Estimativa e identificabilidade de parâmetros em experimentos de qualidade de espuma em meios porosos

Abstract

The use of foam during alternate water and gas injection is known to increase sweep efficiency in advanced oil recovery. This process significantly reduces gas mobility and increases the apparent viscosity of the gas phase, which improves recovery efficiency. Computer simulations can accelerate the design development cycle and increase knowledge about the systems of interest. However, to use the simulators, it is first necessary to properly calibrate the model parameters to the observed experimental data. As the complexity and number of parameters increase, this task becomes challenging. In particular, depending on the available observations and the mathematical structure of the models, some parameters cannot be estimated, that is, they are not identifiable. This work focused on analyzing the identifiability of parameters in foam displacement models during two-phase flow in porous media, with particular interest in calibrating relative permeability and foam parameters. It is shown how the likelihood profiling technique can help analyze the parameters’ identifiability in three different scenarios. The first scenario considers the availability of two experiments, foam quality scan and steady state, while the second scenario studies calibration using only foam quality scan. All parameters were successfully estimated in the first scenario. However, in the second scenario, the likelihood profiling technique found two unidentifiable parameters, the end point of relative gas permeability and the maximum reduction in gas mobility. In the third scenario, another way to adjust all the parameters is suggested using one more experiment of water and gas flow. The results suggest that the likelihood profiling technique is an important tool in the analysis of parameter identifiability