A new method is presented to assess layer connectivity and determine dynamic reservoir properties for use as inputs in a numerical model for a multi-layered reservoir. Large scale formation tester transient measurements and downhole fluid analysis information are used to construct the reservoir model. Pressure and fluid measurements distributed vertically at the wellbore and laterally in different wells are used to delineate the sands and model the complex fluid behaviors. Pressure transients recorded in different wells from a wireline formation tester (WFT) in small scale drill stem test (DST), or from conventional well tests, are then used to constrain the dynamic model properties through a grid-based inversion technique, so that observed test data and the model response match. Our workflow consists in constructing initial reservoir models using the acquired log and test data, and constraining the model properties to the transient test data. Wireline conveyed tests and downhole fluid analysis (DFA) measurements from different wells are inputs to the geomodel, and building steps to assess layer connectivity from pressure gradient and fluid matches. Then a fluid model equation of state (EOS) is built from PVT experiments on the fluids in the laboratory, and tuned to the observed changes against depth in the fluid properties from downhole measurements, such as solution gas-oil ratio or saturation pressures. Initial dynamic layer properties, such as horizontal and vertical permeabilities are typically obtained from logs and cores. In our workflow we calibrate these properties to interval pressure transient tests (IPTT) and conventional well tests. These modeled properties are constrained to match the IPTT or well tests. An optimization routine is conducted, using adjoint simulations, to minimize an objective function composed of observed bottomhole pressures given different flowrates during a transient test and the pressure response of the model. In this optimization step, the skin values are introduced explicitly in the model, but the dynamic properties are modified until a match is obtained. Once the model is constrained, it is then used to develop an optimum development strategy. We present a case study to illustrate our methodology, in a new field development setting, with transient tests and WFT measurements acquired in three different appraisal wells, which traverse multiple channel sands reservoirs. These tests were used to build an optimal model which respected all acquired measurements. We finally show how this calibrated model was used to help make a decision about the development strategy.
展开▼
