The Effect of Including Tracer Data in the EnKF Approach

Abstract This paper presents for the first time: Coupling of the EnKF methodology and tracer data, a new developed tracer simulator which accounts for partitioning gas tracers, and coupling of the EnKF methodology with partitioning gas tracer data. The ensemble Kalman Filter (EnKF) has recently gained popularity as a method for history matching. The EnKF includes online update of parameters and the dynamical states. An ensemble of model representations is used to represent the model uncertainty. Tracers are widely used to increase the understanding of fluid flow. Tracers can be used to label injection fluids, hence, well connections and fluid patterns can be established when the tracer appears in production wells. Tracer data contains valuable information but are typically underexploited; most of the tracer-tests are only evaluated in a qualitative manner, without any kind of comparison to simulation results. This paper brings together tracer and modelling competence by including tracer data as measurements in the EnKF methodology. Gas tracers in reservoirs are partitioning tracers and must be modelled as such. This requires that adequate options are available for modelling of these tracers, both with respect to convection terms and diffusion/dispersion terms in the conservation equation. In this paper we therefore present a new tracer simulator, which includes partitioning of gas tracers. The tracer simulator includes separate time step control and a second order spatial numerical scheme to reduce numerical smearing of the tracer data, and has been coupled with the EnKF methodology. The value of tracer data, and partitioned tracer data, in the EnKF approach is demonstrated on North Sea based examples. The permeability and fault transmissibility multipliers of the reservoirs are estimated by EnKF. These examples show that tracer data can be used successfully in an automatic updating scheme, not only by the traditionally manual updating.