| dc.description.abstract | Due to climate change is it desirable to improve the energy efficiency of production systems. The environmental impact will decrease with increasing energy efficiency and it will even reduce the operational cost. Heat integration is a much used method for improving the energy efficiency and the annualized cost. The usual heat exchanger network synthesis problem is to design a heat exchanger network that minimizes the annualized cost. A set of hot streams which needs to be cooled and a set of cold streams which need to be heated is given. When the possibility of expanding or compressing some of the streams is added, the problem gets more complex. Without pressure manipulation the streams are continuously heated/cooled from the supply temperature to the target temperature. Pressure manipulation changes the temperature of the fluid going through the turbine/compressor, so the stream is split into two parts. This can cause the streams to change identity and makes the modeling and optimization more difficult. The objective is to minimize the hot and cold utilities, minimize compression work and maximize the work produced by turbines. Because of the difference in energy quality is exergy going to be used instead of energy. Fu and Gundersen (2015b) have made some insights and a graphical procedure that finds a good solution to the problem. The procedure and theorems have not been received as well as the authors hoped, so an optimization model not based on their insights was of interest. A MINLP model, which did not utilize the insights from Fu and Gundersen (2015b), was developed in the project thesis. The model handled expansion of hot and cold streams above ambient. In this master thesis have the model been extended to also include compression and below ambient temperatures. The MINLP model did not always find the optimal solution, so a new model with the use of insights was developed. The model became a much simpler LP model and have for all test cases found an equally good or better solution than the first model and the graphical procedure. Both models have been implemented in GAMS and have been tested on several test cases. The concept of simultaneous work and heat integration is quite novel and the models developed in this thesis is just a step in the right direction. The end will probably be an optimization model which designs the work and heat exchange network with respect to the annualized costs. | |
