Improved MINLP Algorithms And Application In Production Scheduling For Oilfield Development

Oil has become indispensable important energy and industrial raw materials in the world economic development. Under the decline trend of global oil reserves, scientific management of reservoir development is particularly important. The scheduling mathematical models of oil field production process are mostly MINLP, including a large number of discrete, continuous variables and nonlinear functions. Complex model and problem scale make that MINLP optimal algorithm be the research focus and difficulty for scholars all over the world.Based on the Generalized Benders Decomposition (GBD) and Outer Approximation algorithm (OA), this work propose hybrid improved MINLP algorithm, and verify the feasibility and validity of the algorithm through numerical experiments. We formulate the mathematical modeling for scheduling problem of oilfield development, and apply the improved algorithm to solve the model. Finally, the decision support system for oilfield production scheduling is developed with the embedded optimization model and algorithm.1 Based on comparing numerical experiments of Benchmark MINLP problems, we summarize the computational performance for OA, GBD and their improved algorithm. Analysising GBD failure reasons with algorithm theory, the improved strategy is put forward, and applied to solve complex MINLP cases. There are certain effect for improved method.2 Based on the OA and GBD algorithm, hybrid cuts algorithm is proposed, combined with OA and GBD cut. Through introducing partial surrogate cut, partial surrogate cut algorithm is presented. Numerical experiments on benchmark problems show the feasibility and efficiency of the algorithm.3 Aimming at the production of a large oilfield development process, production scheduling problem of oil field is presented, and formulated the MINLP mathematical model. The improved algorithm is applied to solving the model. Comparing numerical experiment results show that the improved algorithm can significantly improve the solving efficiency at CPU times and the number of iterations.4 Taking the production management of oilfield development as the background, this work developments the decision support system for scheduling optimization. We embed the presented mathematical model and algorithm in the system. This system can conveniently manage production process data, and supply production optimazation scheduling solution for decision makers in the oilfield.There are theoretical research value of MINLP optimal algorithm for numerical experiments conclusions. The presented MINLP algorithm significantly improves the solving efficiency for oilfield production scheduling optimization, which is well application to solve MINLP instances from real production. The developed optimization decision support system for oilfield production rapidly provides scientific production management solutions for oilfield managers, integrating data and model algorithm.