ASSESSMENT OF THE ERRORS OF VARIOUS METHODS OF STRUCTURAL MODELING OF RESERVOIR DEPOSITS
LAPKOVSKY V.V.1,2, KANAKOVA K.I.1,2, PONOMAREV M.S.2
Keywords: STRUCTURAL MODELING, ERROR MODELS, REFLECTING HORIZONS, SPLINE APPROXIMATION, COKRIGING, CROSS-VALIDATION
Abstract
The article presents an original methodology for constructing a structural model of an oil and gas field, focused on the consistent integration of drilling and seismic data represented in different domains—depth and time. To address the problem of reconciling heterogeneous data, the use of three-dimensional approximation methods—3D splines and joint cokriging—is proposed, within which drilling and seismic data are combined in a single variable space. This approach simplifies the workflow for structural model construction by eliminating the need for explicit generation of average velocity maps or the introduction of corrections to regression relationships. A comparison is also performed of the quality of structural models obtained using the proposed methods—3D spline approximation and joint cokriging—with results produced by traditional approaches, including linear regression and methods based on average velocity maps constructed in various ways. The comparison is based on error analysis using cross-validation techniques. It is shown that the best results are provided by the 3D spline method, which has not previously been applied to structural mapping problems. Comparable accuracy is demonstrated by the traditional approach based on linear regression followed by structural map correction using interpolation and adjustments. The results are demonstrated using the example of modeling the depth of the top of the Bazhenov Formation at the Kazan oil and gas condensate field. The high drilling density made it possible to assess the accuracy of different structural modeling methods under varying levels of well data availability, characteristic of different stages of field development.
|
