Abstract：
Taking the dense distribution area of horizontal wells in the Sulige Gas Field as an example, the superiority and paradox of horizontal well data in 3D geological modeling are analyzed in this paper. For the characteristics of centralized information and large quantity of information, the horizontal wells data have advantage in improving the accuracy of inter-well prediction and characterizing the regularity of geological bodies. However, a clustering effect occurs because of the strong directional and selective drilling of sand bodies in horizontal well data. As a result, the variogram based on the horizontal well data is not consistent with the geological knowledge. In this paper we present two practical approaches of horizontal well data application in 3D geological modeling. Based on the spatial distribution characteristics of the variable range ellipsoid, the range ellipse calculation method is presented. In this method, two different ranges of the horizontal variogram are obtained by horizontal well data in two different directions individually while the vertical range of variogram is obtained by vertical well data. So we can get the variogram with high precision by calculation. However, the application of this method is limited to areas with more than two directions of horizontal wells. The other is the horizontal well data rarefying method. To eliminate the clustering effect, the horizontal well data are rarefied regularly and uniformly multiple times. And then we can get multiple variograms by rarefying the horizontal well data to calculate the variogram of the study area by statistics. Compared to the range ellipse calculation method, the horizontal well data rarefying method has a lower precision and can be applied to the area with
single direction of horizontal wells.
Key words：horizontal wells; variogram; geological modeling; Sulige Gas Field
Corresponding Authors： liulin8812@163.com
Cite this article：LIU Lin. 3D geological modeling based on horizontal well data. Petroleum Science Bulletin, 2020, 01: 17-25.
