Yoo, Jina (2024) Investigation of a coupled optimization concept for heat pump cycles with simultaneous data-driven compressor design. Masterarbeit, FRIEDRICH-ALEXANDER-UNIVERSITÄT ERLANGEN-NÜRNBERG.
PDF
- Nur DLR-intern zugänglich
6MB |
Kurzfassung
Heat pumps are carbon-neutral devices that play a crucial role in decarbonizing heat generation processes and have become key technologies for achieving Net Zero Emission goals by 2050. This study investigates the optimization of heat pumps operating at temperatures exceeding 140°C by integrating data-driven component design and heat pump models, aiming to propose an effective approach to optimize heat pumps while simultaneously selecting the best compressor design. The research questions involve determining the effect of integrating the data-driven models into the process chain, assessing their potential to reduce computational time and deliver reliable results, as well as determining the optimal database size for training them. The study is structured into three parts: System level, component level, and integration level, each employing distinct tools for different model development. At the system level, the Brayton heat pump model is created using Modelon Impact. The component level focuses on constructing a data-driven model or artificial neural network model for designing compressors. The integration level combines all parts into a single system via Python and conducts system optimization. The findings reveal that involving a data-driven model in the process chain significantly reduces the computational time from 4.5 hours to 4 seconds for a single optimization run. This neural network model offers accurate pressure ratio results while the efficiency prediction needs to be improved. In addition, the increase in database size results in higher prediction accuracy for pressure ratio and efficiency. In terms of optimization algorithms, differential evolution is preferred, by achieving the heat pump design with the highest coefficient of performance (COP). In conclusion, this thesis offers valuable insights into the integration of simulation, machine learning, and optimization tools for heat pump design.
elib-URL des Eintrags: | https://elib.dlr.de/204500/ | ||||||||
---|---|---|---|---|---|---|---|---|---|
Dokumentart: | Hochschulschrift (Masterarbeit) | ||||||||
Titel: | Investigation of a coupled optimization concept for heat pump cycles with simultaneous data-driven compressor design | ||||||||
Autoren: |
| ||||||||
Datum: | April 2024 | ||||||||
Open Access: | Nein | ||||||||
Seitenanzahl: | 60 | ||||||||
Status: | veröffentlicht | ||||||||
Stichwörter: | Optimization, Neural Networks, Regression, Surrogate Models, Heat Pump Design | ||||||||
Institution: | FRIEDRICH-ALEXANDER-UNIVERSITÄT ERLANGEN-NÜRNBERG | ||||||||
Abteilung: | TECHNISCHE FAKULTÄT, DEPARTMENT INFORMATIK | ||||||||
HGF - Forschungsbereich: | Energie | ||||||||
HGF - Programm: | Materialien und Technologien für die Energiewende | ||||||||
HGF - Programmthema: | Thermische Hochtemperaturtechnologien | ||||||||
DLR - Schwerpunkt: | Energie | ||||||||
DLR - Forschungsgebiet: | E SP - Energiespeicher | ||||||||
DLR - Teilgebiet (Projekt, Vorhaben): | E - Dekarbonisierte Industrieprozesse | ||||||||
Standort: | Cottbus | ||||||||
Institute & Einrichtungen: | Institut für CO2-arme Industrieprozesse > Simulation und Virtuelles Design Institut für CO2-arme Industrieprozesse | ||||||||
Hinterlegt von: | Gollasch, Jens Oliver | ||||||||
Hinterlegt am: | 11 Jun 2024 12:46 | ||||||||
Letzte Änderung: | 13 Jun 2024 12:29 |
Nur für Mitarbeiter des Archivs: Kontrollseite des Eintrags