Session: 11-19-01: Methods in Computational Heat Transfer and Their Applications

Paper Number: 145030

145030 - Multi-Disciplinary Analysis and Optimization of Micro-Channel Evaporators 

Introduction

Efficient thermal management is necessary to dissipate high heat fluxes in applications ranging from cooling power electronics to nuclear reactors. For these applications, micro-channel evaporators using liquid-vapor phase change offer a compact solution that achieves remarkable heat transfer. However, the optimal geometry for a thermal management system is often unclear, especially when dry-out of the evaporators is a concern. This motivates the development of a design-optimization framework that can help inform the design of heat sinks with micro-channel evaporators.

Contribution

Motivated by the gap identified above, this work develops and validates a multi-disciplinary model of micro-channel evaporators embedded in a solid heat sink. A design-optimization framework is proposed and demonstrated for maximizing the energy transfer from the external heat source to the refrigerant flowing through the micro-channel evaporators under various operating conditions. Notably, the model is analytically differentiated using algorithmic differentiation and the adjoint method, which enables optimization of complex channel trajectories.  The optimization framework is used to investigate the impact of different working fluids, operating conditions, and heat-sink materials on optimal micro-channel designs.

Methodology

The micro-channel evaporators are modeled using quasi-1D two-phase flow equations and the temperature in the heat sink is modeled using the heat equation. The energy transfer from the heat sink to the refrigerant flowing inside the channels is coupled in a consistent and conservative manner. B-Spline control points parameterize the paths of the evaporator channels through the heat sink, and the control point coordinates serve as the design variables in the optimization process. We adopt a gradient-based optimization algorithm because of its favorable scalability with the dimension of the design space, which allows us to achieve complex channel configurations. We use PyMFEM to set up the heat-transfer problem in the heat sink and OpenMDAO for the optimization framework.

Preliminary results

The thermal component of the analysis has been verified using the method of manufactured solutions. The two-phase flow model has been validated by comparing it to experimental results. We conducted two preliminary design optimization studies with 1 and 2 evaporator channels. Each channel was parameterized by 4 control points but their inlet and outlet locations were fixed in space. The internal B-Spline control points were allowed to move only in the direction of the heat-source.  A 4.63% increase in the thermal flux transferred from the heat source was observed when the number of evaporator channels was increased from 1 to 2 in the final optimized design of the heat sink.

Conclusions

In the final paper we will have successfully implemented and verified a multi-disciplinary analysis of micro-channel evaporators. We will also propose an optimization framework that generates an optimized design that can effectively operate in the two-phase regime without experiencing dry-out, which diminishes the performance of these evaporators under high heat loads. The framework also allows the user to control the complexity of the design through the specification of the number of B-Spline control points. Results will be demonstrated for steady-state operating conditions and a pareto front of optimized designs will be presented across a range of refrigerant inlet temperatures and outlet pressures. Our optimization studies will consider different sets of heat sink material properties (aluminum, stainless-steel, copper) and refrigerants (water, R11, R134). We plan to provide a prototype design that is additively manufactured and tested in an experimental setup. Dynamic heat loads are reserved for potential future work.

Presenting Author: Vignesh Ramakrishnan Rensselaer Polytechnic Institute

Presenting Author Biography: The presenting author is a 3rd year Aerospace engineering, Ph.D. student at Rensselaer Polytechnic Institute. He is highly interested in research in the field of numerical methods, high performance computing, and numerical design optimization.

Authors:

Vignesh Ramakrishnan Rensselaer Polytechnic Institute
Arif Mohammad Rensselaer Polytechnic Institute
Jason Hicken Rensselaer Polytechnic Institute
Sandipan Mishra Rensselaer Polytechnic Institute
Shankar Narayan Rensselaer Polytechnic Institute