Optimisation of stand-alone hybrid energy systems supplemented by combustion-based prime movers

Abstract

A comparative analysis is undertaken between a baseline PV/Batt system, meeting a dynamic load profile, and systems hybridised with supplementary combustion-based prime movers such as Internal Combustion Engines (ICEs) or Micro Gas Turbines (MGTs) of 30–65kW rating. This study sheds light for the first time on a number of research questions not addressed in earlier studies. The main contributions of the work are namely to: (i) analyse the effects of the start-up threshold and the type of supplementary prime mover on the Cost of Energy (COE, $/kWh), lifetime CO2 emissions, and (unrecovered) waste heat for a specified reliability (Loss of Power Supply Probability-LPSP); (ii) investigate the effects of including the transient start-up periods of prime movers on systems sizing; and (iii) look into the effects of using two smaller sized (tandem) supplementary prime movers versus a single larger one on the operational characteristics. The research also analyses (iv) the effects of the methods used (e.g. temporal resolution of simulations, Genetic Algorithm (GA) population size) on the COE, lifetime CO2 emissions, and (unrecovered) waste heat.The results of this study indicate that PV/Batt and PV/Batt/ICE systems have comparable COEs but are preferable to PV/Batt/MGT. The minimum starting thresholds of supplementary devices (ICE or MGT) have significant effects on renewable energy penetration, genset running hours, waste heat generation, and Life Cycle Emission (LCE, kg CO2-eq/yr), but insignificant effects on the COE. The results also show that the transient start-up of supplementary devices has a negligible influence on overall system sizing. The COE resulting from the use of larger capacity prime movers (60kW ICE or 65kW MGT) is comparable to deploying two smaller capacity prime movers (30kW) but results in higher renewable energy penetration, an improved duty factor with lower LCEs. Additionally, the COE increases slightly (2∼5%) when the models run at 15min temporal resolution compared to 60min.