John HedengrenBrigham Young University

University of Utah Graduate Seminar

30 Oct 2013

30 Oct 2013

PRISM Group Overview

Monitoring / Intelli-Fields Unmanned Aerial Vehicles Oil and Gas Exploration and Production

Uncertainty Analysis Investment Planning Under Uncertainty Energy Storage and the Smart Grid

Next Generation Simulation with Optimization Solid Oxide Fuel Cells Systems Biology

2

OverviewOverview

30 Oct 2013

PRISM Group OverviewPRISM Group Overview

PRISM: Process Research and Intelligent Systems Modeling

Methods Mixed Integer Nonlinear Programming (MINLP) Dynamic Planning and Optimization Uncertain, Forecasted, Complex Systems

Fit Systems into Standard Problem Formulation

Solver development: Large-scale MINLP (100,000+ variables)

3

max

subjectto , , , 0

h , , 0

30 Oct 2013

PRISM Group Overview

Monitoring / Intelli-Fields Unmanned Aerial Vehicles Oil and Gas Exploration and Production

Uncertainty Analysis Investment Planning Under Uncertainty Energy Storage and the Smart Grid

Next Generation Simulation with Optimization Solid Oxide Fuel Cells Systems Biology

4

OverviewOverview

30 Oct 2013

UAV PlatformUAV Platform

30 Oct 2013

Monitor InfrastructureMonitor Infrastructure

30 Oct 2013

Canal – Change DetectionCanal – Change Detection

30 Oct 2013

Dynamic Optimization with UAVsDynamic Optimization with UAVs

1

2

3

8

Courtesy Sentix Corp

30 Oct 2013

Information SourcesInformation Sources

9

Controlled Airspace

Population Density

Aircraft Performance Model

Geospatial / Urban Dev Air Traffic Density

Define Risk

Compute Risk

Minimize RiskWeather

Airspace + Pop DensityAirspace + Pop Density

Multiple Sources of Information Can Be Utilized

Safety Thresholds

Functions

Courtesy Sentix Corp

30 Oct 2013

Failures to Monitor and PredictFailures to Monitor and Predict

Detect early warning signs

Automate monitoring of critical systems

Give critical data to key decision makers

30 Oct 2013

First use of fiber optic sensors on a subsea pipeline to monitor pressure, strain and vibration in external casing pipe bundle during fabrication.

Bullwinkle Platform

Troika - Gulf of Mexico

30 Oct 2013

Fiber Bragg GratingsFiber Bragg Gratings

30 Oct 2013

Tendon Tension MonitoringTendon Tension Monitoring

30 Oct 2013

Adhesion Testing for Subsea InstallationAdhesion Testing for Subsea Installation

FEA Test Article Pipe for Tension TestingAvoid Local Areas of Inelastic Deformation (>50 ksi)

30 Oct 2013

Uniform Loading in 4 Point BendingUniform Loading in 4 Point Bending

0 10 20 30 40 50 60 70-2000

0

2000

0 10 20 30 40 50 60 70-2000

0

2000

0 10 20 30 40 50 60 70-2000

0

2000

0 10 20 30 40 50 60 70-2000

0

2000

Time (minutes)

Conventional Strain GaugeFiber Optic Strain Gauge

30 Oct 2013

Tendon Band PreparationTendon Band Preparation

Cleaning with a Water Jet Polishing to Bare Metal

Marine Growth (Before) Clean Band (After)

30 Oct 2013

Diver InstallationDiver Installation

Diver with Clamp Riser Preparation

Clamp Installation Clamp Inspection

30 Oct 2013

Drilling and ProductionDrilling and Production

IntelliServ, 2013

30 Oct 2013

PRISM Group Overview

Monitoring / Intelli-Fields Unmanned Aerial Vehicles Oil and Gas Exploration and Production

Uncertainty Analysis Investment Planning Under Uncertainty Energy Storage and the Smart Grid

Next Generation Simulation with Optimization Solid Oxide Fuel Cells Systems Biology

19

OverviewOverview

Smart Grid Optimization

Smart grid integration with solar, wind, coal, biomass, natural gas, and energy storage

Nuclear integration withpetrochemicalproduction, processing, and distribution

District Heating and Cooling District Heating and Cooling

Uncertainty in Natural Gas PricesUncertainty in Natural Gas Prices

Uncertainty in Electricity PricesUncertainty in Electricity Prices

Dynamic Model for Dynamic SystemDynamic Model for Dynamic SystemH

eat

Dem

and

(MM

BTU

/hr)

Electricity Dem

and (MW

)

MW

MW

MW

Simplifying SystemSimplifying System

Create Model: Electric and Heating Demand Model (winter and summer)

Model Predictive Control ApproachModel Predictive Control Approach

Optimize to a Target RangeOptimize to a Target Range

0 2 4 6 8 10 12 14 16 18 201

2

3

4

5

Man

ipul

ated

uopt

0 2 4 6 8 10 12 14 16 18 200

5

10

15

Con

trolle

d

xopt

Optimal sequence of moves given uncertainty in the parameters

Distribution of Controlled Variables

0 2 4 6 8 10 12 14 16 18 201

2

3

4

Man

ipul

ated

uopt

0 2 4 6 8 10 12 14 16 18 200

5

10

15

Con

trolle

d

xopt

Optimize to a LimitOptimize to a Limit

Conservative movement based on worst case CV

Upper Limit

PRISM Group Overview

Monitoring / Intelli-Fields Unmanned Aerial Vehicles Oil and Gas Exploration and Production

Uncertainty Analysis Investment Planning Under Uncertainty Energy Storage and the Smart Grid

Next Generation Simulation with Optimization Solid Oxide Fuel Cells Systems Biology

29

OverviewOverview

Systems BiologySystems Biology

Objective: Improve extraction of information from clinical trial data

Dynamic data reconciliation Dynamic pharmacokinetic models (large-scale) Data sets over many patients (distributed) Uncertain parameters (stochastic)

05

1015

1.9

1.95

21

2

3

4

5

6

7

Log(

Viru

s)

HIV Virus Simulation

time (years)Log(kr1)

1.828

2.364

2.899

3.435

3.970

4.506

5.041

5.576

6.112

6.647

0 5 10 151

2

3

4

5

6

7

8

Time

log1

0 vi

rus

Dynamic Energy System Tools

Solid Oxide Fuel Cell (SOFC)

Toolbox for Object Oriented Modeling in MATLAB, Simulink, and Python

Advanced tools are required for collaborative modeling and high performance computing

Optimization BenchmarkOptimization Benchmark

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 510

20

30

40

50

60

70

80

90

100

Not worse than 2 times slower than the best solver ()

Per

cent

age

(%)

APOPT+BPOPTAPOPT1.0

BPOPT1.0IPOPT3.10

IPOPT2.3

SNOPT6.1

MINOS5.5

Summary of 494 Benchmark Problems

Speed

Success

Speed and Successwith combined approach

Survey of DAE SolversSurvey of DAE Solvers

Software Package Max DAE Index

Form Adaptive Time Step

Sparse Partial‐DAEs

Simultaneous Estimation / Optimization

APMonitor 3+ Open No Yes Yes Yes

DASPK  / CVODE / Jacobian

2 Open Yes No No No

gProms 1 (3+ with transforms)

Open Yes Yes Yes No

MATLAB 1 Semi‐explicit

Yes No No No

Modelica 1 Open Yes Yes No No

DAE = Differential and Algebraic Equation

ConclusionsConclusions

Powerful insights can be gained from modeling and data reconciliation over long periods of historical data

When data, modeling, and optimization are combined, hidden savings are discovered through dynamic optimization

Simulation and optimization can give realistic options to evaluate risks and rewards

Simulation results can then be directly applied in practice to continuously monitor and optimize

AcknowledgmentsAcknowledgments

Extra SlidesExtra Slides