基于双相介质理论的储层参数反演方法

传统基于单相介质理论的储层参数反演方法将孔隙流体与固体骨架等效为单一固体,弱化了孔隙流体的影响,反演结果精度不高.本文提出根据双相介质理论反演储层参数的方法.首先,在前人研究的基础上,利用岩石物理模型建立弹性参数与孔隙度、饱和度、泥质含量等储层参数间的关系,进而将双相介质反射系数推导为储层参数的函数;其次,根据贝叶斯反演理论,在高斯噪声假设的基础上,采用更加符合实际情况的修正柯西分布函数描述反射系数的稀疏性,推导出储层物性参数目标反演函数;最后,应用差分进化非线性全局寻优算法来求解目标反演函数,使得反演结果与实际资料间误差最小.新方法旨在突出流体对介质反射系数的影响,以期得到较高的储层参数反演精度.模型与实际资料测试均表明该方法可行、有效且反演精度较高.

Inversion of reservoir parameters based on dual-phase media theory

Traditional inversion methods of reservoir parameters based on the one-phase media theory regard the medium a single solid-phase body of solid skeleton and pore fluid, which may weaken the effect of pore fluid and lead to a lower credibility in inversion results. To make full use of fluid information in seismic amplitude and improve the precision of reservoir parameter prediction, this paper suggests a new method to invert reservoir parameters based on dual-phase media theory. Firstly, on the basis of previous work, a rock physics model is used to establish the relationship between the elastic parameter and reservoir parameter. Then a new dual-phase reflection coefficient equation is derived, which is the function of reservoir parameters. Secondly, according to Bayesian inversion theory and the Gaussian noise assumption, the modified Cauchy distribution is employed to describe the sparsity of reflection coefficients and deduce the final objective inversion function. Finally the differential evolution (DE) nonlinear global optimization algorithm is adopted to solve the objective inversion function and minimize the error between the inversion results and real data. From the reflection coefficient sensitivity analysis, the sensitivities to the reflection coefficient caused by change values of gas saturation, shale content and porosity are different. While fixing the values of shale content and porosity, we found that with the increase of gas saturation, the difference between reflection coefficient decreases. This trend is more obvious with the increasing incident angle. While fixing the values of gas saturation and porosity, the variation of reflection coefficients with low shale content values (less than 0.3) and under the condition of the incident angle less than 35° are obvious. With fixed values of shale content and gas saturation, the difference of reflection coefficients changing with porosity is more obvious than gas saturation and shale content, and there are still some difference even after 50°. From the objective function convergence analysis, the shape of convergence domain likes an ellipsoid. The porosity is easiest to converge to minimum and the shale content comes second. The convergence rate changes with reservoir parameters, but they all can converge to minimum. According to the analysis of the model, the DE algorithm used to solve the objective function is not sensitive to the initial model and is immune to noise to a certain degree. Even the initial model has great deviation and the signal-noise ratio is 5, our method still has a satisfactory result. We think that reservoir parameters are sensitive to the reflection coefficient and objective function can converge to a minimum in the range of actual incident angle. Using the dual-phase media theory to conduct reservoir parameter inversion is feasible and can achieve the optimal solution. Reservoir parameters can be quantitatively described as the reservoir characteristics. Results show that the inversion method based on dual-phase media theory can highlight the influence of fluid-phase and the method is feasible, effective and accurate. To get better actual application results, there are some problems need to be noted: (1) Inappropriate modeling results can cause a serious deviation to the relationship between reservoir parameters and seismic amplitude. Rock-physics experiments and statistical analysis can effectively ensure the accuracy of the modeling results. (2) High signal to noise ratio and amplitude-preserved seismic data can guarantee the accuracy of the inversion results. (3) The DE algorithm has weak dependence on the initial model, while using inversion results from traditional methods as the initial model can reduce iteration times and computing consumption of the DE algorithm. (4) High-performance computation equipment and high efficiency of forward and inversion algorithms can vigorously promote the application of the dual-phase media theory to large-scale 3D issues.