Optimizer as a CAPE-OPEN Unit Operation - · PDF fileOptimizer as a CAPE-OPEN Unit Operation...

2007 AIChE Annual Meeting - Topical D: 4th CAPE-OPEN US Conference - Salt Lake City, Utah

©IF

P

Extended reserves | Clean refining | Fuel-efficient vehicles | Diversified fuels | Controlled CO2

Optimizer as a CAPE-OPENUnit Operation

2007 AIChE Annual Meeting – Topical D: 4th US CAPE-OPEN Conference

Authors:Martin Gainville, Pascal Roux: IFPAlain Vacher, Philippe Baudet: ProSim SAMichel Pons: Michel Pons Technologie SA (for TOTAL)Didier Paen: RSI - Dynamic Simulations Solutions

2007 AIChE Annual Meeting - Topical D: 4th CAPE-OPEN US Conference - Salt Lake City, Utah 2

©IF

P

Content

� Introduction

� Problem statement

� Optimizer integration and validation

� Case study

� Conclusion


©IF

P

Introduction Current situation

Riser

Riser base

PumaFlow

ECLIPSEREVEAL...

OLGA

LEDA

TACITE

PIPESIM, PIPEPHASE, GAP

...

PROSPER

WELLFLO

...

HYSYSINDISSD-SPICEPRO/IIUniSim...

Reservoir

Well Head


©IF

P

Introduction

� Design and operation of oil & gas production systems require more and more advanced process simulations� Multiple software solutions to address multiple knowledge domains

� For integrated simulations from reservoir to process, seamless integration of different provider solutions is needed

� Numerous issues to be resolved when coupling these solutions� Components interoperability� Data consistency

� Numerical consistency


©IF

P

Introduction

� To address these requirements, TOTAL and IFP have led a research collaborative project� TINA project based on INDISS Process Modelling Environment

� Purpose of the study� Extends modular strategy of INDISS for performing steady-state

optimizations� Using a 3rd party optimization module within INDISS architecture

� In a CAPE-OPEN interoperability framework� CAPE-OPEN PME: INDISS

� TINA pipeline modules are CAPE-OPEN compliant� Optimizer wrapped as a CAPE-OPEN unit operation to be easily

integrated within INDISS architecture


©IF

P

Problem statement� Network variables: x

� Unit operation parameters (ex: choke opening), pressure, flow rate, partial flow rate, ...

� Function to optimize: f(x)� Min of cost, max of Net Present Value, max of production, ...

� Equality constraints� N(y) = 0, y ⊂⊂⊂⊂ x, network equations

� Conservation equations, recycling equations, homogeneous pressure at nodes, ...

� Z(x) = 0, other specifications

� Inequality constraints� C(x) ≤ 0

� Lower bound ≤ x ≤ Upper bound


©IF

P

Problem statement� Solving mechanism: a sequential modular approach and

an "Unfeasible path" strategy

C(x), Z(x), N(y), f(x)

Optimizer

C(x) ≤≤≤≤ 0 ? Z(x) = 0 ? N(y) = 0 ?

Lower bound ≤≤≤≤ x ≤≤≤≤ Upper Bound ?

min f(x) ?

x for new estimate

or

x+dx for gradient estimate

Sequential modular calculation with a specific calculation order


©IF

P

Optimizer integration� OPTI module: optimizer

� Native optimizer of ProSimPlus PME (ProSim SA)� Applied to convergence of NLP optimization problems

� Designed for a sequential modular approach

� Unfeasible path strategy is recommended

� Jacobian Matrix is estimated by numerical differentiation� Physical models are complex and implicit therefore analytical

derivatives are not available

� OPTI wrapped as a CAPE-OPEN unit operation (CO-OPTI)� Specific parameter to check the optimizer convergence� Data transfer using CAPE-OPEN information streams


©IF

P

Optimizer integration� INDISS PME has been adapted: INDISS-TINA

� To check the convergence of the optimizer� To manage tear streams solved by CO-OPTI� To establish convenient calculation order

� CO-OPTI made last UO in the maximum recycle sequence

� CO-OPTI ICapeUnit.Calculate( ) method� Retrieves values of constraints and objective function through

information streams� Transfers data to OPTI that gives new iterative values in return� Updates CO-OPTI outputs variables with the new iterative values

� Including recycle material streams which are flashed

� Updates the convergence status (flag parameter)


©IF

P

Validation on a simple linear problem� Within ProSimPlus PME

� ProSimPlus is a CAPE-OPEN compliant PME� Native OPTI module in ProSimPlus and CO-OPTI used as CAPE-

OPEN UO in ProSimPlus give identical results

� Within ProSimPlus and INDISS using CO-OPTI as a CO UO� Convergence path and final results are identical in both PMEs


©IF

P

Case study

( )γβα ++= DDLC Line2

CPumpPump WBAC +=

PQW Pump ∆=

� Cost optimization of a sub-sea production system� Target function: minimize the design cost of the system� Cost of pipe installation + cost of multiphase pump

� Optimization parameters� Line diameter� Pump useful power (or differential pressure)

� Equality constraint (operational constraint)� Topside arrival pressure equal to 15 bars

� Inequality constraints� Maximum Pump useful power: 1.2 MW� Line diameter: from 8 to 12 inches


©IF

P

Case study� Demo: CO-OPTI within INDISS-TINA


©IF

P

8

9

10

11

12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15Iterations

Line

dia

met

er (i

nch)

0

0.2

0.4

0.6

0.8

1

1.2

Use

full

pum

p po

wer

(MW

)

DIAMETER POWER

Case study� Results

CONSTRAINTS VIOLATION

0.00001

0.0001

0.001

0.01

0.1

1

10

100

1000

10000

0 2 4 6 8 10 12 14 16

Iterations

OBJECTIVE FUNCTION

3.54E+07

3.56E+07

3.58E+07

3.60E+07

3.62E+07

3.64E+07

3.66E+07

0 2 4 6 8 10 12 14 16

Iterations

Upper bounds

Lower bounds


©IF

P

Conclusion

� CO-OPTI integrated and used successfully in INDISS as a CAPE-OPEN Unit Operation� Use of CAPE-OPEN specifications has reduced development time for

integrating CO-OPTI into INDISS-TINA� CO-OPTI easily plugged into INDISS-TINA PME

� Recycle material streams easily handled� CAPE-OPEN information streams useful for data transfer

� CAPE-OPEN enables optimizer integration in sequential modular PMEs


©IF

P

Thank you for your attention

� Acknowledgements to TOTAL, RSI, ProSim SA for their contribution to this work

� Questions?

� Additional explanations will be provided tomorrow morning during the CAPE-OPEN Interoperability Workshop session


©IF

P

Case study� Sensitivity study to the transportation distance

� Optimizations are performed for several distances� 5, 10, 15, 20 and 25 km

� Three design options� Option 1: system designed with the maximum specified pipe

diameter (Pipe diameter of 12 inches), � Option 2: system designed with the maximum pump unit

capacity (Pump unit of 1.2 MW),� Option 3: optimum solution calculated by CO-OPTI


©IF

P

Case study� Normalized overall cost

0.5

0.6

0.7

0.8

0.9

1

0 5 10 15 20 25 30Tie-back distance [km]

No

rmal

ized

ove

rall

cost

Option 1 - Pipe diameter of 12 inches Option 2 - Pump unit of 1.2 MW

Option 3 - 0ptimum solution calculated by CO-OPTI


©IF

P

Case study� Over cost in percentage of the optimum cost

0

2

4

6

8

10

12

0 5 10 15 20 25 30Tie-back distance [km]

Co

st b

enef

it in

per

cen

tag

e o

f th

e o

pti

mu

m o

vera

ll co

st [

%]

Option 1 - Pipe diameter of 12 inches Option 2 - Pump unit of 1.2 MW


©IF

P

Optimizer integration� Schematic view

� Calculation order� {UO1, UO2, UO3, Evaluation module, UO5, UO4, CO-OPTI}

ICapeUtilities.Edit

Optimizer as a CAPE-OPEN Unit Operation - · PDF fileOptimizer as a CAPE-OPEN Unit Operation...

Documents

Transcript of Optimizer as a CAPE-OPEN Unit Operation - · PDF fileOptimizer as a CAPE-OPEN Unit Operation...