Applying the Probabilistic Collocation Method to Surfactant-polymer Flooding

Enhanced oil recovery has achieved great attention during the past few years. However, broad scale implementation requires greater understanding of the relevant uncertainties and their effect on performance. Quantifying this uncertainty is very important for designing these processes, yet traditional methods which are usually based on Monte Carlo simulations require a large number of realizations to produce convergent results. We propose the use of a non-intrusive approach known as the Probabilistic Collocation Method (PCM) to quantify parametric uncertainty for surfactant-polymer flooding. The quantification of uncertainty was performed for surfactant/polymer related state variables such as adsorption rates and residual saturations. The PCM is performed on two reservoir models: a modified section of the SPE10 model and the PUNQ-S3 model. The random input variables PDFs are first approximated using polynomial chaos expansions and then probabilistic collocation is used to produce approximations of the reservoir model using the collocation points obtained via Gaussian quadrature and Chebyshev extrema. These approximations can then be used to produce PDFs for output variables such as the recovery factor. Results show that PCM produces similar results to those obtained via Monte Carlo simulation, which requires a large number of simulations, while requiring significantly lower number of simulation runs.