A simple approach to identify the proper relative permeability model

11th Curtin University Technology, Science and Engineering (CUTSE) International Conference

工业技术（总论）

Al-Mossawy, Mohammed Idrees^1

Petronas University of Technology, Malaysia Curtin University Malaysia, Miri, Sarawak, Malaysia^1

关键词: Berea sandstone;    Fluid displacement;    History matching;    Relative permeability;    Relative permeability model;    Reservoir simulation;    Unsteady-state flow;    Water saturations;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/495/1/012074/pdf
DOI  :  10.1088/1757-899X/495/1/012074

学科分类：工业工程学

来源: IOP

【 摘 要 】

Relative permeability is one of the crucial input data for any reservoir simulation study. Experimentally, relative permeability tests can be executed by either steady state or unsteady-state flow procedures. Steady-state procedures have the advantage of easy calculations and the drawback of tedious long processes. Unsteady-state procedures take less time but need complex calculations. To shorten calculations of unsteady-state procedures, various models have been developed based on measuring permeability and saturation at initial and final stages of the flow process. This paper presents a simple approach to assess unsteady-state relative-permeability models for simulating a certain fluid displacement process. The Approach includes comparing the experimental average water saturation at the moment of breakthrough ((Swavg)Bth) with modelling (Swavg)Bth that can be estimated from the fractional-flow curve. The presented approach aims to identify the proper relative permeability model before conducting full reservoir simulation studies. This will improve the reservoir simulation process and shorten the history matching. The approach has been applied on six experiments of displacing brine by nitrogen gas in core samples of Berea sandstone. The results shown that two of the displacing experiments can be simulated with MBC model, one experiment can be simulated by Corey's model, and one experiment can be simulated by Kam and Rossen model with error percent less than 1%. Modelling of the other two experiments gave error percent greater than 1% for MBC, Corey, Pirson, and Kam and Rossen models.