Robust optimization for the preliminary design of solar organic Rankine cycle (ORC) systems

Organic Rankine cycle (ORC) powered by solar energy is a viable and effective option for a high efficiency conversion of solar thermal energy into electricity at a distributed scale but recurring fluctuations of the thermal energy often force solar-based ORC systems to operate at part-load conditions. With the aim of including the effects of the expected variations of the heat source and heat sink characteristics, even during the design phase, a novel optimization approach for the preliminary design of ORC systems integrated with concentrating solar collectors is presented and analysed. In particular, the minimization of the expected levelized cost of energy (LCOE) of the ORC unit is chosen as an objective function, while the generation of various scenarios is proposed to face the expected fluctuations on the heat source mass flow rate and temperature and cooling inlet temperature. In this way, the expected off-design performances are involved during the design step, giving robustness to the optimal design solution. The proposed methodology is tested by referring to the solar ORC system configuration of the Ottana solar facility. Firstly, the effect of a robust optimization on the preliminary ORC design is investigated by considering an increased number of scenarios for each of the three most significant ORC input variables (heat source mass flow rate and temperature and ambient temperature). Subsequently, the proposed methodology was applied to an ORC design case by considering a concurrent variation of the three variables and three different working fluids. The results of this study demonstrate that a multi-scenario approach drives towards an ORC configuration with lower performance under design conditions, but less sensitive to the variation of the main inputs. Less expensive solutions are therefore achieved by the proposed methodology, but the annual energy production obtained is comparable with those achieved by adopting a single scenario approach, with a consequent reduction of the LCOE.