Experimental investigation and ANN analysis of a four-intersecting-vane rotary expander in a micro-scale organic Rankine cycle system

Organic Rankine cycles (ORCs) are often used for power generation from low-temperature waste heat. Although the ORC is a well-established technology for medium- to large-scale applications, several issues need to be resolved before it can be widely used in micro-scale applications, including the development of a reliable, lowcost, and efficient expander. A small-scale ORC experimental setup was built using R134 as the working fluid. This paper aims to investigate the performance of a four-intersection-vane rotary expander in a small-scale ORC system. The characteristics include filling factor, isentropic efficiency, and shaft power. The experiments are tested at suction temperatures up to 70 degrees C, rotational speeds up to 900 rpm and suction pressures up to 11 bar (abs). The expander demonstrated shaft power of up to 73 W, minimum filling factor of 1.9, and isentropic efficiency of up to 45.6 %. The impacts of lubrication on the performance of the expander are discussed. Besides the experimental work, an Artificial Neural Network (ANN) and Genetic Algorithm (GA) modelling approach was proposed to achieve higher accuracy in mapping the expander's performance. The developed model is evaluated with different parameter settings, train functions and learning rates to increase the prediction accuracy.