Near-Well Reservoir Monitoring Through Ensemble Kalman Filter

In the management of reservoirs it is an important issue to utilize the available data in order to make accurate forecasts. In this paper a novel approach for frequent updating of the near-well reservoir model as new measurements becomes available is presented. The main focus of this approach is to have an updated model usable for forecasting. These forecasts should have initial values that are consistent with recent measurements. The novel approach is based on utilizing a Kalman filter technique. The idea behind the Kalman filter is to incorporate the information from the measurements into the current estimate of the state of the model, taking into account the uncertainty that belongs both to the state of the model and the measurements. The uncertainty of the model is updated simultaneously with the model itself. A benefit of this approach compared to usual history matching is that the initial values for the forecasts will be in better agreement with the current measurements. Originally, the Kalman filter had shortcomings for large, non-linear models. During the last decade, however, Kalman filter techniques has been further developed, and applied successfully for such models within oceanographic and hydrodynamic application. This work is based on use of the ensemble Kalman filter. The ensemble Kalman filter is easy to implement, and have some good properties for non-linear problems. Here, we demonstrate the use of this technique within nearwell reservoir monitoring, focusing on its performance in forecasting the future production. Introduction Several different smart well systems are available with different functionality. The simplest systems consist of sliding sleeves which only can be open or shut and without any monitoring. The most advanced systems consist of infinitely variable chokes and extensive monitoring like pressure, temperature, multi-phase metering, and resistivity and seismic sensors for tracking near well fluid contacts. The smart well systems are motivated by the possibility of improved reservoir management. Remote choking or shutting zones with poor performance will cause an immediate response on the well performance without any expensive well intervention. Another benefit of smart well systems is improved reservoir monitoring. Smart wells systems add value by enhancing workflow cycles containing the key elements of measurement, modeling and control. Several papers have been presented where the possible benefits of using smart well systems have been quantified. In all these papers the reservoir model is assumed to be known. However, a key element in the measurement, modeling and control loop is how to update the near well reservoir model based on the measurements. This is the focus of the present paper. A novel approach for updating a near-well reservoir model based on measurements in the well will be presented. The approach applies a Kalman filter technique and both the reservoir properties and the state of the reservoir is updated. Benefits of this approach is that the initial values of the forecasts will be in better agreement with the current measurements and that the methodology is well suited for frequent updating of the near well model. An alternative methodology for updating the near-well reservoir model consist in finding the reservoir properties which gives the least difference between measured data and model results within a given time interval. We start by describing the reservoir model. Then the ensemble Kalman filter methodology applied to near well reservoir modeling is presented. Examples of application of the methodology are given, and finally some conclusions are drawn. The reservoir model The model equations for two-phase, immiscible porous-media flow with isotropic permeability are (see, e.g. Aziz & Settari )