Conference Year

2018

Keywords

Pillow Plate Heat Exchanger, Multi-Objective Optimization, Computational Fluid Dynamics, Non-Uniform Rational B-Splines

Abstract

Miniaturization of Plate Heat Exchangers (PHXs) is becoming a central research topic in order to utilize less material and less refrigerant charge to attain similar heat transfer performance, and hence contribute significantly into energy conservation and lower environmental impact. Thus, it is greatly desirable to obtain new designs to achieve this goal. Pillow Plate Heat Exchanger (PPHX) is a type of PHXs with a 3D complex wavy structure, but yet an economical manufacturing process positioning itself as a potential strong competitor among other types of PHXs. PPHXs have the advantage of simple manufacturing process which gives it great design flexibility, and allows new designs to be created simpler and less costly. However, PPHXs are more commonly found in chemical and process industry. Research on PPHXs in HVAC&R is very limited. It is desired to make use of PPHXs advantages in HVAC&R applications. This can be done by creating more efficient designs. The thermal-hydraulic performance of PPHXs is primarily altered by the weld shape, size, and pattern, as well as the pillow height. The shape, and size of the weld is one of the most sensitive parameters affecting the thermal-hydraulic performance of PPHXs. As the weld size is smaller and more streamlined, the pressure drop is reduced significantly. However, the heat transfer area is also reduced using a more streamlined weld shape. In this study, new designs for PPHXs are investigated using different weld shapes that are represented using Non-Uniform Rational B-Splines (NURBS. Each control point in the NURBS curve is a design parameter in the optimization problem. The optimization problem has 11 design parameters. The whole CFD simulation is automated using Parallel Parameterized CFD (PPCFD). Since the CFD simulation of 3D PPHXs is computationally very expensive, the automated CFD simulations and Approximation Assisted Optimization (AAO) reduce the computational time and resources required significantly. A meta-model, using Kriging method, is calculated and verified using random samples from the design space. Multi-Objective Genetic Algorithm (MOGA) utilizes the verified meta-model to calculate optimum designs which have the optimum weld shape and size. The potential enhancement can be up to 50% improvement in heat transfer coefficient and 20% reduction in pressure drop as compared to a selected PPHX baseline design. The optimum designs are also compared to optimum designs of PPHXs with circular spot welds. The potential improvement can be up to 20% in both heat transfer coefficient, and pressure drop.

Share

COinS