Abstract:

There are certain correlations among formation lithology, physical properties, hydrocarbon potential and electrical property (logging curve). Different logging curves often reflect different formation properties. It is difficult to describe the mathematical relationship among these curves through one single function. Multiple regression technique is a useful tool for dealing with the correlation among variables. At present, it has been applied widely. Multiple linear regression model is often used in various literature to describe the relationship of formation electrical property. Although it may acquire effective results for analyzing simple formation characteristic relationship, the effects of this model are often unsatisfactory due to the increasing of various geological influencing factors. Therefore, on the basis of conventional linear multiple regression model, nonlinear multiple regression models were established in this paper, and the complex nonlinear problem was simplified to a linear problem which is relatively simple to handle, which enhances the feasibility of multivariate regression technique. Meanwhile, optimal methods to select multiple regression parameters and models were established, which provides technical support for obtaining the optimum regression results. At last, the concrete meaning of log reconstruction technology in geology and engineering operation was discussed, and the feasibility is proved by citing three application cases.

Key words: natural gamma-ray, shale content, logging interpretation, model reconstruction, reservoir inversion