|M.Sc Student||Aloush Hila|
|Subject||Heat-integration Synthesis for Real Process Streams using|
a Genetic Algorithm
|Department||Department of Chemical Engineering||Supervisor||Professor Daniel Lewin|
|Full Thesis text|
An effective heat exchanger network (HEN) is a key element in reducing both external energy consumption, as well as CO2 emissions. Over the past decades, the field of HEN study has developed extensively, from sequential methods in the 1970’s to simultaneous methods in the 1990’s. But while the computational means and techniques continued to evolve, the process problems handled remained largely the same - constant heat capacity process streams with no phase changes. During the last decade, a new generation of HEN studies have emerged, taking into account the non-constant values of the thermodynamic properties of the process streams, either by dividing the stream to linear segments, or by adopting a cubic correlation, thus providing a HEN that can be implemented in real processes with a better fit. However, using cubic correlation might result in violations in the minimum temperature driving force, in the actual process.
This work proposes a more realistic representation of the thermodynamic properties, using actual data points for each process stream as provided by a simulation program. The T-H curves are fitted using a cubic spline, enabling closer matching of phase changes as well as non-linearity in the flowing heat capacities, while maintaining the minimal driving force within the exchangers. The actual HEN structure is determined using a genetic algorithm, which progressively improves a population of solutions, such that those with lower total annual cost are retained and refined. Multiple utilities and stream splits are supported.