Evolutionary Algorithm Applied to Gas Lift Optimization in ... · PDF fileRafael G. Barroeta, Senior Consultant, SPT Group Carlos A Beltran, Senior Consultant, SPT Group. Kjetil Havre,

32nd Gas-Lift WorkshopThe Hague, The Netherlands

February 2 - 6, 2009

This presentation is the property of the author(s) and his/her/their company(ies).It may not be used for any purpose other than viewing by Workshop attendees without the expressed written permission of the author(s).

Evolutionary Algorithm Applied to Gas Lift Optimization in a Fully Dynamic Online Production Support System

Rafael G. Barroeta, Senior Consultant, SPT GroupCarlos A Beltran, Senior Consultant, SPT GroupKjetil Havre, Chief Scientist, SPT Group

Feb. 2 - 6, 2009 2009 Gas-Lift Workshop 2

Outline of presentation

•

Introduction•

Lobito Tomboco – Production Management System

•

The dynamic production model of Lobito Tomboco•

Gas-Lift Optimization based on steady state lift curves

•

MEPO Experimental Design and Optimization•

MEPO applied to OLGA on-line application

•

Results•

Conclusions


Introduction

Gas lift applications

•

Stabilization of casing heading and unstable wells with feedback control

•

Find minimum steady-state lift gas rate required for stable operation of a deep water riser or a well

0

2

4

6

8

10

12

0.0 0.5 1.0 1.5 2.0 2.5Gas lift rate (kg/s)

Oil

prod

uctio

n ra

te (k

g/s)

Oil production rate - Stable flow

Oil Production rate - Unstable flow

Region of optimum operation

•

Optimize steady-state oil production by allocating lift gas to different producers/wells

•

The Lobito Tomboco (LT) field is located in Block 14, offshore Angola (West African Coast). LT’s production is routed to the BBLT Drilling and Production platform which also receives production from the Benguela Belize field.

•

Block 14 is operated by Cabinda Gulf Oil Co. Ltd., a subsidiary of Chevron with 31%. The other partners include Agip Angola 20%, Sonangol 20%, Total Angola 20% and Portugal's Petrogal Exploration 9%.

•

LT is an oil-dominated system with a GOR ~ 650 SCF/STB and currently contributes the highest output of oil in Angola’s Block 14.

•

BBLT is Chevron’s largest offshore platform and is the fifth tallest freestanding structure in the world.

Lobito Tomboco

SPT’s involvement in Lobito Tomboco


Lobito Tomboco – Production Management System

•

The LT PMS provides operational support to:–

engineers

–

operators

of the Lobito Tomboco production system

•

The LT PMS is built to replicate the dynamic behavior of multiphase flow between the well perforations and the outlets of first stage separators


SW components in the LT PMS

Database

Web-based GUI

OPC Data Server

Field Instrumentation

APIS

Simulation Engine

1

2

3

4

5

OLGA


Lobito Tomboco production model Production Network

Subsea Center A

Benguela Belize DPP

WellsTA3P1TA3P2TA6P3TA6P4

Subsea Center C

Subsea Center B

10”

Production8”

Test12”

Water Injection6”

Gas Lift

WellsLB3P4LB3P5LB3P6LB3P7

InjectorsTA3I1TA3I2TA6I1TA6I2

InjectorsLB3I4LB3I5LB3I6

ProducersLDN1LC3P2LC3P3

InjectorsLC3I1LC3I2LC3I3LC3I4

Gas-Lift optimization based on steady state lift curves

•

Hl is total amount of gas available in each cluster.

•

Gas lift rates cannot be negative

•

The gas lift rate per well can be defined to be less than a certain value U (typically 4-6 MMscdf)

•

Assuming price (p) =1 and cost (k)=0

Basic Optimization Formulation:

N

igigioi kQQpQ

1)(max

NiQgi ,10 NiUQgi ,1,

N

ilgi HQ

1


1. Find gas lift performance curves for each well from off-line OLGA simulations. Curves paramterized as function of WHP and gas injection rate


Typical gas lift preformace curveFeb. 2 - 6, 2009 2009 Gas-Lift Workshop 10

2. Parameterize upper bound on gas injection per flowline as a function of differential pressure over the flowline


Typical Flowline Performance Curve


3. Use quadratic programming solver with GL objective function and well & flow line constraints


N

igigioi kQQpQ

1)(max

NiQgi ,10 NiUQgi ,1,

N

ilgi HQ

1


Main Conclusion:

The results of GLOT were not consistent with offline OLGA simulations of the complete flow network

Possible Reasons:

•

Simplification in the interaction of wells, i.e. interconnections of wells and flow lines

•

Use of a simplified version of the quadratic programming solver



MEPO – Experimental Design

•

Sensitivity analysis: Identify the most influential uncertainties.–

Traditionally the engineer has performed a one-parameter- at-a-time approach to:

•

Serve as screening purposes by producing Tornado diagrams•

Model checking–

In Experimental Design (ED) several parameters are varied simultaneously according to a predefined pattern. With this technique, more information than that obtained with the one-at-a-time method can be developed with fewer simulation runs.

–

The theory (ED) ensures that the parameter sets are constructed so that the maximum information can be obtained from a minimum number of simulation runs.


MEPO - History Matching/Optimization

•

The optimization methods of MEPO can be used for either history matching or optimization:–

MEPO’s optimizer searches for a minimum or a maximum of a quality function (objective function), i.e. it attempts to decrease or increase the value.

–

History Matching: In a history matching problem, the objective is to minimize the difference between observed and simulated data.

–

Optimization: A typical optimization problem would be to optimize an uncertainty variable (say well location or number of wells) by maximizing a parameter, say, maximize the recoverable oil reserves or an economic parameter.

•

The Objective function is the numerical description of the system to be optimised. It may be either a Mismatch parameter or a Watch parameter.


MEPO - Objective Function

where,

i: references an Objective element, e.g. the oil rate at a particular well

j: references the time step at which an observed value exists

s: defines simulated value at time step j in Objective element i

o: defines observed value at time step j in Objective element i w: is the corresponding weight factor

σ: is the standard deviation (the measurement error) of Objective element i

Objective functionA

Mismatch parameter is a function of the difference between the simulated value and the corresponding observed data i.e. typically used for history matching


Objective Function (cont.)

Objective functionA Watch parameter consist of a single numerical value specified at certain time e.g. cumulative field oil production at the end of field history or the corresponding NPV i.e. typically used for an optimizing process where the objective is to maximize e.g. the recoverable reserves


Evolution Strategy

•

Principles of biological evolution–

Mutation–

Recombination–

Selection

•

A class of Evolutionary Algorithms

•

Robust•

Applicable also for–

discontinuities–

non-linearities

of the search space


1 5 10

Qua

lity

of th

e M

atch

Generation (Iteration)

acceptable

unacceptable

φ

k

Example: (2+4) Evolution Strategy

Some acceptable models


Stagnation region

Optimal region

Iterations

Glob

al er

ror

Destabilization

Stagnation

Profile of Evolutionary Optimal Search•

Global optimization involves search of a large data space that may include local optima and stagnation regions that do not contain desirable solutions

•

The search can be enhanced by destabilization (manual or automatic) so as to redirect the search from stagnation regions

•

MEPO provides statistical criteria that helps detection of stagnated search and mechanism


MEPO workflow

OLGA


MEPO applied to edpm on-line application

MEPOcli


Uncertainty parameter definition


Uncertainty parameter definition


Watch parameter definition


Cycle preferences

We are not doing HM, this is project for oil maximization !!!


Running MEPOcli through edpm


Maximize oil production (ES 4,2)– Early life

Gas injection rates are reduced

Oil production is

increased

Gas lift is not required


Maximize oil production (ES 4,2)– Mid life

Feb. 2 - 6, 2009 2009 Gas-Lift Workshop 29Optimized values

Optimum

Trend curves from OLGA


Conclusions

•

A case was setup to evaluate the feasibility and accuracy of a MEPOcli-OLGA-edpm solution.

•

The results are by far more consistent with OLGA test cases than previous approaches based on Steady State lift-curves and QP -

solver

•

MEPO’s suggested rates result in a higher topsides oil rate

•

It is very promising as there is no evidence of other gas lift optimization tools that combine fully-transient online models with sophisticated optimization algorithms


Applications of OLGA on-line and MEPO

•

Optimization of well routing•

Optimization of subsea and topsides choke settings

•

Optimize steady-state gas lift rate required for stable operation of deep water riser or well

•

Parameter estimation for retuning of on-line applications with or without history matching

•

Engineering and experimental design to cover ranges of parameter variations –

present

feasible production envelope


Some ideas for further use of the OLGA - MEPO in flow assurance consulting

•

Off-line OLGA model tuning to field instrumentation including uncertainty in instrumentation, well tests, reservoir parameters

•

Use MEPO experimental design and OLGA to conduct studies and establish operational envelopes at different field life stages

•

Experimental design can simplify the work on several flow assurance studies that involve a lot of trial and error runs or simply a lot of simulations

•

MEPO enables sensitivity analysis with OLGA as an kernel in an automatic manner

•

MEPO reperesents simulation results in alternative ways


Thanks


CopyrightRights to this presentation are owned by the company(ies) and/or author(s) listed on the title page. By submitting this presentation to the Gas-Lift Workshop, they grant to the Workshop, the Artificial Lift Research and Development Council (ALRDC), and the American Society of Mechanical Engineers (ASME), rights to:

–

Display the presentation at the Workshop.–

Place it on the www.alrdc.com web site, with access to the site to be as directed by the Workshop Steering Committee.

–

Place it on a CD for distribution and/or sale as directed by the Workshop Steering Committee.

Other uses of this presentation are prohibited without the expressed written permission of the company(ies) and/or author(s) who own it and the Workshop Steering Committee.


DisclaimerThe following disclaimer shall be included as the last page of a Technical Presentation or Continuing Education Course. A similar disclaimer is included on the front page of the Gas-Lift Workshop Web Site.The Artificial Lift Research and Development Council and its officers and trustees, and the Gas-Lift Workshop Steering Committee members, and their supporting organizations and companies (here-in- after referred to as the Sponsoring Organizations), and the author(s) of this Technical Presentation or Continuing Education Training Course and their company(ies), provide this presentation and/or training material at the Gas-Lift Workshop "as is" without any warranty of any kind, express or implied, as to the accuracy of the information or the products or services referred to by any presenter (in so far as such warranties may be excluded under any relevant law) and these members and their companies will not be liable for unlawful actions and any losses or damage that may result from use of any presentation as a consequence of any inaccuracies in, or any omission from, the information which therein may be contained.The views, opinions, and conclusions expressed in these presentations and/or training materials are those of the author and not necessarily those of the Sponsoring Organizations. The author is solely responsible for the content of the materials.The Sponsoring Organizations cannot and do not warrant the accuracy of these documents beyond the source documents, although we do make every attempt to work from authoritative sources. The Sponsoring Organizations provide these presentations and/or training materials as a service. The Sponsoring Organizations make no representations or warranties, express or implied, with respect to the presentations and/or training materials, or any part thereof, including any warrantees of title, non- infringement of copyright or patent rights of others, merchantability, or fitness or suitability for any purpose.

Documents

Evolutionary Algorithm Applied to Gas Lift Optimization in ... · PDF fileRafael G. Barroeta, Senior Consultant, SPT Group Carlos A Beltran, Senior Consultant, SPT Group. Kjetil Havre,