Determining safety distance in process design

On 9 July 1976, one kilogram of 2,3,7,8 tetrachlorodibenzo-dioxin (TCDD) was released

through a rupture disk at the ICMESA plant in Seveso, Italy. That was not only the day when the

hazard of a toxic cloud potentially spreading over the whole commu-nity, but it was also the beginning of a huge change in the regulatory and methodological approach to process safety. Seveso Directives I (1982), II (1997) and III (2012) have introduced the concept of risk in the industry and have addressed the quantitative risk assessment (QRA) approach for siting of potentially hazardous installations.

Previously, a prescriptive approach was the general method used to manage safety and occupa-tional aspects of the industrial world. The methodological change

the safety and occupational health laws of the European Union. Through New Approach and Global Approach, the European Commission in 2000 also introduced individual responsibility for the site owner to provably certify the acceptability of risk. In the industrial sectors poten-tially affected by major hazards, such as the oil and gas and petro-chemical/chemical industries, this process has been implemented rela-tively more quickly than in others, due to the industries’ cultural back-ground and their high potential hazards. The need to minimise risk and a progressively growing consciousness about “friendly safety” (Kletz, 2010) have led to the adoption of techniques and meth-odologies which are capable of

Safety distance determination is a key design issue that may have a dramatic

impact on a refinery construction project

RENATO BENINTENDI, ANGELA DEISY RODRIGUEZ GUIO and SAMUEL MARSH

Amec Foster Wheeler

reducing post-incident measures and able to develop increasingly sustainable approaches because of their inherent low hazard and potential for harm. The key concept of ‘inherent safety’, which had been introduced several years earlier (Kletz, 2010) is “the limitation of effects by changing designs or reac-tion conditions rather than by adding protective equipment that may fail or be neglected”.

QRA studies in the industry have traditionally been implemented as separate, stand-alone tasks, often not synchronised with design devel-opment. A possible outcome of this for the design team is to be delayed while implementing suitable design and layout changes, which gener-

protective measures, a non-harmo-

impact on project cost and, last but not least, an ineffective achievement of safety targets. This is often the case with plant/equipment siting. The traditional approach consists essentially of the adoption of prescriptive distances, which may in fact be unsafe, or which may lead to the available space being used in a less than optimised manner. Amec Foster Wheeler’s experience includes a long project execution history, throughout which the necessity to develop risk-based,

safety distances between plant units, between main equipment and occupied areas, has increased in importance. This article describes this evolution and presents a state-of-the-art, quantitative risk assessment approach to safety distance determination.

Background of the methodology of the separation distance assignment Early guidance about safety distances was given by Armistead (1952), Backurst and Harker (1973), and Anderson (1982). In 1976, the Dow Chemical company included safety distances in its Fire and Explosion Index (FEI) Guide. Developed in the 1980s, the Mond Fire Explosion and Toxicity Index method is an extension of the origi-nal Dow Index method. Exxon (1998) issued some safety design

-tive values for layout spacing. Similar separation distance tables have been given by Mecklenburgh (1985) and Industrial Risk Insurers. Mecklenburgh also carried out a categorisation of the most important hazardous scenarios to be used in support of plant layout.

Prescriptive separation distances for small and large tanks containing

the Health and Safety Executive in 1998 and, for LPG, in 2013. The US Center for Chemical Process Safety (CCPS) (2003) has provided typical separation distances between vari-ous elements in open-air process facilities. These tables are based on historical and current data from

and insurance sectors. The data were developed based on experi-ence and engineering judgment and, as clearly stated in the CCPS textbook, not always on calculations.

On the other hand, risk- and consequence-based methods have increased in importance and this has

and standards. In 1996, the

www.eptq.com PTQ Q1 2015 31

www.eptq.com PTQ Q1 2015 33

-

-

-

-

The FEATHER model

-

-

Determination of Safety Distances

Risk Management Program Guidance for Offsite Consequence Analysis

-

-

-

-

et al

Safety distance as part of inherently safer design

-

Chemical Hazard Engineering Guidelines -

-Process

Plants: A Handbook for Inherently Safer Design

-

-

Safety distance

-

34 PTQ Q1 2015 www.eptq.com

Amec Foster Wheeler aimed at auto-matically identifying the hazard scenarios and providing frequency and safety distances, along with iso-contour diagrams. Safety

from the release or blasting (BLEVE) -

mable, heat-radiation, overpressure endpoint. This software has been programmed in Microsoft Visual Basic and incorporates API’s physi-cal-chemical database and

-trated in Figure 1, where the light

the output data or intermediate data automatically calculated or uploaded by the software.

Chemical substances

octane, crude oil, hydrogen, carbon

along with the corresponding hazard scenarios.

Flow models

are calculated according to adiaba-

liquid state at the outlet because the Fauske and Epstein critical length (1988) for phase transition is not

-lated through Torricelli’s formula.

Figure 1 Flow chart of FEATHER software

(LEL, ERPG, combustion, heat...)

Design case

Plant units/modules

Main equipment/piping

Substance(s) selection

Process data

Data collection

Substances and equipment data

Process and layout data

Hazardous properties

Hazard intrinsic scenario

Field hazard process scenario

Ignition sources

Jet fire Near/medium/far-field flow(flame length, heat, radiation) (light/heavy gases)

(yes/no)

Specific data requirements(particular endpoints)

(thermodynamic, toxic)

Hazardous material identity

Main process data

Equipment/pipe selection

Module(s) congestion data

Programme substances properties database

Flammability, toxicity

Multiple hazard data(flammability and toxicity)

Failure data(hole size, failure rates...)

Outflow model BLEVE

LPG

Carbon dioxide

Ammonia

One phase

Two phase

Diked/unidiked pool fire

Frequency

Safety distance

Vaporisation flow

Congested-space blast Flash fire Toxic cloud Open-space blast

Frequency Safety distance

www.eptq.com PTQ Q1 2015 35

(module and units) has been modelled according to the method provided by Puttock. The user is requested to provide geometrical and congestion data. The software

35

45

40

30

25

20

15

10

5

10 20 30 40 50 60

FEATHER

PHAST

Figure 2 Propane jet fire; comparison of FEATHER vs PHAST

300

350

250

200

150

100

50

Radia

tive s

afe

ty d

ista

nce, m

0

0 20000 40000 60000 80000 100000

FEATHER

PHAST

Figure 3 Heptane pool fire; comparison of FEATHER vs PHAST

Figure 4 Fireball; comparison of FEATHER vs PHAST

120

200

180

160

140

100

80

60

40

20

0

FEATHER

PHAST

0 100 200 300

DispersionDispersion modelling has been approached by tuning a blending of sequential models, taking into account the initial jet momentum/

-

(Britter and McQuaid, 1988), and the

Wind and Pasquill weather catego-ries data are selected by the user.

Pool evaporation and strippingMacKay and Matsugu’s (1987) formula has been adopted because of its validation against experi-ments. For crude oil, gasolines, diesels and kerosenes, the Reid vapour pressure can be used to estimate the mass of vapour evapo-rating from the liquid. It has been assumed that all of the toxic gas is stripped from the liquid in order to be conservative. Once this mass of toxic vapour is known, dispersion models have been applied.

Hazard scenariosHazard scenarios are automatically

the characteristics of the substances.

Pool fire

have been modelled. The evapora-tion effect has been considered according to the methodology outlined above. The TNO (2005) model has been adopted.

Jet fire

Flame dimensions and the radiative

according to TNO (2005). A light or

Flash fire and toxic release

considering the distance to substance lower explosive limits. This is conservative and reasonable. Therefore, toxic release has been modelled in the same way, just

Open space explosion

The TNT method has been selected for modelling open space explo-sion. Despite the claimed poor accuracy stated in the literature,

comparison with DNV PHAST has shown very good results.

Congested space explosion

Explosion in congested space

36 PTQ Q1 2015 www.eptq.com

and Related Industries, John G Simmonds & Co,

Inc., New York, 1952.

2 Anderson F V, Plant Layout In: Kirk R E,

Othmer D F, 1982, op. cit., vol. 18, 23.

3 Backhurst J R, Harker J H, Process plant

design, American Elsevier, New York, 1973.

4 Benintendi R, Turbulent jet modelling for

hazardous area classification, Journal of Loss

Prevention in the Process Industries, 2010, vol

23, issue 3, 373–378.

5 Britter R E, McQuaid J, Workbook on the

Dispersion of Dense Gases, HSE Contract

Research Report No. 1.7, 1988.

6 Cox A W, Lees F P, Ang M L, Classification of

Hazardous Locations, IChemE, 1993.

7 Crowl D, Louvar J, Chemical process safety

- Fundamentals with applications, New Jersey,

Prentice Hall PTR, 2002.

8 Fauske H K, Epstein M, Source term

considerations in connection with chemical

accidents and vapour cloud modelling, Journal

of Loss Prevention in the Process Industries, vol

1, April1988.

9 Ivings M J, Clarke S, Gant S E, Fletcher B,

Heather A, Pocock D J, Pritchard D K, Santon R,

Saunders C J, Area Classification for secondary

releases from low pressure natural gas

systems, Health and Safety Executive Research

Report RR630, 2008.

10 Kletz T, Amyotte P, Process Plants: A

Handbook for Inherently Safer Design, CRC

Press Taylor & Francis Group, 2010.

11 Kawamura P I, MacKay D, The Evaporation

of volatile liquids, J. of Hazardous Materials,

1987, 15, 365-376.

12 Kletz T, Amyotte P, Process plants: A

Handbook for Inherently Safer Design, 2nd ed,

CRC Press Taylor & Francis Group, 2010.

13 Marsh S, Guidelines for the determination

of safety distances with respect to fire,

explosion and toxic hazards, Foster Wheeler,

2013.

14 Mecklenburgh J C, Process Plant Layout,

John Wiley & Sons, New York, 1985.

15 TNO, Method for the Calculation of Physical

Effects (Yellow Book), Ed: van den Bosch C J H,

Weterings R A P M, 2005.

Renato Benintendi is Principal Consultant,

Loss Prevention with Amec Foster Wheeler,

Reading, UK. He holds an advanced degree in

chemical engineering from the University of

Naples, Italy, as well as a master’s degree in

environmental and safety engineering.

Angela D Rodriguez Guio is a Senior Process

Safety Engineer. She holds a bachelor’s degree

in chemical engineering from Universidad

Nacional de Colombia, a postgraduate degree

in occupational health and safety from

the Universidad Distrital Francisco Jose de

Caldas and an MSc in process safety and loss

prevention from the University of Sheffield, UK.

Samuel Marsh is a Process Engineer with Amec

Foster Wheeler. He holds a master’s degree in

chemical engineering from the University of

Manchester, UK.

very good within the sensitivity analysis results. The software is not intended to replace validated soft-ware adopted in QRA and consequence assessment studies. Nevertheless, it can be considered a

-tion of initial equipment spacing.

ConclusionAmec Foster Wheeler is implement-ing a risk-based approach to safety distance determination early in the design of process plant. Spacing of equipment and separation distance

has been traditionally approached by means of prescriptive distances,

risk-based methodology has been used and software has been devel-oped, which includes and integrates validated models and provides satisfactory predictive results in terms of frequency and safety distances. The method is considered a step forward in the implementa-tion of inherently safer design.

Based on a paper presented at the IChemE

HAZARDS 24 Conference, Edinburgh, 7-9 May

2014.

Further reading

1 Armistead G, Safety in Petroleum Refining

automatically calculates whether a

module/unit and assumes that

reasonable and conservative hypothesis.

BLEVE

BLEVE has been modelled accord-ing to the method provided by CCPS.

Accuracy and validityFEATHER works according to the

Typically, a frequency of 10-4/yr is

which can be changed. Accordingly, a dual option has been imple-mented, which allows for the provision of the iso-contours for the

of the possible incidents. The soft-

with DNV PHAST results. Some

Figures 2, 3 and 4, and in Tables 1 and 2, showing the calculation of distances to acceptable radiation

-

-

masses. The comparability is also

Distance from Pipe rack Assumed operating Distance Distance pressure bar, g FEATHER, m tables, mTo Heat exchanger 10To Columns, accumulators, drums 10To Rundown tanks 20 90 (jet fire) 100To Moderate hazard reactors 150 ÷ 300 (fireball) 10To Intermediate hazard reactors 15To High hazard reactors 25

Comparison of FEATHER distances (to 8 kw/m2) with tabulated (prescriptive) distances: jet fire and fireball

Table 1

Distance from Intermediate Assumed Distance Distance hazard pumps substance FEATHER, m tables, mTo Columns, accumulators, drums 10To Pipe racks 10To Heat exchangers Heptane 15÷35 15To Moderate hazard reactors 10To Intermediate hazard reactors 10To High hazard reactors 10

Comparison of FEATHER distances (to 8 kw/m2) with tabulated (prescriptive) distances: pool fire

Table 2