Abstract
Topology optimization (TO) is a powerful method with high degree of design freedom. However, TO is difficult to perform effectively for complex problems represented by flow boiling. This paper builds a framework that avoids time-consuming two-phase simulation and generates structures with high flow boiling heat transfer performance. In this study, the necessary numerical methods are modeled and integrated into a multi-fidelity workflow, including low-fidelity TO and high-fidelity evaluation. Based on these models, the flow boiling heat transfer performance of low-fidelity TO structures is parametrically investigated under different heat flux, subcooling and inlet velocity. The results show that the high temperature at corners of the heat sink leads to accumulation of vapor that severely degrades the heat transfer performance. Based on this result, the high temperature at corners in the low-fidelity temperature field is considered to be a feature with poor high-fidelity performance. This feature is verified by comparing performance of three TO structures, and interestingly, the results show that structures with high low-fidelity performance do not lead to high high-fidelity performance. Then this feature is quantified as an index. By comparing this index under the low-fidelity condition, the performance under the high-fidelity condition can be judged. Twelve TO structures are evaluated by this index, and the results show that there is a significant positive correlation between the index and flow boiling heat transfer performance. The new index can characterize the high-fidelity flow boiling performance of low-fidelity TO structures.
