Sensitivity Analysis and Optimization of Alkaline-Surfactant Flooding in a Thin Clastic Reservoir

Ghadami, Nader (Petronas Carigali Sdn. Bhd.) | Das, Apurba K. (Petronas Carigali Sdn. Bhd.) | Tunio, Kamran H. (Petronas Carigali Sdn. Bhd.) | Sabzabadi, Ali (Petronas Carigali Sdn. Bhd.)

OnePetro 

Abstract

Chemical EOR is one of the promising methods to improve the oil recovery. However, due to high cost of the process, there are challenges to minimize the cost and maximize the oil recovery. Some influencing parameters should be taken into account in a systematic approach to find their impact on oil recovery and accordingly optimizing the process.

In this study, we present a robust optimization workflow of alkaline-surfactant (AS) flooding into a thin clastic reservoir of a field in the Malay Basin. There are coreflood experiments and pilot tests on this field that can be quite helpful to provide a basis to find out the appropriate range of input parameters. Optimization work is based on response surface methodology (RSM) and particle swarm optimization (PSO) technique that aid us to indicate the optimum oil recovery from chemical flooding. In order to get the utmost advantage of this workflow, the waterflooding should be optimized prior to the chemical flooding optimization to maximize the sweep efficiency and oil recovery from the chemical flood.

Evaluation of coreflood and pilot tests indicated that some parameters need supplementary evaluation to investigate their effect on reservoir performance and flow dynamics. These parameters include residual oil reduction by chemical, relative permeability curves, chemicals adsorption, chemical concentration, slug size, injection rate, and initiation time of chemical injection. Based on the result of tornado chart, residual oil reduction and injection rate exhibited highest and lowest impact on oil recovery. RSM was used to explore the relationship between input variables and objective function. Some design parameters such as chemical concentration, slug size and initiation time were examined in this stage. Afterwards, proxy models have been built using polynomial regression and neural network methods. The results showed that the proxy model by neural network method revealed better performance for prediction of the simulation results. The proxy model was used to calculate the oil recovery for any combination of input parameters. Besides, it was used to assess the parameter sensitivity and identify the impact of any input parameter on oil recovery. At the next stage, PSO method was utilized to optimize the oil recovery by chemical flooding. It was found that the optimized water injection rate and pattern for water flooding scenario need further optimization to improve the sweep efficiency and thereby oil recovery by AS flooding at later stage. Running numerous simulation cases is normally expected to optimize the process by conventional methods and the proposed PSO approach can be used to reduce the number of runs significantly. Sensitivity analysis provided a very good understanding about reserve ranges for the different influential parameters. Optimizing the cost of chemical flooding and improving oil recovery are other outcomes of this study.