Optimization Of Large-scale Multi-source Water Injection System Based On Intelligent Computing

Water injection system is the main power system for oilfield production,and it is also a high-energy-consumption and high-investment pipe network system.How to realize the energy-saving operation and economic construction of the water injection system is an urgent problem that needs to be solved in major oilfields.Multi-source water injection system is a complex fluid network system with complex structure,large number of nodes and wide coverage area.In the established system the operation status of each water injection station is mutually restricted by the communication network,and in the new system the location planning of the water injection station and water distributing station is coupled with the direction of the pipeline.Especially for thousands of large-scale water injection systems,it is very difficult to carry out operation management and planning design.At present,the operation management and construction scheme design of oilfield water injection system rely on manual experience,and it is hard to achieve high-efficiency comparison and optimization of multiple schemes,making it difficult to ensure the optimality of decision-making schemes,which is easy to cause waste of energy and investment.In this paper taking large-scale multi-source water injection system as the research object,the research on hydraulic analysis,operation optimization and layout optimization is carried out and the specific research contents are as follows.Firstly,based on the analysis of the network structure characteristics of water injection system,combined with parallel search idea and breadth-first search algorithm(BFS),a bidirectional BFS algorithm for base loop is proposed.The storage structure of the algorithm is optimized and its optimality is proved.Through the complexity analysis and case calculation,the algorithm has significant advantages in terms of computational efficiency compared with the classical methods for base loop solving such as naive BFS algorithm,adjacency matrix transformation method and cotree method.With the minimum total friction loss of the pipeline as the objective function,an optimization model of initial flow distribution in hydraulic analysis is established,and is solved by Lagrange multiplier method.Based on the above,combining Hardy-Cross method the hydraulic analysis calculation method of the large-scale multi-source loop-tree water injection system is formed.Secondly,with the minimum operating unit consumption as the objective function,the optimization model of operation parameters for large-scale multi-source water injection system is established.Combining the local mining ability of particle swarm optimization and the global exploring ability of fireworks algorithm,the fireworks operator is embedded into the process framework of particle swarm optimization algorithm,and the abandonment-supplement operator,the modified explosion operator and the novel mutation operator are constructed.Then the hybrid particle swarm optimization-fireworks(PS-FW)algorithm is proposed.Through the numerical experiments,the performance of PS-FW algorithm and other 8 excellent survival algorithms are compared and the influence of the main control parameters for PS-FW algorithm on the performance is analyzed.Based on Poincare cycle theorem and measure method,the convergence theorem of stochastic optimization algorithm is proposed,and the global convergence of PS-FW algorithm is proved.Combining PS-FW algorithm and hydraulic analysis calculation method,the optimization model of operation parameters is solved efficiently.Thirdly,considering the interaction between multi-source supply in the water injection system,taking the minimum unit consumption of the system as the objective function,the flow restriction,pressure limitation,and pipe flow characteristics as the constraints,and the start-stop state and operation displacement of water injection pump as decision variables,the mathematical model for the operation schemes optimization of large-scale multi-source water injection system is established.Through the relaxation transformation and the derivation of the constraint conditions,the dynamic programming method for calculating the number range of feasible open pumps is proposed,and the branch and bound method for optimizing the initial feasible scheme is formed.On the basis of narrowing the feasible solution space by using the bounded idea,combined with the PS-FW algorithm and hierarchical optimization idea,the hybrid hierarchical-boundary-particle swarm-fireworks solving method is constructed,which realizes the efficient and accurate solution of the optimization model.Fourthly,based on the proposed obstacle implicit characterization method,vector product method and bounding box simplification method,an obstacle-bypassing pipeline routing optimization method is formed.Considering the length of water injection pipeline and the connectivity of pipeline network,the reliability calculation method of the water injection system is established.Making the maximum reliability and minimum construction cost as objective functions,a mathematical model for obstacle layout optimization of large-scale multi-source water injection system is established.An adaptive Cauchy perturbation operator,a Gauss mutation operator and a probability transfer operator are constructed.An improved particle swarm optimization algorithm(MPSO)is proposed.Its global convergence is proved and its excellent high-dimensional optimization ability is verified by numerical experiments.Based on the dimension reduction planning and modularization idea,the grid dissection set partition method is proposed,and its theoretical basis and complexity are analyzed.In order to reduce the dimension of the problem,the fuzzy set of adjacent position method is proposed and its optimality proof for solving the membership relationship of the injection well is given.Combing the advantages of the MPSO algorithm,the grid partition method and the fuzzy set of adjacent position method,the combined optimization solution strategy is presented.Finally,based on the research results obtained,using C# program development language,relying on Visual Studio 2010 program development platform and component GIS development technology,the intelligent decision-making system software for large-scale multi-source water injection system is developed.Its main functional modules include file management,production data management,graphical modeling,water injection system hydraulic analysis,operation parameters optimization,operation scheme optimization and layout optimization.It can realize the calculation and decision-making functions of water injection system,such as hydraulic analysis,operation optimization and layout optimization,which can provide scientific operation and planning scheme for oilfield production and construction.