CFD-based multi-objective optimization of operating conditions in a small-scale ORC radial-inflow turbine using response surface methodology

Abstract

This study presents a CFD–RSM framework for constructing surrogate-based operating maps and optimizing the operating conditions of a manufactured small-scale radial-inflow turbine used in an Organic Rankine Cycle (ORC) system. Unlike studies focused mainly on turbine redesign, the proposed approach addresses operating-point selection for an existing experimentally characterized turbine with fixed geometry and working fluid. Four operating variables are considered: rotational speed, inlet temperature, inlet pressure, and outlet pressure, with admissible ranges constrained by the coupled operation of the feed pump, condenser, and turbine. A design of experiments with 355 operating points is generated using Latin Hypercube Sampling and evaluated using steady-state three-dimensional CFD simulations with real-gas thermophysical properties. Polynomial response surface models are then fitted to predict turbine isentropic efficiency and power output across the admissible operating range. The results show that a first-order response surface provides the most robust prediction, with validation errors below 5% against independent CFD simulations. The surrogate model is subsequently used for single- and multi-objective optimization. The optimized operating points reveal a clear trade-off between efficiency and power output: the maximum-efficiency and maximum-power conditions do not coincide, while the nominal design condition remains close to the favorable Pareto-optimal region. The resulting operating map enables rapid evaluation of candidate operating points and provides a transferable decision-support methodology for reducing trial-and-error testing and supporting experimental planning and operation of small-scale ORC systems.