Publications

Publications / SAND Report

Geomechanical Simulation of Big Hill Strategic Petroleum Reserve - Calibration of Model Containing Shear Zone

Park, Byoung P.

A finite element numerical analysis model , that consists of a realistic mesh capturing the geometries of Big Hill (BH) Strategic Petroleum Reserve (SPR) site using the multi - mechanism deformation (M - D) salt constitutive model and including data taken daily of the wellhead pressure and level of the oil - brine interface , has been upgraded . The upgraded model contains the shear zone to examine the interbed behavior in a realistic manner. The salt creep rate is not uniform in the salt dome, and creep test data for BH salt is limited. Therefore, a model calibration is necessary to simulate the geomechanical behavior of the salt dome. Cavern volumetric closures of SPR caverns calculated from sonar survey reports are used for the field baseline measurement. The structure factor, A 2 , and transient strain lim it factor, K 0 , in the M - D constitutive model are used for model calibration. An A 2 value obtained experimentally from the BH salt and K 0 value of WIPP salt are used as the baseline values. To adjust the magnitude of A 2 and K 0 , multiplication factors A2F an d K0F are defined, respectively. The A2F and K0F values of the salt dome and salt drawdown layer of elements surrounding each SPR cavern have been determined through a number of back fitting analyses. The trendlines of the predictions and sonar data match up well for BH 101, 103, 104, 106, 110, 111, 112, and 113. The prediction curves are close to the sonar data for BH 102 and 114. However, the prediction curves for BH 105, 107, 108, and 109 are not close to the sonar data. An inconsistency was found in the sonar data, i.e. the volume measured later is larger than that before in some time intervals , even if the leached volume is taken into account , for BH 101, 104, 106, 107, and 112. Project discussions are needed to determine possibilities on how to resolve the issues and determine the best path forward for future computer modeling attempts.