Assessment of Optimal Operating Conditions in a SAGD Project by Design of Experiments and Response Surface Methodology

Abstract

The steam-assisted gravity drainage (SAGD) is likely the most efficient and important thermal recovery in-situ method to produce extra-heavy oil and bitumen reservoirs. Indeed, a huge expansion of commercial SAGD applications is taking place, particularly in the Alberta oil sands of Canada. Numeric reservoir simulators are available for predicting SAGD performance indicators and are used as tools to support reservoir management decisions. Those decisions are related to the selection of optimal values of controllable variables, including operating conditions such as preheating period, sub-cooling temperature, maximum steam injection pressure, maximum steam injection rate and steam quality, and temperature. In order to make unbiased decisions, the optimization process should be done considering the stochastic character of reservoir variables. However, the high computational time associated to the complex numeric solution of reservoirs under the SAGD recovery process makes the integration of reservoir uncertainty to the SAGD decision-making process an almost impossible task. Thus, a calibrated-proxy is used in this work as an efficient substitute of the numeric simulator to accomplish such a task. Design of experimental techniques and response surface methodology allowed the construction of a simple model by fitting a quadratic model to reservoir simulator outputs extracted from a chosen set of simulation cases. The main purpose of this work was to optimize the production and injection constraints of a SAGD well pair, based on an Athabasca oil sands data set, in order to maximize the net present value in presence of reservoir uncertainty. The production and injection constrains considered in the problem were: injection pressure, maximum steam flow rate, and sub-cooling temperature; and the reservoir uncertainty was represented by vertical permeability, porosity, thickness, horizontal to vertical permeability, and initial oil saturation. The results indicate that experimental design and response surface techniques are excellent tools to quickly obtain valuable information about the SAGD performance.