Pulser Model: Updated Framework and Lessons Learned from Applications in Forecasting CO2 Flood Performance

Abstract

Abstract Reservoir management of Carbon Dioxide-Enhanced Oil Recovery (CO2-EOR) floods is essentially an optimization problem-maximizing the oil recovery while simultaneously minimizing the CO2 injection volumes. Liu et al. (2014) proposed a parametric model of recovery factor as a function of miscible CO2 injected in dimensionless form. Incorporating the lessons learned from application of the model for production forecasting of multiple CO2 floods across the Permian basin, the updated model has better predictive power over the life of the flood. Results from applications to actual field data are also presented. The Pulser Model is linearized, which allows for faster solution and robust convergence. Algorithmic nuances for regression are described. A physical interpretation of the model parameters is explored so as to facilitate physics-constrained optimization. This allows calibration of parameters to data such as heterogeneity (from core) and CO2 storage capacity of the rock. Sensitivity analysis of model parameters is conducted on field data sets to evaluate relative and physical significance of the model parameters.