Sensitivity analysis of Computer-aided molecular design problem for the development of novel working fluids for power cycles

Abstract

In recent years there is a large availability of low-temperature heat sources in different applications such as waste heat in chemical industries and refrigeration plants as well as renewable energy sources suchas biomass combustion, geothermal and solar heat sources. Power cycles are an important technical tool to convert this waste heat into usable energy. So far the low-temperature heat cannot be utilized efficiently for electricity generation.In order to optimize the heat transfer process and the power generation, the influence of the working fluid, the cycledesigns and the operating conditions is vital. Multi-criteria database search and Computer Aided Molecular Design(CAMD) can be applied to generate, test and evaluate promising pure component/mixture candidate as process fluids to help optimize cycle design and performance. The problem formulation for the development of novel working fluids is anadvanced CAMD challenge both in terms of data and computational demand, because includes process related as wellas property related equations.In CAMD problems the identification of target properties is often based on expert knowledge. To support identification of relevant target properties, in this study, we propose the use of sensitivity analysis. The sensitivity analysis, among others,provides using information concerning the influence of a certain property of a candidate working fluid on the performanceof the power cycle, i.e. the net power output in this particular application. This information allows the ranking ofsignificance of properties and also the identification of a set of properties which are relevant for the design of a workingfluids.In this study the CAMD problem for the development of novel working fluids for organic Rankine cycles (ORC) isformulated as a mathematical optimization problem. It integrates both a system model for the ORC and property models,such as GC+-based models for estimation of pure component properties, Peng-Robinson equation of state for estimationof enthalpy, entropy, etc.. The system consists of coupled mass and energy balances for a pump, a condenser, a turbineand an evaporator.A global sensitivity analysis is performed to determine which input parameters (e.g. properties of working fluids, cycle design parameters, etc) have important effects on the net power output. To this end, two methods have been usednamely Monte-Carlo based linear regression (so called SRC method) as well as Morris Screening techniques. Theapplication of two global sensitivity analysis methods is highlighted for a case study involving power-cycle design forenergy recovery from low-heat water stream at temperature 120C and peressure 4 bar. Overall this contribution presentsa new approach for the systematic identification of target properties of CAMD problems based on sensitivity analysis.which is validated for the development of novel working fluids of organic Rankine cycles for low temperature heat sources.