Interpretation of partitioning interwell tracer tests using EnKF with coarse-scale constraints

The success of enhanced recovery processes depends largely on the accurate characterization of the location and distribution of the target oil. The Partitioning Interwell Tracer Test (PITT) is designed to identify the residual oil saturation during waterflooding. In this test a conservative and a partitioning tracer are injected into the reservoir. The conservative tracer travels with the water phase whereas the partitioning tracer dissolves in both water and oil. The separation between the conservative and the partitioning tracer at the producer can be used to infer the residual oil saturation distribution. The interpretation of partitioning tracer tests in the presence of mobile oil is considerably more difficult because the separation of the tracer response depends not only on the partitioning coefficients but also on the spatial distribution of the saturation itself. We propose a two-stage Ensemble Kalman Filter (EnKF) that utilizes the tracer data in conjunction with coarse-scale saturation constraints to identify the fine-scale saturation distribution and associated uncertainties. The coarse-scale saturation distribution can be conveniently obtained via an inversion of the waterflood response, for example water-cut data at the producers. The tracer data are first used to condition the ensemble of fine scale model realizations which are further improved via the coarse-scale constraint. The scale-decomposition of the inverse problem significantly improves the performance of the EnKF and prevents filter divergence as more data are assimilated. We demonstrate the effectiveness of our approach using two and three-dimensional examples. First, we derive the coarse-scale saturation from a streamline-based inversion of the waterflood response. Next, the two-stage EnKF is used to interpret the PITT using the coarse-scale saturation constraint. A performance comparison of our proposed approach with the standard EnKF clearly indicates better characterization of the reservoir heterogeneities and improved estimates of the spatial distribution of bypassed oil.