Dynamic simulation and experimental validation of an Organic Rankine Cycle model

The aim of this thesis is to develop a methodology to model and simulate the dynamics of Organic Rankine Cycle (ORC) power plants, and to demonstrate and validate this methodology by performing experiments on a laboratory-scale ORC plant. Typical plant models for ORCs provide steady state analysis of thermodynamic cycles and losses, but for a system that has complex starting mechanisms and undergoes fluctuations in operating conditions due to environmental effects, dynamic models are needed to predict how the system will behave. Two main questions are interesting to an ORC designer. What is the start-up and shut-down time of the plant? And: What effect does an unprecedented slow or sharp transient in one or multiple physical variables have on the system? There are a number of modelling libraries available for the simulation of Rankine cycles, however there is no complete package that covers all potential dynamic scenarios and is fully documented with guidance on how to develop a stable dynamic model. Issues such as component selection and parameterisation criteria, compilation of stable models in large systems, simulation initialisation strategies, and heat transfer model selection present many challenges to the inexperienced modeller. This thesis aims to address these issues and document strategies to overcome them. Existing modelling libraries do not include extensive heat transfer models that can accurately simulate the heat exchange that occurs during dynamic transients in ORC heat exchangers. Void fraction is a critical variable for the dynamics of a closed thermodynamic cycle that depends on accurate heat transfer models, and models that switch between single-phase and two-phase heat transfer correlation are beneficial here. Development of an extension of the existing heat transfer models to better model these effects is another aim of this thesis. A smaller ORC laboratory that is separate to the main facility used in this project was available early in the thesis to test initial heat exchanger and cycle modelling results. An intermediate project goal was to use this laboratory to model and analyse a novel small-scale solar cogeneration unit that uses cheap and available components to heat water and produce power using a scroll expander. The completion of this goal was seen as a fundamental step toward understanding the physical characteristics of an ORC that is producing power, and to observe the system dynamics. The results of the study include a typical day's power output for various times of the year, and show the competitiveness of this type of system. The major contributions that originate from the main body of work on the larger ORC facility are: Development of extended pipe models that include an extra wall for heat transfer to a shell or the environment; Development of detailed, deterministic plate heat exchanger models with descriptive parameters that can be quickly configured by an inexperienced modeller. These include extensions to heat transfer models and a new phase switching method that is more accurate and robust than those currently available; Implementations of a number of heat transfer correlations into the existing libraries, and discussion on when and how the best place to use them is; Development of an extended expansion valve model and discussion on how best to characterise, parameterise, and initialise it; Tutorial on how to choose the correct models and initialisation parameters for a compiled organic Rankine cycle model; and Discussion on initialisation strategies for the compiled ORC model, and presentation of a preferred initialisation method. The developed models were configured for simulation of the target experimental ORC. Experimentally stable operating state points representative of the entire operating range of the ORC were chosen as target conditions to compare steady state and dynamic simulation results against. The steady state points at pressure and temperature sensors between each component in the ORC were used to record data for six steady state cases. Two pressure sensor locations were chosen as representative of the ORC low-side and high-side pressures, based on their proximity to the primary pressure drop driver, the expansion valve. Twelve dynamic test cases using variations in expansion valve closure and pump speed were used to record dynamic test data for comparison against simulation data. A pressure deviation of less than 6% was observed in all steady state test cases except one. The outlying highest deviation case showed 16% deviation in the low- side pressure variable. The simulation results for the dynamic test cases matched up closely to the experimental data. The observed deviation between simulation and experiment results for settling time was less than 20 seconds for the worst matching cases. In five out of twelve cases, the dynamic pressure transient matched the experimental data with almost zero deviation. The approach taken in this project can be used as a guide for future researchers and model developers to overcome the aforementioned issues in order to further advance research in the area of dynamic modelling of Organic Rankine Cycles.