Automatic Parameter Fitting And Its Application In Shale Gas Reservoir Dynamic Analysis

Automatic parameter fitting is a commonly applied method in well testing analysis.It can be used to determine the reservoir characteristics and the parameter characteristics of wells,which is an important part of the analysis of reservoir models.As a complex gas reservoir,the shale gas has special internal structure,flow mechanism and development storage mode,which is different from the conventional reservoir.However,the correct judgment and interpretation of these complex structures requires rich practical experience,and it is more difficult to determine the parameters of the structures.The traditional fitting algorithm has the problem of low fitting speed and low accuracy.Therefore,in order to overcome the errors caused by the subjective factors and improve the fitting efficiency,intelligent optimization algorithm is applied to the analysis of model parameters.The algorithm is simple,with few adjustable parameters and fast convergence rate.It can improve the efficiency of the interpretation work without calculating the error gradient on the basis of random search in the parameter space.Application quality of the automatic fitting results is determined by measured data,well test model and algorithm,most important of these is algorithm.In this thesis,the automatic fitting algorithm and the application of the model have been studied.First,this thesis analyzes the basic principles of the intelligent optimization fitting algorithm,and studies the particle swarm optimization,simulated annealing algorithm and PSOSA(particle swarm optimization annealing algorithm).Through the study concluded that:?the PSO has the parallel global search ability,strong global convergence,but the local convergence is weak;?The simulated annealing algorithm has strong local search ability,but poor in the global;?The use of particle swarm algorithm can quickly converge to the global optimum,and the simulated annealing algorithm is used again to find the new optimal solution in the global optimal neighborhood,it can overcome the premature convergence of particle swarm optimization and use the advantages of the particle swarm algorithm with fast global convergence and the simulated annealing algorithm jump out from solution.Then,the homogeneous reservoir vertical well model and shale gas multilevel fracturing horizontal well model are analyzed.The mathematical expressions of two well test models are deduced by numerical calculation and Stehfest numerical inversion.Finally,the automatic fitting of well test models is realized by MATLAB programming.The fitting shows that:For homogeneous reservoir vertical well models,three kinds of fitting algorithms are applied,of which the best is hybrid PSOSA algorithm,and then is particle swarm optimization;For the complex seepage machine of shale gas reservoir,the particle swarm optimization algorithm can reach the fitting precision and the fitting effect is good.