withfuel contents below the Lower Flammability Limit (LFL) or above the Upper Flammability Limit(UFL), the laminar burning velocity equals zero, i.e. it will not burn. In an explosion, the flamewill accelerate and become turbulent. The turbulent burning velocity is much larger than thelaminar one due to much better mixing of reactants and products. FLACS uses correlations forboth laminar and turbulent burning velocities that origin from experimental work.

In industrial applications, the reaction zone in a premixed flame is thin compared to practical gridresolutions. It is therefore necessary to model the flame. In FLACS, the flame zone is thickened byincreasing the diffusion with a factor β and reducing the reaction rate with a factor 1/β. Hence,the flame model in FLACS is called the β-model.

8.6.1 The FLACS flame model

The diffusion coefficient D for fuel comes from the transport equation for fuel:

D =µeffσfuel

(8.61)

Furthermore, it is possible to define a dimensionless reaction rate W. In the β-model, D and Ware adjusted as follows:

W∗ = Wβ = W

lLT∆g

(8.62)

D∗ = Dβ = D∆g

lLT(8.63)

From an eigenvalue analysis of the burning velocity (Arntzen, 1998), the following relation be-tween the diffusion coefficient D and a dimensionless reaction rate W is obtained for χq = 0.05:

WD = 1.37S2u = W∗D∗ (8.64)



χqis the quenching limit of the progress variable χ. D∗ and W∗ depend on the grid size and theburning velocity as follows:

W∗ = c1βSu

∆g(8.65)

D∗ = c1βSu∆gDβ (8.66)

The reaction rate of fuel is modelled by the following expression:

Rfuel = −W∗ρ min(δH(χ− χq), χ, 9− 9χ

)(8.67)

where δH is the Heaviside step function.

8.6.2 Burning velocity model

The burning velocity is laminar when the flame is smooth and governed by molecular diffu-sion. This is typically the case in the very early phase of an explosion (e.g. spark ignition ofa combustible cloud under quiescent conditions). A short period of time after the ignition, theflame becomes quasi-laminar when instabilities lead to wrinkling of the flame. After a transitionperiod, the flame reaches the turbulent regime.

The laminar burning-velocity depends on the type of fuel, fuel-air mixture and pressure. For eachfuel, the laminar burning velocities at different equivalence ratios are tabulated. The laminarburning-velocity of a mixture of fuels is estimated by taking the volume-weighted average. Thepressure dependency on the laminar burning velocity is described as:

SL = S0L

(PP0

)γP

, (8.68)

where γP is a fuel dependent parameter. In the quasi-laminar regime, the turbulent burningvelocity is given by:

SLQ = SL + 8S0.284L u′0.912 l0.196

I (8.69)

The correlation for the turbulent burning velocity ST is a simplification of a general expressionpresented by Bray (2000) and reads:

ST = 15S0.784L u′0.412 l0.196

I (8.70)

FLACS selects burning velocity as follows:

Su = max(SL min

(SQL, ST

))(8.71)

8.7 Modelling of jet sources

To model the conditions of a pressurised reservoir which is gradually emptied through a nozzle,a simple procedure may be utilised which calculates the sonic flow rate through the nozzle. Byassuming isenthalpic expansion, it is possible to calculate expansion of a sonic flow analytically.Further air entrainment can be accounted for using simplifying assumptions. Finally, the calcu-lated mass flow mixes in a well-stirred reactor with a constant volume and a constant ventilationrate. Additional values for the turbulence quantities must be calculated in order to use the datain FLACS.

Calculation of 5 stage analytic dispersion:


8.7 Modelling of jet sources 231

1. Reservoir (stagnation)2. Nozzle (sonic)3. Jet (outlet)4. Air entrainment5. Well-stirred reactor

Let Θ(u, h, T, p, ρ, A) be a vector describing the necessary leakage parameters, then

• Θa refers to the ambient condition,• Θ2 refers to the outlet condition,• Θ1 refers to the nozzle condition and• Θ0 refers to the stagnation condition.

Initial reservoir conditions:

Pressure: p0 is specified.Temperature: T0 is specifiedVolume: V0 is specifiedDensity: ρ0 = p0

RT0Total mass: m0 = ρ0V0Heat exchange coefficient: hwall is =specified.

Reservoir conditions at time t + dt:

Total mass: m0,t+dt = m0 − m1dtTemperature: T0,t+dt = T0 − (Q0 + m1h1)dtWall heat flux: Q = hwall(T0 − Twall)Density: ρ0 = m0/V0Pressure: p0 = ρ0RT0

Nozzle (sonic) conditions:

Effective nozzle area: A1 is specified.Temperature: T1 = T0(2/(γ + 1))Pressure: p1 = p0(T1/T0)γ/(γ−1)

Density: ρ1 = p1RT1

Sound speed: c1 =√

γRT1Velocity: u1 = c1Enthalpy: h1 = cpT1Mass flow: m1 = ρ1u1 A1

Jet (outlet) conditions:

Velocity: u2 = u1 − p1−p2ρ1u1

Enthalpy: h2 = h1 + 12 (u2

1 − u22)

Temperature: T2 = T1 + 12

u21−u2

2cp

Pressure: p2 = paDensity: ρ2 = p2

RT2

Effective outlet area: A2 = A1ρ1u1ρ2u2

Mass flow: m1 = ρ1u1 A1



Air entrainment condition:

Pressure: p3 = paTemperature: T3 = TaDensity: ρ3 = p3

RT3

Velocity: u3 = u2f2f3

Effective area: A3 = A2ρ2u2ρ3u3

Mass flow: m3 = ρ3u3 A3

Well-stirred reactor condition:

Volume: V4 is specified.Ventilation rate: Vair is specified.Incremented fuel mass: mfuel = mfuel + mfueld fIncremented air mass: mair = mair + maird fIncremented mixture mass: mmix = mfuel + mairDensity: ρ4 = mmix

V4+(Vfuel+Vair)dtTotal mass inside volume: m4 = V4ρ4Fuel mass: mfuel,t+dt = m4

mfuelmtotal

Air mass: mair,t+dt = m4mair

mtotal

8.8 Numerical Schemes

FLACS is a computational fluid dynamics (CFD) code solving the compressible conservationequations on a 3D Cartesian grid using a finite volume method. The conservation equationsfor mass, momentum, enthalpy, and mass fraction of species, closed by the ideal gas law, areincluded. The conservation equations can be represented in general as:

∂

∂t(ρφ) +

∂

∂xj(ρuiφ)− ∂

∂xj

(ρΓφ

∂

∂xj(φ)

)= Sφ (8.72)

The in-house development started around 1980, primarily aimed at simulating the dispersionof flammable gas in process areas, and subsequent explosions of gas-air mixtures. Hjertager(1985, 1986) describes the basic equations used in the FLACS model, and Hjertager, Bjørkhaug &Fuhre (1988) present the results of explosion experiments to develop and validate FLACS initially.During the course of more than 25 years of development and evaluation of the FLACS software,the numerical methods have been steadily modified and revised.

The inherent capability of FLACS has been performing explosion and dispersion calculationsto help in the improvement of oil and gas platform safety with initial focus on the North Sea.Significant experimental validation activity has contributed to the wide acceptance of FLACS asa reliable tool for prediction of natural gas explosions in real process areas offshore and onshore.

The numerical model uses a second order scheme for resolving diffusive fluxes and a second-order κ scheme (hybrid scheme with weighting between 2∧{nd} order upwind and 2∧{nd} ordercentral difference, with delimiters for some equations) to resolve the convective fluxes.

The time stepping scheme used in FLACS is a first order backward Euler scheme. Second orderschemes in time have been implemented, but are generally not used due to short time steps.Based on extensive validation, guidelines for time stepping have been established in order toget accurate results. These are based on CFL-numbers based on speed of sound (CFLC) andflow velocity (CFLV). For explosion calculations CFLC=5 and CFLV=0.5 must be applied (whichmeans that the pressure can propagate 5 cells and the flow 0.5 cells in each time step) to achieve


8.9 Linux Quick Reference 233

good results. For dispersion calculations, the guidelines are less strict as the results do not dependmuch on the time steps. Normally it is recommended to increase the time steps by a factor of 4(CFLC=20 and CFLV=2). When grid is refined near leak, it is also recommended to ignore therefined region (i.e. multiply CFLC-number with the refinement factor).

The SIMPLE pressure correction algorithm is applied (Patankar, 1980), and extended to handlecompressible flows with additional source terms for the compression work in the enthalpy equa-tion. Iterations are repeated until a mass residual of less than 10−4 is obtained.

8.9 Linux Quick Reference

This section summarises some relevant information for users that run FLACS under the Linuxoperating system. Further information concerning Linux may be found at e.g. www.linux.org

8.9.1 Distributions

FLACS work on most recent Linux distrobutions. An updated list of distributions, on whichFLACS v9.0 has been tested, is given in Hardware and Software requirements.

8.9.2 Desktop environments

FLACS works independently of the Desktop environment. The most popular environments are:

• KDE

• Gnome

Most of the developers at GexCon prefer KDE.

8.9.3 Shell

A command shell is command line interface computer program to an operating system (OS). Themost popular shells are:

• bash , setup file in home directory: .bashrc• C shell , setup file in home directory: .cshrc

8.9.4 Text editors

To create, read, write, or edit text files, e.g. Flacs input file the user must know how to use a texteditor. Recommended text editors are:

• vi / vim

• Emacs

• gedit

• kate


http://www.linux.org/

http://www.linux.org/

http://www.kde.org

http://www.gnome.org

http://tiswww.case.edu/php/chet/bash/bashtop.html

http://en.wikipedia.org/wiki/C_shell

http://www.vim.org/

http://www.gnu.org/software/emacs/

http://www.gnome.org/projects/gedit/

http://kate-editor.org/


Warning:

With Emacs: Remember to have an empty line at the end of the file. VIM adds an extra lineautomatically.

Attention:

Notepad++ is recommended if you like to edit your text file in Windows. Notepad++ doesnot change your text file, which can be the case for other editors.

8.9.5 Communication with other computers

• ssh (SSH client) is a program for logging into a remote machine and for executing com-mands on a remote machine.

• scp copies files between hosts on a network. It uses ssh for data transfer.• ftp is the user interface to the Internet standard File Transfer Protocol. The program allows

a user to transfer files to and from a remote network site.

8.9.6 Help in Linux

Most commands in Linux have related manual pages. These can be displayed by:

> man <command>

For instance:

> man ls

The help and info commands provide less extensive outputs than man, whereas apropos alsoincludes the man output for related commands. Most commands do also show help by writing

> <command> --help

For instance

> r1file --help

8.9.7 Useful commands

Recent commands are saved in a history file, located in the users home directory, and it is listedwith the command:

> history

User can define aliases. The alias commands are usually wanted to be executed every time ashell is used and are therefore usually added to the shell setup-file (.bashrc or .cshrc). alias canbe used to make new short-cuts or to change output of already existing commands. The aliassyntax differ slightly from shell to shell. See:

> man alias


8.9 Linux Quick Reference 235

The cd command changes the working directory:

> cd DIR

To move to the parent directory:

> cd ..

ls lists contents of directories:

> ls

There are a lot of options to the ls command. See

> man ls

Other frequently used commands:

• chgrp : Changes group.• chmod : Changes permissions.• chown : Changes ownership.• cp : Copies files from one place to another, or duplicates one file under a different name.• diff : Compare files line by line.• df : Keeps track of your hard disk space.• du : Lists the file sizes in kilobyte.• exit : Ends the application.• find : Looks for files with particular content.• free : Outputs the amount of free RAM on the system.• grep : Finds words in files.• gunzip : Expands files.• gzip : Compresses files.• kill : Terminates a process.• less : Views file contents.• mkdir : Creates directories.• more : Views file contents.• mv : Moves or renames files.• ps : Lists running processes on the system.• pwd : Prints the path of the working directory.• rm : Deletes files permanently.• rmdir : Deletes empty directories.• tail : Views the last part of a file. It is an useful command for monitoring Flacs running

files.• tar : Assembles files into a package and extract a package.• top : Provides live summary of running processes.• | : Pipe command, used together with other commands.



8.9.8 Permissions

Permissions are best showed by an example with a directory containing only one file (file.txt)and one directory (DIR). ls -al gives the following results:

drwxr-sr-x 2 idar gexcon 4096 2008-05-17 13:01 DIR/-rw-r--r-- 1 idar gexcon 102030 2008-05-17 14:01 file.txt

where:

• -rw-r-r- indicates the file permissions.• 1 indicates that there is one file.• idar indicates that the file belongs to the user idar.• gexcon indicates the group.• 102030 is the size of the file in bytes.• 2008-05-17 indicates the date the file was created/modified/moved.• 14:01 indicates the time the file was created/modified/moved.• file.txt is the file name.

The first character in file persmission is ’d’ if it is a directory and ’-’ else. The next nine charactersindicate the permissions, where the first three are for the user that owns the file, the next threeare for the group, and the last three are for others. There are three possible attributes:

• r : Read permission.• w : Write permission.• x : Execute permission.

In the example,

-rw-r--r--

only the user has both read and write permission, i.e. can modify the file. Members of the groupand others can read the file.


Chapter 9

Nomenclature

238 Nomenclature

9.1 Roman letters

A Area m2

a Moles of O2 in a stoichiometric reaction -c Speed of sound ms−1

cp Specific heat capacity at constant pressure JK−1kg−1

cv Specific heat capacity at constant volume JK−1kg−1

C1ε Constant in the k− ε equation; typically C1ε = 1.44 -C2ε Constant in the k− ε equation; typically C2ε = 1.92 -C3ε Constant in the k− ε equation; typically C3ε = 0.8 -CD Drag coefficient -Cµ Constant in the k− ε equation; typically Cµ = 0.09 -d Diameter mD Diffusion coefficient m2s−1

f Sub-grid obstructions turbulence generation factor -E+ Constant in wall functions; typically E+ = 11 -F Force NFD Drag force NFw Wall friction force NF/O Fuel-oxidant ratio, see definition -g, ~g Gravitational acceleration (scalar, vector) ms−2

h Specific enthalpy Jkg−1

h Heat transfer coefficient WK−1m2

h Height of the atmospheric mixing layer mIp Pressure impulse Pa · sIT Relative turbulence intensity -k Turbulent kinetic energy m2s−2

L Monin-Obukhov length scale ml Length mlLT Mixing length in the β-model, lLT = Cµk3/2ε−1 mM, Mk Molecular weight of a mixture, specie kgmol−1

m Mass kgm Mass rate kgs−1

N Total number density -n Number density -p Absolute pressure Pap0 Ambient pressure PaP Gauge pressure, overpressure Pa, barQ Heat JQ Heat rate Js−1

R, Rk Gas constant of a mixture, specie R = Ru/M Jkg−1K−1

Ru Universal gas constant 8.314Jmol−1K−1

Rfuel Reaction rate for fuel kgm−3s−1

r Radius mrox Stoichiometric fuel-oxidant ratio on mass basis -rair Stoichiometric fuel-air ratio on mass basis -SL Laminar burning velocity ms−1

SQL Quasi-laminar burning velocity ms−1

ST Turbulent burning velocity ms−1

T Absolute temperature Kt Time s


9.3 Subscripts 239

U0 Reference, characteristic velocity ms−1

ui,~u Mean velocity (ith component, vector) ms−1

u′ Root mean square of velocity ms−1

u∗ Friction velocity ms−1

V Volume m3

V Volume rate m3s−1

W∗ Dimensionless reaction rate -X Mole fraction -x Length coordinate mY Mass fraction -y Wall distance my+ Dimensionless wall distance in wall functions -z Distance above the ground mz0 Aerodynamical roughness length m

9.2 Greek letters

α Volume fraction -α Thermal diffusivity m2s−1

β Transformation factor in the β-model, see The FLACS flame model -βi Area porosity in the i th direction -βv Volume porosity -γ Isentropic ratio -γp Pressure exponent for the laminar burning velocity, see correlation. -∆ Control volume length mδH Heaviside step function. δH(a− b) = 1 if a ≥ b. δH(a− b) = 0 if a < b. -δij Kronecker delta function, δij = 1 if i = j. δij = 0 if i 6= j. -ε Dissipation of turbulent kinetic energy m2s−3

εg Surface roughness mζ Surface tension Nm−1

κ Von Karman constant; typically κ = 0.41. -λ Conductivity Wm−1K−1

µ Dynamic viscosity Pa · sµt Dynamic turbulent viscosity Pa · sµeff Effective viscosity, µeff = µ + µt Pa · sν Kinematic viscosity m2s−1

ξ Mixture fraction -ρ Density kgm−3

σ Prandtl-Schmidt number, see overview. -σij Stress tensor, see equation. Nm−2

τ Time scale sτe Integral time scale in turbulent flows sτw Wall shear stress Nm−2

Φ Equivalence ratio, see definition. -φ General variable -χ Progress variable, see definition. -

9.3 Subscripts


240 Nomenclature

a Ambient -cv Control volume -D Drag -g Ground -f Flow -i Specie index, spatial index -j Spatial index -L Laminar -n Control volume index -o Sub-grid objects -p Particle -stoich Stochiometric -T Turbulent -v Volume -w Wall -

9.4 Dimensionless groups

Bi Biot number, Bi = hdpλp

. -Da Damköhler number, Da = τe

τchem. -

Fr Froude number, Fr = inertial forcegravity force = u

l2g . -

Le Lewis number, Fr = αD = λρcpD = Sc

Pr -Ma Mach number, Ma = u

c -Nu Nusselt number, Nu = hl

λ -Pr Prandtl number, Pr = ν

α = µcpλ -

Re Reynolds number, Re = inertial forceviscous force = ρlu

µ -Sc Schmidt number, Sc = ν

D = µρD -

St Stokes number, St = τpτf

-

We Weber number, We = inertial forcesurface tension force = ρu2dp

ζ -

9.5 Abbreviations

AIT Auto Ignition Temperature -CAD Computer Aided Design -CASD Computer Aided Scenario Design -CFD Computational Fluid Dynamics -CFL Courant-Friedrichs-Levy -CFU Central Processing Unit -CMR Christian Michelsen Research -CP8 Complex Polyhedron -CV Control Volume -DDT Deflagration-to-Detonation Transition -DESC Dust Explosion Simulation Code -DNS Direct Numerical Simulation -ER Equivalence Ratio -FLACS FLame ACceleration Simulator -GTC General Truncated Cone -HSL Health and Safety Laboratory -


9.6 FLACS variables 241

HVAC Heating, Ventilating, and Air-Conditioning -LES Large Eddy Simulation -LFL Lower Flammability Limit -MIE Minimum Ignition Energy -PI Ignition Probability -QRA Quantitative Risk Analyses -RAM Random Access Memory -UFL Upper Flammability Limit -

9.6 FLACS variables

CFLC CFL number based on speed of sound -CFLV CFL number based on flow velocity -CS Speed of sound, c ms−1

DPDT Rate of pressure rise, dpdt Pas−1

EPK Turbulence ratio, εk s−1

EPS Dissipaton of turbulent kinetic energy, ε m2s−3

EQ Equivialence ratio, finite bounded, EQ = (F/O)(F/O)+(F/O)stoich

-ER Equivialence ratio, Φ -DIMP Drag impulse,

∫ t0

∣∣∣~FD

∣∣∣ dt Pa · s

DRAG Drag value,∣∣∣~FD

∣∣∣ Pa

FDOSE Dose, integral of mole fraction of fuel,∫ t

o Xfueldt sFLUX Mass flux, m

A kgm−2s−1

FMIX Mixture fraction, ξ -FMOLE Mole fraction of fuel, Xfuel -FUEL Mass fraction of fuel, Yfuel -FVAR Variance of mixture fraction, ξ2 -GAMMA Isentropic gas constant, γ -H Enthalpy, h Jkg−1

K Turbulent kinetic energy, k Jkg−1

LT Turbulent length scale, (output), lLT mMACH Mach number, Ma -MU Effective dynamic viscosity, µeff Pa · sNUSSN Nusselt number, Nu -OX Mass fraction of oxygen, YO2 -P Gauge pressure, overpressure, P barPIMP Pressure impulse,

∫ t2t1

Pdt bar · sPMAX Maximum over pressure, Pmax barPROD Mass fraction of products, Yprod barRET Turbulent Reynolds number, ρulLT

µt-

RFU Combustion rate, Rfuel kgm−3s−1

RHO Density, ρ kgm−3

RTI Relative turbulence intensity (input), IT -T Temperature, T KTAUWX Wall shear stress in x direction, τw,1 Nm−2

TAUWY Wall shear stress in y direction, τw,2 Nm−2

TAUWZ Wall shear stress in z direction, τw,3 Nm−2

TLS Turbulence lenght scale (input), lLT mTURB Root mean square of velocity, u′ ms−1


242 Nomenclature

TURBI Relative turbulence intensity (output), IT -U Velocity component in x direction, u1 ms−1

UDIMP Drag impulse in x direction,∫ t

0 FD,1dt Pa · sUDRAG Drag value in x direction, |FD,1| PaUFLUX Mass flux in x direction, ρu1 kgm−2s−1

UVW Absolute value of velocity, |~u| ms−1

V Velocity component in y direction, u2 ms−1

VDIMP Drag impulse in y direction,∫ t

0 FD,2dt Pa · sVDRAG Drag value in y direction, |FD,2| PaVFLUX Mass flux in y direction, ρu2 kgm−2s−1

VVEC Velocity vector, ~u ms−1

W Velocity component in z direction, u3 ms−1

WDIMP Drag impulse in z direction,∫ t

0 FD,3dt Pa · sWDRAG Drag value in z direction, |FD,3| PaWFLUX Mass flux in z direction, ρu3 kgm−2s−1


Chapter 10

References

244 References

Abdel-Gayed, R.G. & Bradley, D. (1981). A two-eddy theory of premixed turbulent flame prop-agation. Philosophical Transactions of the Royal Society of London, Series A, 301: 1-25.

Abdel-Gayed, R.G., Al-Khishali, K.J. & Bradley, D. (1984). Turbulent burning velocities andflame straining in explosions. Proceedings of the Royal Society of London, Series A, 391: 389-413.

Abdel-Gayed, R.G., Bradley, D. & Lawes, M. (1987). Turbulent burning velocities: A generalcorrelation in terms of straining rates. Proceedings of the Royal Society of London, Series A,414: 389-413.

Abdel-Gayed, R.G., Bradley, D., Hamid, M.N. & Lawes, M. (1984). Lewis number effects onturbulent burning velocity. Twentieth Symposium (Int.) on Combustion: 505-512.

Abdel-Gayed, R.G., Bradley, D., Lawes, M. & Lung, F.K-K (1986). Premixed turbulent burningduring explosions. Twenty-first Symposium (Int.) on Combustion: 497-504.

Altman, I.S. (1999). On condensation growth of oxide particles during gas-phase combustionof metals. Seventeenth International Colloquium on the Dynamics of Explosions and ReactiveSystems (ICDERS), 25-30 July 1999, Heidelberg, Germany.

Amyotte, P.R., Chippett, S. & Pegg, M.J. (1989). Effects of Turbulence on Dust Explosions. Jour-nal of Loss Prevention in the Process Industries, 14: 293-310.

Arntzen, B.J. & Hansen, O.R. (1997). Improved thermodynamics in FLACS, modelling the effect of in-ert gases. Report Christian Michelsen Research, CMR-97-F30009, Bergen, Norway.

Arntzen, B.J. (1993). Combustion modelling in FLACS93. Report Christian Michelsen Research,CMR-93-F25043 (Confidential).

Arntzen, B.J. (1997). Modelling of steady and transient flows in FLACS96. Report Christian MichelsenResearch,CMR-97-Fxxxxx (to be issued in 1997?) (Confidential).

Arntzen, B.J. (1998). Modelling of turbulence and combustion for simulation of gas explosions in com-plex geometries. Dr. Ing. Thesis, NTNU, Trondheim, Norway.

Arntzen, B.J., Salvesen, H.C., Nordhaug, H.F., Storvik, I.E. & Hansen, O.R. (2003). CFD mod-elling of oil mist and dust explosion experiments. Proceedings Fourth International Seminaron Fire and Explosion Hazards, 8-12 September 2003, Londonderry, Northern Ireland, UK:601-608.

ATEX 1999/92/EC (1999). Directive 1999/92/EC (ATEX 118a) of the European Parliament and theCouncil, on minimum requirements for improving the safety and health protection of workers po-tentially at risk from explosive atmospheres.

ATEX 94/9/EC (1994). Directive 94/9/EC (ATEX 100a) of the European Parliament and the Council, onthe approximation of the laws of the member states concerning equipment and protective systemsintended for use in potentially explosive atmospheres.

Babrauskas, V. (2003). Ignition Handbook. Fire Science Publishers, Issaquah, USA.

Bardon, M.F. & Fletcher, D.E. (1983). Dust explosions. Scientifig Progress, 68: 459-473.

Bartknecht, W. (1971). Brenngas- und Staubexplosionen. Forschungsbericht F45, Bundesinstitutfür Arbeitsschutz, Koblenz, Germany (in German).

Bartknecht, W. (1978). Explosionen: Ablauf und Schutzmaßnahmen. Springer-Verlag, Berlin (inGerman, English translation: Bartknecht, 1981).

Bartknecht, W. (1981). Explosions: course prevention protection. Springer-Verlag, Berlin.


245

Bartknecht, W. (1993). Explosionsschutz: Grundlagen und Anwendung. Springer-Verlag,Berlin (in German).

Barton, J. (2002). Dust explosions: Prevention and protection. IChemE, Rugby.

Batchelor, G.K. & Townsend, A.A. (1947). Decay of vorticity in the isotropic turbulence. Pro-ceedings of the Royal Society of London, Series A, 190: 534-550.

Batchelor, G.K. & Townsend, A.A. (1948a). Decay of isotropic turbulence in the initial period.Proceedings of the Royal Society of London, Series A, 193: 539-558.

Batchelor, G.K. & Townsend, A.A. (1948b). Decay of isotropic turbulence in the final period.Proceedings of the Royal Society of London, Series A, 194: 527-543.

Beck, H., Glienke, C. & Möhlmann, C. (1997). BIA-Report 13/97: Combustion and explosion char-acteristics of dusts. HVBG, Saint Augustin, Germany.

Beér, J.M., Chomiak, J. & Smooth, L.D. (1984). Fluid dynamics of coal combustion: a review.Progress in Energy and Combustion Science, 10: 177-208.

Bjerketvedt, D., Bakke, J.R. & van Wingerden, K. (1993). Gas explosion handbook, Report Chris-tian Michelsen Research, CMR-93-A25034, Bergen, Norway.

Bosch, C.J.H. van den, Weterings, R.A.P.M. (Editors) (1996) Methods for the calculations ofphysical effects. Due to releases of hazardous materials, Report TNO CPR14E.

Bradley, B., Lau, A.K.C. & Lawes, M. (1992). Flame stretch rate as a determinant of turbulentburning velocity. Philosophical Transactions of the Royal Society of London, Series A, 338: 359-387

Bradley, D. (1992). How fast can we burn? Twenty-fourth Symposium (Int.) on Combustion: 247-262.

Bradley, D. (2002). Problems of Predicting Turbulent Burning Rates. Combustion Theory andModelling, 6: 361-382.

Bradley, D., Chen, Z. & Swithenbank, J.R. (1988). Burning rates in turbulent fine dust-air ex-plosions. Twenty-second Symposium (Int.) on Combustion: 1767-1775.

Bradshaw, P. (1994). Turbulence: the chief outstanding difficulty of our subject. Experiments inFluids, 16: 203-216.

Bray, K.N.C. (1990). Studies of the turbulent burning velocity. Proceedings of the Royal Society ofLondon, Series A, 431: 315-335.

Butler, T.D., Cloutman, L.D., Dukowicz, J.K., Ramshaw J.D. (1981). Multidimensional numer-ical simulation of reactive flow in internal combustion engine. Progress in Energy and Com-bustion Science, 7: 293-315.

Cashdollar, K.L. (2000). Overview of dust explosibility characteristics. Journal of Loss Preventionin the Process Industries, 13: 183-199.

Cashdollar, K.L., Zlochower, I.A., Green, G.M., Thomas, R.A. & Hertzberg, M. (2000).Flammability of methane, propane, and hydrogen gases. Journal of Loss Prevention inthe Process Industries, 13: 327-340.

Cassel, H.M. (1964). Some fundamental aspects of dust flames. Report of Investigation 6551, U.S.Department of the Interior, Bureau of Mines, Washington.


246 References

Catlin, C.A. & Lindstedt, R.P. (1991). Premixed turbulent burning velocities derived from mix-ing controlled reaction models with cold front quenching. Combustion and Flame, 85: 427-439.

Cesana, C. & Siwek, R. (2001). Operating instructions 20-l-apparatus, 6.0. Adolf Kühner AG.

Chen, C.H. & Jaw, S.Y. (1997). Fundamentals of turbulence modeling. In Combustion: An Interna-tional Series. Taylor & Francis, Washington.

Clift, R., Grace, J.R. & Weber, M.E. (1978). Bubbles, drops and particles. Academic Press, NewYork.

CMR (1996). FLACS 96 User’s GUIDE. Report Christian Michelsen Research, CMR-96-F20073,Christian Michelsen Research AS, Bergen, Norway.

Coulsen, J.M. & Richardson, J. F. (1977). Chemical engineering. Pergamon Press, Oxford.

Crowe, C., Sommerfeld, M & Tsuji, Y. (1998). Multiphase flows with droplets and particles. CRCPress, Boca Raton.

Dahoe, A.E. & de Goey, L.P.H. (2003). On the determination of laminar burning velocity fromclosed vessel gas explosions. Journal of Loss Prevention in the Process Industries, 16: 457-478.

Dahoe, A.E. (2000). Dust explosions: a study of flame propagation. PhD-thesis, Delft University ofTechnology, Delft, Holland.

Dahoe, A.E., Cant, R.S. & Scarlett, B. (2001). On the decay of turbulence in the 20-litre explosionsphere. Flow, Turbulence and Combustion, 67: 159-184.

Dahoe, A.E., Cant, R.S., Pegg, R.S. & Scarlett, B. (2001). On the transient flow in the 20-litre ex-plosion sphere. Journal of Loss Prevention in the Process Industries, 14: 475-487.

Dahoe, A.E., Hanjalic, K. & Scarlett, B. (2002). Determination of the laminar burning velocityand the Markstein length of powder-air flames. Powder Technology, 122: 222-238.

Dahoe, A.E., van der Nat, K., Braithwaite, M. & Scarlett, B. (2001). On the sensitivity of themaximum explosion pressure of a dust deflagration to turbulence. KONA, 19: 178-195.

Dahoe, A.E., Zavenbergen, J.F., Lemkowitz, S.M. & Scarlett, B. (1996). Dust explosions inspherical vessels: the role of flame thickness in the validity of the ´cube-root law´. Journalof Loss Prevention in the Process Industries, 9: 33-44.

Eckhoff, R.K. (1984). Relevance of using (d p/d t)max data from laboratory-scale tests for pre-dicting explosion rates in practical industrial situations. VDI-Berichte, 494: 207-217.

Eckhoff, R.K. (2003). Dust explosions in the process industries. Third edition, Gulf ProfessionalPublishing, Amsterdam.

Elghobashi, S. & Truesdell, G.C. (1993). On the two-way interaction between homogeneousturbulence and dispersed solid particles. I: Turbulence modification. Physics of Fluids A(Fluid Dynamics), 5: 1790-1801.

Elghobashi, S. (1991). Particle-laden turbulent flows: direct simulation and closure models. Ap-plied Scientific Research, 48 :301-314.

Elghobashi, S. (1994). On predicting particle-laden turbulent flows. Applied Scientific Research,52: 309-329.

EN 1127-1 (1997). Explosive atmospheres – Explosion prevention and protection, Part 1: Basicconcepts and methodology. European Standard, CEN, Brussels, _XXX_Month, 1997.


247

EN 14491 (2006). Dust explosion venting protective systems. European Standard, CEN, Brussels,March 2006.

EN 50281-3 (2000). Equipment for use in the presence of combustible dust, Part 3: Classification ofareas where combustible dusts are or may be present. European Standard, CEN, Brussels, _-XXX_Month, 1997.

Ertesvåg, I.S. (2000). Turbulent strøyming og forbrenning: Frå turbulensteori til ingeniørverkty. ISBN82-519-1568-6. TAPIR Forlag, Trondheim (in Norwegian).

Gibbs, G.J. & Calcote, H.F. (1959). Effect of molecular structure on burning velocity. Journal ofChemical and Engineering Data, 4: 226-237.

Glassman, I. (1996). Combustion. Third Edition. Academic Press, San Diego.

Golder, D. (1972). Relations among stability parameters in the surface layer. Boundary-Layer Me-teorology 3: 47-58.

Gore, R.A. & Crowe, C.T. (1989). Effect of Particle Size on Modulating Turbulent Intensity. In-ternational Journal of Multiphase Flow, 15: 279-285.

Goroshin, S. & Lee, J. (1999). Laminar dust flames: A program of microgravity and groundbased studies at McGill. Fifth International Microgravity Combustion Workshop, 18-20 May1999, Cleveland, Ohio.

Gülder, O.L. (1990). Turbulent premixed flames for different combustion regimes. Twenty-thirdSymposium (Int.) on Combustion: 743-750.

Han, J., Arya, S. P., S. Shen, Lin, Y.-L. (2000) An estimation of turbulent kinetic energy and en-ergy dissipation rate based on atmospheric boundary layer simimlarity theory. NASA/CR-2000-212298

Han, O-S., Yashima, M., Matsuda, T., Matsui, H., Atsumi, M. & Ogawe, T. (2000). Behaviourof flames propagating through Lycopódium dust clouds in a vertical duct. Journal of LossPrevention in the Process Industries, 13: 449-457.

Han, O-S., Yashima, M., Matsuda, T., Matsui, H., Atsumi, M. & Ogawe, T. (2001). A study offlame propagation mechanisms in Lycopódium dust clouds based on dust particles’ be-haviour. Journal of Loss Prevention in the Process Industries, 14: 153-160.

Hansen, O.R. & Åsheim, O. (1995). Simulation of explosion experiments using FLACS93 version 2.0.Report Christian Michelsen Research, CMR-95-F20004 (Confidential).

Hansen, O.R. & Storvik, I. (1993). FLACS93 version 1.0 – validation report.Report ChristianMichelsen Research, CMR-93-F25052 (Confidential).

Hansen, O.R. & Teigland, R. (1994). A multi-block extension of FLACS for BLAST calculations.Report Christian Michelsen Research, CMR-94-F25053 (Confidential).

Hansen, O.R. & Teigland, R. (1995). Validation of the BLAST option in FLACS.Report ChristianMichelsen Research, CMR-95-F20005 (Confidential).

Hansen, O.R. & van Wingerden, K. (1993). Modelling of the effects of water spray on gas explosions– implementation in FLACS93.Report Christian Michelsen Research, CMR-93-F25048 (Confi-dential).

Hansen, O.R. (1994). Grid dependency study using FLACS.Report Christian Michelsen Research,CMR-94-F25047 (Confidential).


248 References

Hansen, O.R. (1997). Improvement of the waterspray model in FLACS.Report Christian MichelsenResearch, CMR-97-F30012 (Confidential).

Hansen, O.R., Skjold, T. & Arntzen, B.J. (2004). DESC – A CFD-tool for dust explosions. Inter-national ESMG Symposium, Nürnberg, 16-18 March 2004.

Hattwig, M. & Steen, H. (2004). Handbook of explosion prevention and protection. Wiley-VCH Ver-lag, Weinheim.

Hauert, F., Vogl, A. & Radant, S. (1994). Measurement of turbulence and dust concentration insilos and vessels. Proceedings Sixth International Colloquium on Dust Explosions, Shenyang,August 29 – September 2, 1994: 71-80.

Hauert, F., Vogl, A. & Radant, S. (1996). Dust cloud characterization and the influence on thepressure-time histories in silos. Process Safety Progress, 15: _XXX_PageRange.

Hertzberg, M., Zlochower, I.A. & Cashdollar, K.L. (1992). Metal dust combustion: explosionlimits, pressures, and temperatures. Twenty-fourth Symposium (Int.) on Combustion: 1827-1835.

Hinze, J.O. (1972). Turbulent fluid and particle interactions. Progress in Heat and Mass Transfer, Vol-ume 6, Proceedings of the International Symposium on Two-Phase Systems, Hetsroni, G., Side-man, S. & Hartnett, J.P. (Eds.), Pergamon Press, Oxford: 433-452.

Hinze, J.O. (1975). Turbulence. Second Edition. McGraw-Hill, New York.

ISO 6184-1 (1985). Explosion Protection Systems – Part 1: Determination of explosion indices of com-bustible dusts in air. ISO.

Jarosinski, J. & Podfilipski, J. (1999). Combustion mechanism of dust clouds in microgravity.Fifth International Microgravity Combustion Workshop, Cleveland, Ohio.

Jarosinski, J., Podfilipski, J. & Pu, Y.K. (1999). Visualisation of dust explosions under micro-gravity conditions. Sevententh International Colloquium on the Dynamics of Explosionsand Reactive Systems (ICDERS), Heidelberg, Germany.

Kauffman, C.W., Srinath, S.R., Tezok, F.I., Nicholls, J.A. & Sichel, M. (1984). Turbulent and ac-celerating dust flames. Twentieth Symposium (Int.) on Combustion: 1701-1708.

Kobayashi, H., Kawabata, Y. & Maruta, K. (1998). Experimental study on general correlation ofturbulent burning velocity at high pressure. Twenty-seventh Symposium (Int.) on Combustion:941-948.

Kobayashi, H., Nakashimi, T., Tamura, T., Maruta, K. & Niioka, T. (1997). Turbulence mea-surements and observations of turbulent premixed flames at elevated pressures up to 3.0MPa. Combustion and Flame, 108: 104-117.

Koch, D.L. (1990). Kinetic Theory for a Monodisperse Gas-Solid Suspension. Physics of FluidsA, 2 (10): 1711-1723.

Krause, U. & Kasch, T. (2000). The influence on flow and turbulence on flame propagationthrough dust-air mixtures. Journal of Loss Prevention in the Process Industries, 13: 291-298.

Kulick, J.D., Fessler, J.R. & Eaton, J.K. (1994). Particle response and turbulence in fully devel-oped channel flow. Journal of Fluid Mechanics, 277: 109-134.

Kundu, P.K. and I.M. Cohen (2004). Fluid Mechanics. Amsterdam, Elsevier Academic Press,Amsterdam.

Kuo, K.K. (2005). Principles of Combustion. Second Edition. John Wiley & Sons, New Jersey.


249

Langeland, T. (1993). FLACS INTERFACES – file interface description.Report Christian MichelsenResearch, CMR-93-F40004, Bergen, Norway.

Langeland, T. (1993). FLACS Interfaces, File Interface Description .Report Christian MichelsenResearch, CMR Report CMR-93-F40004, Bergen, Norway.

Langeland, T. (1995). FLACS INTERFACES – installation guide.Report Christian Michelsen Re-search, CMR-95-F40003, Bergen, Norway.

Langeland, T. (1997). Flowvis Version 3.2, User’s Guide.Report Christian Michelsen Research,CMR-97-F50005, Bergen, Norway.

Lauder, B.E. & Spalding, D.P. (1974). The numerical computation of turbulent flows. ComputerMethods in Applied Mechanics and Engineering, 3: 269-289.

Law, C.K. (1993). _XXX_Title. In: Peters, N. & Rogg, B. (Eds.) , Reduced Kinetic Mechanisms forApplication in Combustion Systems, Springer, Berlin 1993.

Lee, J.H.S. (1988). Dust explosion parameters: their measurement and use. VDI-Berichte, 701:113-122.

Lee, J.H.S., Pu, Y.K. & Knystautas, R. (1987). Influence of turbulence on closed volume explo-sion of dust-air mixtures. Archivum Combustionis, 7: 279-297.

Lipatnikov, A.N. & Chomiak, J. (2002). Turbulent flame speed and thickness: phenomenology,evaluation, and application in multi-dimensional simulations. Progress in Energy and Com-bustion Science, 28: 1-74.

Loth, E. (2000). Numerical approaches for motion of dispersed particles, droplets and bubbles.Progress in Energy and Combustion Science, 26: 161-223.

Mannan, S. (2005). Lees’ loss prevention in the process industries.Third Edition. Elsevier Butter-worth Heinemann, Amsterdam.

Meinköln, D. (1999). Oxide layer effects in metal particle combustion. Fifth International Micro-gravity Combustion Workshop, 18-20 May, Cleveland, Ohio.

Metghalchi, M., Keck, J.C. (1980). Laminar burning velocity of propane-air mixtures at hightemperature and pressure. Combustion and Flame, 38: 143-154.

Mintz, K.J. (1993). Upper explosive limit of dusts: experimental evidence for its existence undercertain circumstances. Combustion and Flame, 94: 125-130.

Mitgau, P. (1996). Einfluß der Turbulenzlänge und der Schwankungsgeschwindigkeit auf die Verbren-nungsgeschwindigkeit von Aerosolen. Abteilung Reaktionskinetik, Göttingen, Max-Planck-Institut für Strömungsforschung.

Motif (_XXX_Year). Motif 1.2 User’s Guide Sun Microsystems. Inc. 2550 Garcia Avenue, MountainView, California 94043 U.S.A. Part No: 801-5363-10

Muhammadi, B. & Pirinneau, O. (1994). Analysis of the k-epsilon turbulence model. John Wiley &Sons, Masson, Paris.

Nagy, J. & Verakis, H.C. (1983). Development and control of dust explosions. In the series: Occupa-tional Safety and Health, 8, Marcel Dekker, New York.

Nagy, J., Conn, J.W. & Verakis, H.C. (1969). Explosion development in a spherical vessel. Report ofInvestigation 7279, U.S. Department of the Interior, Bureau of Mines.

Nagy, J., Seiler, E.C., Conn, J.W. & Verakis, H.C. (1971). Explosion development in closed vessel.Report of Investigation 7502, U.S. Department of the Interior, Bureau of Mines, Washington.


250 References

NASA (2005). ThermoBuild – interactive tool using the thermodynamic database at NASAGlenn Research Center.

NFPA 655 (2007). Standard for prevention of sulphur fires and explosions – 2007 edition. NationalFire Protection Association, Quincy, MA.

NFPA 68 (2007). Standard on explosion protection by deflagration venting – 2007 edition. NationalFire Protection Association, Quincy, MA.

NORSOK Standard Z-013 (2001). Risk and emergency preparedness analysis. Rev. 2, NorwegianTechnology Centre.

Patankar, Suhas V. (1980). Numerical Heat Transfer and Fluid Flow. Taylor & Francis, 1980

Peirano, E. & Leckner, B. (1998). Fundamentals of turbulent gas-solid flows applied to circulat-ing fluidized bed combustion. Progress in Energy and Combustion Science, 24: 259-296.

Popat, N.R., Catlin, C.A., Arntzen, B.J., Lindstedt, R.P., Hjertager, B.H., Solberg, T., Saeter, O. & Van den Berg, A.C. (1996).Investigation to improve the accuracy of computational fluid dynamic based explosionmodels. Journal of Hazardous Material, 45: 1-25.

prEN14373 (2002). Explosion suppression systems. CEN, Brussels (draft March 2002). _XXX_-Update

prEN14460 (2002). Explosion resistant equipment. CEN, Brussels (draft April 2002). _XXX_Update

prEN14491 (2002). Dust explosion venting protective systems. CEN, Brussels (draft June 2002)._XXX_Update

prEN15089 (2004). Explosion isolation systems. CEN, Brussels (draft November 2004). _XXX_-Update

Press, W.H., Teukolsky, S.A., Vetterling, W.T. & Flannery, B.P. (1992). Numerical recipes in C.The art of scientific computing. Second edition. Cambridge. _XXX_Publisher_and_-FORTRAN_Edition

Pu, Y.K. (1988). Fundamental characteristics of laminar and turbulent flames in cornstarch dust-airmixtures. PhD Thesis, McGill University, Quebec.

Pu, Y.K., Jarosinski, J., Johnson, V.G. & Kauffman, C.W. (1990). Turbulence effects on dust explo-sions in the 20-liter spherical vessel. Twenty-third Symposium (Int.) on Combustion: 843-849.

Pu, Y.K., Jarosinski, J., Tai, C.S., Kaufmann, C.W. & Sichel, M. (1988a). The investigation of thefeature of dispersion induced turbulence and its effect on dust explosions in closed vessels. Twenty-second Symposium (Int.) on Combustion: 1777-1787.

Pu, Y.K., Li, Y.C., Kauffman, C.W. & Bernal, L.P. (1989). Determination of turbulence parameters inclosed explosion vessels. Twelvth International Colloquium on the Dynamics of Explosions andReactive Systems (ICDERS), Ann Arbor, Michigan, 23-25 July: 107-123.

Pu, Y.K., Mazurkiewicz, J., Jarosinski, J. & Kaufmann, C.W. (1988b). Comparative study of theinfluence of obstacles on the propagation of dust and gas flames. Twenty-second Symposium (Int.)on Combustion: 1789-1797.

RASE (2000). The RASE project – explosive atmospheres: risk assessment of unit operations and equip-ment; report: methodology for the risk assessment of unit operations and equipment for use inpotentially explosive atmospheres. EU Project No. SMT4-CT97_2169. 17 March 2000.

Rogers, C.B. & Eaton, J.K. (1991). The effect of small particles on fluid turbulence in a flat-plate,turbulent boundary layer in air. Physics of Fluids A, 3: 928-937.


251

Salvesen, H.-C. & Storvik, I.E. (1994). Subgrid modelling of drag forces for flow through louvres andgrating. Report Christian Michelsen Research, CMR-94-F25017 (Confidential).

Salvesen, H.-C. (1995). Modelling of jet release of liquefied gas under high pressure. Report ChristianMichelsen Research, CMR-95-F20062 (Confidential).

Salvesen, H.-C. (1996a). Subgrid modelling of flow through louvres and grating – drag and deflectionforces. Report Christian Michelsen Research, CMR-96-F20052 (Confidential).

Salvesen, H.-C. (1996b). FLACS validation simulations of explosion experiments in 1:5 scale M24module using new subgrid model for flow through louvres. Report Christian Michelsen Research,CMR-96-F20058 (Confidential).

Sand, I.Ø. & Bakke, J.R. (1989). Wall-function boundary conditions in the solution of the Navier-Stokes and the energy equations. Report Christian Michelsen Research, CMI-25110-3 (Con-fidential).

Senecal, J.A. & Beaulieu, P.A. (1998). KG: new data and analysis. Process Safety Progress, 17: 9-15.

Shirolkar, J.S., Coimbra, C.F.M. & Queiroz McQuay, M. (1996). Fundamental aspects of mod-eling turbulent particle dispersion in dilute flows. Progress in energy and combustion science,22: 363-399.

Siwek, R. (1977). 20-L Laborapparatur für die Bestimmung der Explosionskenngrößen brennbarerStäube. Thesis, HTL Winterthur, Switzerland.

Siwek, R. (1988a). Zuverlässige Bestimmung explosionstechnischer Kenngrößen in der 20-Liter-Laborapparatur. VDI Berichte, 701: 215-262.

Siwek, R., (1988b). Reliable determination of safety characteristics in the 20-litre apparatus. Pro-ceedings of Conference on Flammable Dust Explosions, St. Louis, 2-4 November 1988.

Siwek, R., van Wingerden, K., Hansen, O.R., Sutter, G., Schwartzbach, Chr., Ginger, G. & Meili, R. (2004).Dust explosion venting and suppression of conventional spray driers. Eleventh Interna-tional Symposium on Loss Prevention, Prague, 31 May to 3 June 2004.

Skjold, T, Arntzen, B.J., Hansen, O.R., Taraldset, O.J., Storvik, I.E. & Eckhoff, R.K. (2005a).Simulating dust explosions with the first version of DESC. Process Safety and EnvironmentalProtection, 83: 151-160.

Skjold, T. & Hansen, O.R. (2005). The development of DESC – A dust explosion simulationcode. International European Safety Management Group (ESMG) Symposium, 11-13 October2005, Nürnberg, Germany, 24 pp.

Skjold, T. (2003). Selected aspects of turbulence and combustion in 20-litre explosion vessels.Cand. Scient. Thesis, Department of Physics, University of Bergen, Norway.

Skjold, T. (2007). Review of the DESC project. Journal of Loss Prevention in the Process Industries,20: 291-302.

Skjold, T., Arntzen, B.J., Hansen, O.J., Storvik, I.E. & Eckhoff, R.K. (2005b). Simulation ofdust explosions in complex geometries with experimental input from standardized tests.Journal of Loss Prevention in the Process Industries, 19: 210-217.

Skjold, T., Larsen, Ø. & Hansen, O.R. (2006b). Possibilities, limitations, and the way ahead fordust explosion modelling. HAZARDS XIX, 28-30 March 2006, Manchester, UK: 282-297.


252 References

Skjold, T., Y.K. Pu, Arntzen, B.J., Hansen, O.J., Storvik, I.E., Taraldset, O.J. & Eckhoff, R.K. (2005c).Simulating the influence of obstacles on accelerating dust and gas flames. Poster presenta-tion Twentieth International Colloquium on the Dynamics of Explosions and Reactive Systems(ICDERS), 31 July to 5 August 2005, Montreal, Canada.

Skrbek, L. & Stalp, S.R. (2000). On the decay of homogeneous isotropic turbulence. Physics ofFluids, 12: 1997-2019.

Sokolik, A. S., Karpov, V. P. & Semenov, E. S. (1967). Turbulent combustion of gases. Combus-tion, Explosion, and Shock Waves, _XXX_Volume: 36-45 (Translated from Fizika Goreniya iVzryva, 3: 61–76, 1967].

Squires, K.D. & Eaton, J.K. (1991). Preferential concentration of particles by turbulence. Physicsof Fluids A, 3: 1169-1178.

Steinberg, T.A., Wilson, D.B. & Benz, F. (1992). The combustion phase of burning metals. Com-bustion and Flame, 91: 200-208.

Strehlow, R.A. (1979). Fundamentals of Combustion. Robert E. Kreiger publishing company, NewYork.

Tai, C.S., Kauffman, C.W., Sichel, M. & Nicholls, J.A. (1988). Turbulent dust combustion in ajet-stirred reactor. Progress Astronautics and Aerodynamics, 113: 62-86.

Tennekes, H. & Lumley, J.L. (1972). A First Course in Turbulence. The MIT Press, Cambridge.

Truesdell, G.C. & Elghobashi, S. (1994). On the two-way interaction between homogeneousturbulence and dispersed solid particles. II: Particle dispersion. Physics of Fluids A, 6: 1405-1407.

Tsinober, A. (2001). An informal introduction to turbulence. Kluwer Academic Publishers, Dor-drecht.

Turns, S.R. (1996). An introduction to combustion: concepts and applications. McGraw-Hill, NewYork.

Vagelopoulos, C.M., Egolfopoulos, F.N. & Law, C.K. (1996). Further considerations on the de-termination of laminar flame speeds with the counterflow twin-flame technique. Twenty-fifth Symposium (Int.) on Combustion: 1341-1347.

van der Wel, P.G.J. (1993). Ignition and Propagation of Dust Explosions. PhD Thesis, Delft Univer-sity Press, Delft.

van Wingerden, K. (1996). Simulations of dust explosion using a CFD-code. Proceedings Sev-enth International Colloquium on Dust Explosions, Bergen: 6.42-6.51.

van Wingerden, K., Arntzen, B.J. & Kosinski, P. (2001). Modelling of dust explosions. VDI-Berichte, 1601: 411-421.

Veynante, D. & Vervisch, L. (2002). Turbulent combustion modelling. Progress in Energy andCombustion Science, 28: 193-266.

Williams, F.A. (1986). Lectures on applied mathematics in combustion: Past contributions andfuture problems in laminar and turbulent combustion. Physica D, 20D: 21-34.

Yuan, Z. & Michaelides, E.E. (1992). Turbulence modulation in particulate flows – a theoreticalapproach. International Journal of Multiphase Flow, 18 (5): 779-785.

Zevenbergen, J.F. (2004). Report on turbulence measurements in the 20-litre sphere: For partnersDESC project. The Explosion Group of Delft University of Technology.


253

Zhen, G. & Leuckel, W. (1995a). Influence of transient injection induced turbulent flow on gasand dust explosions in a closed vessel. Loss Prevention and Safety Promotion in the ProcessIndustries, 2: 257-268.

Zhen, G. & Leuckel, W. (1995b). The dynamic flow condition of dust dispersion induced turbu-lence. Proceedings of Tenth Symposium of Turbulent Shear Flows, 3 (25): 7-12.

Zhen, G. & Leuckel, W. (1996). Determination of dust-dispersion-induced turbulence and its in-fluence on dust explosions. Combustion Science and Technology, 113-114: 629-639.

Zhen, G. & Leuckel, W. (1997). Effects of ignitors and turbulence on dust explosions. Journal ofLoss Prevention in the Process Industries, 10: 317-324.


Index

a1file, 202abbreviations, 240Aerodynamic roughness length, 77air leak, 87ASCII file, 200–202assembly, 45average pressure, 124axis show, 55

best practice, 206beta model, 229blank, 47block, 103Boundary conditions, 70Boundary, Bernoulli, 70Boundary, Eqchar, 70Boundary, Euler, 70, 71Boundary, Nozzle, 70, 71Boundary, Plane Wave, 70Boundary, Symmetry, 70Boundary, Wind, 70, 72, 227bugs and problems in CASD, 110Burning velocity, 230

CASD, 32casd command line options, 32cc file, 147CFLC, 67CFLV, 67change colour table, 166cloud file, 140cloud interface, 140cn file, 148cofile, 198colour scheme, 37Combustion modelling, 229comerge, 198command input CASD, 35contours, 173

database maintenance, 42database, connect to , 41database, creating, 41dbfutil, 42delete grid line, 57

delete instance, 51delete subtree, 51diffuse leak, 86Dissipation, transport equation, 224dose, 124dumpfile, reading old, 111

Enthalpy, transport equation, 224Equivalence ratio , 222equivalent stoichiometric gas cloud, 87error, 153example, 20, 180export, Flowvis , 163exporting geometry CASD, 106

fan leak type, 87Flame model, 229flash, 195Flowvis, 156Fluid flow equations, 224font settings, 179Fuel lean mixture, 222Fuel rich mixture, 222Fuel, transport equation, 224Fuel-oxidant ratio , 222

gas cloud, 83gas cloud definition, 140gas composition, 83Gas Explosion Programme (GEP), 2gas monitor region, 102gas position and volume, 83Gas Safety Program (GSP), 3Gas thermodynamics, 222geo2flacs, 188geometry appearance, 174geometry import, 111, 188geometry menu, 39geometry, new, 40geometry, open, 40gexcon colour convention, 37global objects, 43gm, 192grating, 100Gravity constant, 75

INDEX 255

grid, 56, 192grid guidelines, 58grid smooth, 57grid stretch, 57Ground roughness, 77

heavy hydrocarbons, 111hiding geometry, 174hinged panel, 81

Ideal gas law, 223ignition, 90import grid, 58import, Flowvis, 163inactive panel, 81Initial conditions, 75Initial temperature, 76instance, add, 46interpolate, 179invisible, 47

jet, 193jet leak, 86Jet sources, 230job numbers, 5

leak, 85leak buildup time, 87leak excess area, 154license terms, 180local object, properties, 47local objects, 43LOD, 107louvre panel, 95

macro file, 108maintenance, 19mark subtree, 51Mass balance, 224mass residual, 138, 153material, 40matrix, 47Mixing of gases, 222Mixture fraction, 222Mixture fraction, transport equation, 224modify page, 166Momentum equation, 224Monin-Obukhov length scale, 227monitor panel, 79, 81monitor points, 62move grid line, 57moving plot elements, 180multiblock, 103

new Flowvis presentation, 160, 162

numercal schemes, 232

object create, 41object open, 41object, open, 48online help, 180optimize computer load, 117options, 179overlay panel, 81

panel drag coefficient, 83panel maximum travel distance, 83panel opening pressure difference, 82panel porosity, 82panel pressure, 124panel sub sizes, 83panel type imp, 82panel weight, 82panels, 80particle traces, 175Pasquill class, 78pdf, 164plastic panel, 81plot domain, 171plot specification, Flowvis, 170Pool leakage file, 144Pool setup file, 142popout panel, 81porcalc, 61porsity calculation, 61Portable Document Format, 164position grid line, 56Prandtl-Schmidt numbers, 226pressure relief panels, 78printing, Flowvis, 163Progress variable, 222project, new, 40project, open, 40

Q8, 87Q9, 87

r1file, 200r3file, 201range colours, 166rdfile, 197rename, 48rotate, 175run manager, 15run script, 117

scale object, 47scenario, 61scenario file, 132script, 117


256 INDEX

select object, 47shocked flow, 193simulation volume, 56single colours, 166Speed of sound , 223Stoichiometric mixture, 222Stoichiometric reaction, 223stopping simulations, 117substitute job, 161, 165substitute object, 47substitute subtree, 51suction, 87support, 19

Toxic dose, 128Toxic probabilty of death, 128Toxic probit function, 128toxic substances, 125Turbulence model, 225Turbulence viscosity, 225Turbulent kinetic energy, transport equation,

224two-phase flow, 195

unblank, 47unspecified panel, 81user defined species, 102utilities, 188

value on curve, 178variable appearance, 172variables, 241vector plot, 174Velocity, friction , 227ventilation time stepping, 136verify porosity, 178verify porosity option, 178vessel burst, 138visible, 47

Wall functions, 226warning, 153water spray, 91Wind, 72, 77, 227Wind buildup time, 73Wind direction, 73Wind speed, 72windows service license manager, 13
