Title: Towards a port-Hamiltonian approach to study Stirling-cycle devices (PDF)
Keywords: port-Hamiltonian systems theory, thermodynamics, GENERIC framework, Stirling-cycle devices, renewable energy
Abstract: Energy conversion devices which are based approximately on the Stirling cycle comprise ‘engines’ which can use low-temperature heat sources such as solar thermal energy and industrial waste heat as well as ‘refrigerators’ which can be used for air-conditioning. Currently, these devices are not very popular but they could play an important role in dealing with global challenges such as pollution and climate change. Port-Hamiltonian systems theory provides a modular approach to modeling physical systems. These systems are defined based on a geometric object called Dirac structure which encodes the exchange of power between different aspects of the system. Port-Hamiltonian systems are naturally passive making them attractive representations of complex systems arising from network modeling as well as modeling of distributed parameter systems. Their structural properties turn out to be useful for accurate numerical computations, model generation, model simplification, optimization and control. In order to apply this theory to nonequilibrium thermodynamic systems such as Stirling-cycle devices, a previously mentioned connection to the GENERIC framework is explored in this thesis. By using the total exergy as a Hamiltonian function, irreversible and reversible processes can be modeled on an equal footing. The structure of GENERIC port-Hamiltonian systems ensures that the reversible dynamics conserve the exchanged exergy and the irreversible dynamics destroy some of the exchanged exergy. A simple lumped-parameter example shows how the graphical modeling language of bond graphs can be used to represent GENERIC port-Hamiltonian systems. Simulation results are obtained based on a software implementation which automatically turns equations into results. Manually writing problem-specific code is avoided by leveraging symbolic computation and code generation.
Software: git repository
Supervisors: Sigrid Leyendecker, Paul Kotyczka, Tobias Scheuermann