Dynamic Modeling of Heat Power System : Modeling of a Heat Power System Using Physical and Data-driven Methods and Investigation of a Moving Boundary Method

Abstract: Our society is becoming more and more electrified every day. However, a significant portion of the world’s electricity generation relies on the combustion of fossil fuels to produce heat, which is subsequently harnessed to generate electricity. One way of generating electricity from heat is by utilizing a Rankine cycle. The basis of a Rankine cycle is to heat a liquid to its boiling point, which causes an increase in pressure that is used to spin a turbine and a generator. Many industries, such as transportation and manufacturing, produce large amounts of waste heat that needs to be removed from the main process. A Rankine cycle variant called an organic Rankine cycle can be used in a heat power system to generate electricity from lower-temperature waste heat, which increases efficiency since less heat is wasted. This thesis focuses on constructing a dynamic model of Climeon’s heat power system called HP300. The HP300 utilizes an organic Rankine cycle to generate electricity. Dynamic modeling is valuable because it provides a deeper understanding of the system, which is beneficial for its development and improvement. Moreover, a system model has the potential to enhance the system’s performance by using advanced control methods. The HP300 consists of four main components: a pump, a turbine, an evaporator, and a condenser. Each component will be modeled individually, and the complete model will be constructed by combining the component models. Additionally, an in-depth investigation of an advanced modeling method for heat exchangers is to be conducted. The constructed model in this thesis has an average error of 4%. The pump and turbine were modeled as steady-state models, and the evaporator and condenser were modeled with data-driven state-space models. The most important output of the model is the power generated by the turbine. The power was modeled with an average error of 6%. The turbine model performs best for pressure ratios of 1.75 and above. The model for the condenser had larger errors than the evaporator since it had fewer input variables. Improving the model of the condenser would decrease the overall errors of the model.