Summary
Two methods for characterizing reservoir pore pressure and reservoir permeability during UBD while applying active tests are presented and evaluated. Both methods utilize a fast, dynamic well fluid-flow model that is extended with a transient reservoir model. Active testing of the well is applied by varying the bottomhole pressure in the well during the drilling operations.
The first method uses the Levenberg-Marquardt optimization algorithm to estimate the reservoir parameters by minimizing the difference between measurements from the drilling process and the corresponding model states. The method is applied after the drilling process is finished, using all the recorded measurements. The second method is the ensemble Kalman filter, which simulates the drilling process using the dynamic model while drilling is performed, and updates the model states and parameters each time new measurements are available. Measurements are used that usually are available while drilling are used, such as pump rates, pump pressure, bottomhole pressure, and outlet rates.
The methods are applied to different cases, and the results indicate that active tests might improve the estimation results. The results also show that both estimation methods give useful results, and that the ensemble Kalman filter calculates these results during the UB operation.
Introduction
During UBD, the well pressure is kept below the reservoir pore pressure, and reservoir fluids flow into the well. The flow rate from the reservoir depends on the pressure difference between the reservoir pore pressure and the well pressure, in addition to other reservoir parameters, such as permeability and porosity. The viscosity and compressibility of the reservoir fluids also influence the influx rate.
The influx of reservoir fluids causes variations in the annulus section of the well, because of changes in well fluid composition and well fluid-flow rate. By measuring some of the fluid-flow parameters of the well, such as pressures changes and rate changes, the reservoir parameters causing the influx might be identified. This is the principal idea that also is the basis for well testing and transient reservoir analysis. Identification of the reservoir properties close to the well gives important information for planning the well-completion design. If highly productive zones can be located, then the use of smart completion can be better utilized.
Reservoir characterization during UBD has received attention from several research groups in recent years. Kardolus and van Kruijsdijk (1997) developed a transient reservoir model based on the boundary-element method. This model was compared with a transient analytical reservoir model. One of their findings was that the transient analytical reservoir model could be used for evaluation of the parameters in the reservoir. In a following study, van Kruijsdijk and Cox (1999) presented a method for identifying the permeability in a horizontal reservoir based on measurements of the reservoir inflow. The flow effects caused by the reservoir boundaries were included in the flow calculations.