Abstract
The acquisition of contamination-free fluid samples in hydrocarbon reservoirs drilled with oil-base mud (OBM) is challenging due to the presence of multiple fluid phases as well as partial-to-complete miscibility between reservoir fluids and OBM. Throughout the sampling process, varying concentrations of OBM contained in the sampled fluid will lead to changes in observed (apparent) fluid properties. Similarly, sand-face transient pressure measurements are affected by OBM invasion as the invasion process itself modifies both fluid viscosity and fluid density in the near-wellbore region due to mixing between different hydrocarbon components.
We use a commercial adaptive-implicit compositional numerical simulator to model the filtrate cleanup process during fluid sampling and to compare the predicted pressure and apparent fluid properties at the sand-face against observed field measurements. A history-matching approach is used to estimate formation permeability and permeability anisotropy.
We apply the proposed workflow to three sets of field measurements of sink probe pressure, observation probe pressure, gas-oil ratio (GOR), and flow rate acquired with a formation tester in light-oil formations. Since the formation is invaded with oil-base mud filtrate that is assumed free of gas, GOR can be used to discriminate between fluids. We use a dimensionless fluid contamination function to relate transient GOR measurements to sample fluid quality. The successful comparison of simulations to field measurements helps us to diagnose and quantify adverse data-acquisition conditions such as plugging and noisy transient data. It is found that numerical simulations are a reliable way to verify the internal consistency of the transient measurements of flow rate, pressure, and GOR in the presence of biasing acquisition problems.
We perform sensitivity analyses to identify the dominant governing parameters such as formation properties, formation-tester flow rates, relative permeability, and radial extent of mud-filtrate invasion, on transient measurements of sand-face pressure and sampled fluid contamination. Our observation is that transient pressure, GOR, and density variations are sensitive to both the radial extent of mud-filtrate invasion and the rate of fluid cleanup. If the radial length of invasion is large, the total pumped volume must be increased in order to retrieve representative fluid samples. This can be achieved either by increasing the duration of the test, using higher rates of fluid withdrawal, or both.