On performance optimization for oil-injected screw compressors using different evolutionary algorithms

Energy consumed for pressurizing air makes up a significant proportion of total electrical energy consumption worldwide. To reduce the carbon footprint, it is necessary to have air compressors, which can operate efficiently over a large range of pressures and flow including full load and part load conditions. Several studies have been performed in this area including some which monitor the performance of a large number of compressors to develop strategies for their designs (often with the intent to achieve performance optimization).

This paper focuses on the design and performance optimization of geometrical and oil parameters of oil-injected screw compressors using different evolutionary algorithms such as genetic algorithm (GA), covariance matrix adaptation evolution strategy (CMA-ES), and so on. A comparison of the performance of these algorithms is presented. SCORG and GT-SUITE (commercial software tools for screw compressor thermodynamic simulations and optimization) are used in the integrated model producing promising results. The feasibility of the optimum outcomes generated by these algorithms is critically evaluated from a machine and system design point of view. Finally, in the context of optimization presented here, the simplex converges fastest as compared to other algorithms. In the future study, the system design limitations are to be incorporated as constraints for the optimization along with the objective to improve energy efficiency.