Anda di halaman 1dari 2

PRODUCTION TECHNOLOGY

Evolutionary algorithms deliver promising results to gas lift optimization problems


A 243 bopd increase on a 56-well problem is seen in computer simulation when an evolutionary algorithm is used with gas lift optimization.
Dr. Tapabrata Ray and Dr. Ruhul Sarker, University of New South Wales Evolutionary algorithms are a class of stochastic optimization algorithms inspired by the process of natural evolution. They are suitable for complex, real life optimization problems that have so far been considered intractable. Gas lift is known to be one of the most economic methods for artificially lifting oil.1 This article reports the application of an evolutionary algorithm to find optimal gas injection volumes for oil extraction subject to constraints on gas availability, oil extraction targets and well characteristics. This article also provides insights on the benefits offered by our multi-objective optimization approach, which simultaneously minimizes gas injection while maximizing oil extraction. Results show significant improvement over existing practices for two standard benchmarks of six-well and 56-well problems. For the six-well problem, the algorithm delivered a solution with 35 bopd more12,775 bbl improvement on a yearly basisand for the 56-well problem our solution was better off by 243 bopd88,695 bbl improvement on a yearly basisthan results previously reported in literature. ing complex nonlinear multi-objective optimization problems where the objective and constraint functions are nonlinear and/or discontinuous, and the users are interested in obtaining a set of tradeoff solutions. In practice, it is common to collect production data at discrete points for each welli.e., amount of gas injected and its corresponding volume of oil extractionand use a piece-wise linear3 or a quadratic function4 to capture their relationship. For a typical well, the volume of oil extraction increases with the amount of gas injected up to a certain level and then decreases. Such behaviori.e., the limit of safe gas injection/ oil extraction volumesis well specific. As a result, an inappropriate gas allocation to different wells, under limited gas availability, will reduce the overall oil production volume and, hence, profitability from the reservoir.
METHODOLOGY The above problem can be formulated as a single objective optimization problem such as maximizing oil extraction volume subject to the constraints on total gas availability and limits of safe gas injection volumes into each well. Mathematically, the problem can be represented as:

F O O R P
availability, and GiL , GiU denote the lower and upper bounds on gas injection volumes into the ith well. For simplicity, we have avoided the functional relationship between Oi and Gi which is nonlinear. However, the interested readers may find the mathematical relationship in work previously published by Ray and Sarker.5 Since the total amount of gas available is likely to vary on a daily basis, the problem can also be formulated as a bi-objective optimization problem as follows: Maximize Minimize and Maximize Subject to and Where, Oi denotes the volume of oil extracted from the ith well, Gi refers to the amount of gas injected to the ith well, N is the number of wells in the reservoir, GMax represents the total gas Subject to the same set of constrains listed above. We have developed an evolutionary algorithm to solve the above two mathematical programming models. This algorithm is a variant of Nondominated Sorting Genetic Algorithm (NSGAII)6,7 and has a major difference in the process of population reduction. In the process of population reduction, our method insists on maintaining not only the end points of the objective space but also maximum and minimum values of the variables. Our process is certainly more computationally expensive than NSGA-II and can be thought of as a diversity maintaining mechanism which might be useful for problems where the diversity in the variable space is important. The complete details of the algorithm are outlined in Ray and Sarkers previous work.5 Further details on multi-objective optimization and evolutionary multi-objective algorithms can be found in a survey paper by Coello Coello.8
World Oil APRIL 2008
1

BACKGROUND Oil is a finite and scarce resource which needs to be produced optimally. Evolutionary Algorithms (EA) are one of the frontrunners for optimization methods and tools and have been used extensively in recent years to solve a number of real life optimization problems.2 Fundamentally, EAs are a class of stochastic optimization methods inspired by nature, where a set of solutions undergo improvements over iterationsoften referred to as generations. They are particularly attractive for solv-

PRODUCTION TECHNOLOGY

CASE STUDIES oil extraction versus gas usage 5.0 The case studies include data to be approximated using 4.5 a six-well problem and a nonlinear functions to repre56-well problem presented in sent real life instances more 4.0 Buitrago et al.9 The gas injecclosely. Additional constraints 3.5 tion volume versus oil extracor objectives can be handled 3.0 tion data is interpolated using seamlessly using the proposed a piecewise linear function. approach. WO 2.5 The maximum injection volB A 2.0 umes into each well are limLITERATURE CITED 1 Ayatollahi, S., Narimani. M. and M. Moshfeghian, ited to 6.00 MMcf of gas. To 1.5 Intermittent gas lift in Aghajari Oil Field, a mathestablish that the EA could ematical study, Journal of Petroleum Science and En1.0 gineering, 42, 2004, pp. 245-255. consistently deliver accept2 Goldberg, D., Genetic Algorithms: In Search, Optimi0.5 sation and Machine Learning. Addison Wesely, 1989. able solutions, 96 indepen3 Kosmidis, V., Perkins, J. and E. Pistikopoulos, A dent runs were conducted for mixed integer optimization formulation for the 0.0 well scheduling problem on petroleum fields, each problem with the follow0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Computers and Chemical Engineering, 29, 2005, pp. Oil extraction, bpd ing set of parameters: popula1523-1541. 4 Camponogara, E. and P. Nakashima, Optimization tion size of 100; generations Fig. 1. Solutions to the multi-objective formulation of the sixof lift-gas allocation using dynamic programming, IEEE Transactions on Systems, Man and Cybernetics of 100 and 200; probability well problem. Part A: Systems and Humans, 36, 2006, pp. 407-414. of crossover of 0.7 and 0.9; 5 Ray, T. and R. Sarker, Genetic algorithm for solving a gas lift optimization problem, Journal of Petroleum probability of mutation of 0.1 Science and Engineering, Vol. 59, 2007, pp. 84-96. 6 and 0.2; distribution index for Table 1. Performance of EA on benchmark Deb. K., Pratap, A., Agarwal, S. and T. Meyarivan, A fast and elitist multi-objective genetic algorithm: crossover as 10 and 20; distri- problems NSGAII, IEEE Transaction on Evolutionary Computation, 6, 2002, pp. 181-197. bution index of mutation as 7 Deb. K., Pratap, A., Agarwal, S. and T. Meyarivan, 6 Well 56 Well 10 and 20; and finally with A fast elitist nondominated sorting genetic algorithm for multi-objective optimization: NSGAII, Buitrago et al. 3,629.00 bpd Buitrago et al. 21,789.9 bpd random seeds of 0.2, 0.4 and Proceedings of the Parallel Problem Solving from Nature Best EA 3,663.99 bpd Best EA 22,033.4 bpd 0.6. The average CPU time VI Conference, Paris, France, 2000, pp. 849-858. 8 Coello Coello, C., A comprehensive survey of evoWorst EA 3,653.90 bpd Worst EA 21,222.4 bpd for an optimization run with lutionary-based multi-objective optimization techAverage EA 3,660.20 bpd Average EA 21,622.3 bpd niques, Knowledge and Information Systems, 1, No. a population size of 100 and Median EA 3,660.77 bpd Median EA 21,651.2 bpd 3, 1999, pp. 269-308. 9 Buitrago, S., Rodriguez, E. and D. Espin, Global a generation of 100, which optimization techniques in gas allocation for concorresponds to evaluation of tinuous flow gas lift systems, SPE35616, presented at the SPE Gas 10,000 candidate solutions, is around single-objective optimization problems Technology Conference, Calgary, Canada, 1996. 10 3.04 seconds on a desktop computer on a daily basis with differing total gas LINGO, 1999. Lingo System Inc., http://www.lindo.com (IBM ThinkCentre: Intel Pentium 4, availability figures. It is clear from Fig. 1 2.8 GHz, 1 GB RAM). Source Codes that point Ai.e., a 2,500-bpd extracwere complied using Visual C++ Ver- tion with 2.00 MMcfis possible to sion 6.0, Enterprise Edition. achieve whereas point B, which requires For the six-well problem, our best re- 3,500-bpd extraction with 2.25 MMcf, The authors sult is around 35 bpd better than those is not achievable. If we need to deliver Tapabrata Ray obtained reported by Buitrago et al. and can be 3,500 bpd, a minimum of 3.50 MMcf his PhD in Ocean Engiseen in Table 1. Also, for the 56-well of gas would be necessary. neering and Naval Arproblem, our best result is around chitecture from Indian 253 bpd better than those reported by CONCLUSIONS Institute of Technology Kharagpur. He is currently Buitrago et al. The corresponding gas In this article we have introduced a a senior lecturer with the injection volumes are 0.475524996, practical gas lift optimization problem School of Aerospace, Civil 0.743445098, 1.350921920, and solved the single and multi-objective and Mechanical Engineer0.827749079, 1.199584681 and versions of the problem using our multiing, University of New 0.000212049 MMcf, respectively. Since objective evolutionary algorithm. For South Wales, Canberra, Australia. He has pubmore than 80 technical papers in the area a piecewise linear model has been used both the six-well and the 56-well prob- lished of multidisciplinary design optimization, robust in the studies by Buitrago et al., as well lems, our method yeilded better results design and surrogate modeling. as our present EA, the results can be than that of previous reports. We have verified using a linear programming tool also extended our study to include the Ruhul Sarker obtained PhD in Operations such as LINGO.10 A comparison with results of multi-objective formulations his Research from DalTech LINGO results indicates that our EA is of the problem to highlight some of the (former TUNS), Dalhousie able to locate optimal solutions to the additional benefits that can be derived University, Canada. He is six-well problem with 99% accuracy. using our multi-objective evolutionary currently a senior lecturer the School of IT&EE, The results of our multi-objective algorithm. The consistency and the effi- in University of New South formulation are shown in Fig. 1, which ciency of our algorithm have been dem- Wales, Canberra, Australia. is particularly useful for planning, as onstrated through multiple runs. Our His main research areas it provides direct information on the multi-objective approach is attractive as are applied operations research and evolutionoptimization. He has published more than minimum amount of gas required to it eliminates the need to solve gas lift ary 140 technical papers. He is the lead author of achieve any given oil extraction tar- optimization problems on a daily basis. the book Optimization Modeling: A Practical Apget, and eliminates the need to resolve Our surrogate-assisted approach allows proach published by Taylor and Francis.
2 APRIL 2008 World Oil

F O O R P

Gas used, MMcf

Anda mungkin juga menyukai