ADVERTISEMENT

Parallel Simulation and Cloud Computing Can Optimize Large-Scale Field Development

You have access to this full article to experience the outstanding content available to SPE members and JPT subscribers.

To ensure continued access to JPT's content, please Sign In, JOIN SPE, or Subscribe to JPT

Field-development optimization is challenging because of the large number of control parameters, model complexity, and subsurface uncertainties. The complete paper discusses a study in which the authors propose a joint field-development and well-control-optimization work flow using high-performance parallel simulation and commercial cloud computing, and demonstrate its application through an offshore oilfield development.

Introduction

Designing a field-development plan to maximize the profitability of the asset involves accounting for many factors. Commonly, a field-development plan follows a staged process in which several alternatives are appraised and the one with the best economics is selected. The selected plan is then optimized while considering schedule and cost. Earth modeling and reservoir simulation are often used to provide probabilistic ranges of oil in place or net recovery for a given design at each phase of the process. This requires effective cross-functional teamwork to achieve quality decisions by relying on knowledge and skills spanning different disciplines.

Proposed Work Flow

Closed-loop or real-time reservoir management are terms often used to describe field-development optimization. These involve rapid data assimilation to define or update subsurface uncertainties and follow optimization processes to find the optimal combination of control parameters. These processes are not single-step but are executed iteratively during reservoir life. The focus of this paper is to describe the optimization process for maximizing economical values by selecting the best development plan under given ranges of subsurface uncertainties.

The work flow proposed by the authors controls topside facilities, the number of wells, their trajectories, the drilling sequence, and completion strategy simultaneously, while considering subsurface uncertainties and constraints. The authors use a next-generation reservoir simulator and commercial cloud computing to explore the possibility of achieving an optimized development scenario within reasonable time and cost constraints. The combination of high-performance simulators and virtually unlimited scalability in the cloud provides opportunities for joint field-development optimization without reducing the number of control parameters, enabling the developed work flow to be applicable to various asset-development scenarios.

The proposed work flow was applied to the Olympus field case, which is an optimization benchmarking problem developed by the Netherlands Organization for Applied Scientific Research using a synthetic North Sea-type reservoir. The study objective was to improve the net present value (NPV) after 20 years of operation by controlling the number and location of platforms and the number of injectors and producers as well as their trajectories and drilling sequence.

The large number of control parameters and subsurface uncertainties make the optimization process challenging. Three optimization techniques—genetic algorithm (GA), particle swarm optimization (PSO), and ensemble-based optimization (EnOpt)—were tested and their performances compared. Best results, in terms of NPV improvement, were obtained by using the mixed-integer genetic algorithm method.

More than 10,000 simulation runs were required by the method to reach optimal development of well location, trajectory, drilling sequence, and other aspects. The high-performance parallel simulator and cloud computing made this possible. The estimated cost of the commercial cloud service is almost negligible compared with the improvement in the economic value of the optimized asset-development plan. The developed work flow and parameterization technique are flexible in well-trajectory configuration and completion design, allowing application to primary depletion as well as waterflooding.

The complete paper first describes the optimization problem for an offshore asset by defining control parameters, constraints, economic values, and objectives. Next, the authors present the proposed field-development optimization work flow including base design setup, parameterization of the decision variables, and feasibility of the joint optimization of all development design and well-­control parameters. Three optimizers were tested and evaluated on the basis of their efficiencies in solving the presented optimization problem. Finally, the authors summarize their learnings by showing the performance of the proposed work flow to achieve the selected optimized design. The application of the proposed work flow under subsurface uncertainties enabled finding an optimized development plan maximizing an NPV S-curve with no risk of negative value. A final optimization test involving thousands of simulation runs demonstrated the efficiency of using computational power. Numerous charts, plots, screen shots, and tables are used to illustrate the objectives, constraints, work flow, and results.

Conclusions

  • The topside facility, number of injectors and producers, well trajectory, completion type, drilling sequence, and well control are simultaneously optimized, resulting in an improved NPV S-curve with reasonable field-development design.
  • The proposed work flow has flexibility and generality to apply varieties of asset-development planning with customization in input for economic values, well-location pattern, completion design, and rig scheduling.
  • Sensitivity analysis was demonstrated to find parameter interactions to the objective function, and identified parameters that can be optimized sequentially.
  • Performance and efficiencies of the three optimizers were evaluated, among which GA was shown to be the most-effective and -robust technique for this case study (Fig. 1).
  • Effect of subsurface uncertainty in field-development optimization has been addressed, and a sequential uncertainty refinement method was proposed, resulting in a robust field-development design that includes the number of model realizations.
  • The study demonstrated the viability of solving a complex, large-scale field-optimization problem using commercially available computer resources.
  • The challenge remains to reduce the computational load because multimillion grids with complex physics-simulation models require significant machine resources.
Fig. 1—Comparisons of optimization-technique performance with a P50 oil-in-place simulation model.

 

This article, written by JPT Technology Editor Judy Feder, contains highlights of paper SPE 191728, “Large-Scale Field Development Optimization Using High-Performance Parallel Simulation and Cloud Computing Technology,” by Shusei Tanaka, SPE, Zhenzhen Wang, SPE, Kaveh Dehghani, SPE, Jincong He, SPE, Baskar Velusamy, and Xian-Huan Wen, SPE, Chevron, prepared for the 2018 SPE Annual Technology Conference and Exhibition, Dallas, 24–26 September. The paper has not been peer reviewed.

Parallel Simulation and Cloud Computing Can Optimize Large-Scale Field Development

01 September 2019

Volume: 71 | Issue: 9

No editorial available

ADVERTISEMENT


STAY CONNECTED

Don't miss out on the latest technology delivered to your email weekly.  Sign up for the JPT newsletter.  If you are not logged in, you will receive a confirmation email that you will need to click on to confirm you want to receive the newsletter.

 

ADVERTISEMENT

No editorial available

ADVERTISEMENT

ADVERTISEMENT

ADVERTISEMENT