Plate Heat Exchangers (PHXs) are used in a wide variety of applications including, but not limited to Heating, Ventilation, Air-Conditioning, and Refrigeration (HVAC&R). PHXs are favored by the HVAC&R industry due to their compactness, flexible sizing, close approach temperature, and good heat transfer performance. PHXs are increasingly utilized and becoming very competitive in two-phase flow applications due to their desirable thermal-hydraulic characteristics. Pillow plate heat exchanger (PPHX) is a type of PHXs that consists of wavy plates that are welded together with a certain pattern using spot welding, sealed at the edges, and then inflated in a hydroforming process. PPHXs have an economic advantage over other types of PHXs due to their simple manufacturing process. Additionally, the complex wavy structure of the pillow plates creates an excellent heat transfer medium and thus, if their performance is optimized, they can potentially replace other types of PHXs in a wide range of applications. However, very limited research is done regarding the use of PPHXs in the HVAC&R and no research on their optimization is found in literature.

The first objective of this thesis is the optimization of PPHXs using four design parameters including their basic geometry parameters using Parallel Parameterized Computational Fluid Dynamics (PPCFD) and Approximation Assisted Optimization (AAO). The potential improvement in thermal-hydraulic performance is expected to be at least 50% as compared to existing designs. The second objective is to perform a comprehensive multi-scale analysis with topology and shape optimization integrating Non-Uniform Rational B-Splines (NURBS) to obtain a novel PPHX design with at least 20% improvement in thermal-hydraulic performance as compared to optimal chevron PHXs designs. This will improve energy efficiency significantly on the component level and potentially on the system level.

Finally, a comprehensive literature survey shows a significant gap regarding PHXs modeling with respect to combining robustness, accuracy, flexibility, and convenient speed into a single model. A current PHX computer model in literature is significantly improved in the aforementioned aspects using a novel algorithm. The model is also used as a component on a system level modeling to evaluate the performance of PHXs on the system level.