Here is the abstract you requested from the Thermal_2007 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|A Preliminary Study of Optimization Techniques for Compact Thermal Model Development of Multi-Heat Source Components|
|Keywords: Thermal, Computational Fluid Dynamics , Compact Thermal Models|
|The overall objective of the present study is to explore the possibility of CTM development for opto-electronic components using methodologies that employ either optimization or reduced order modeling approaches. The first route taken is to develop the methodology required in order apply DELPHI  (Development of Libraries and Physical Models for an Integrated Design Environment) based optimization approach for CTM generation of multi heat source components. The DELPHI based methodology employs optimization of the conduction and convection resistances in order to generate the final network of resistances. In essence, the optimization is the minimization of the error between the experimental and calculated values of the junction temperature and heat fluxes by varying the value of the resistances. Once the error is minimized, the resulting network of resistances can be used as a boundary condition independent CTM. Since optimization is the back bone of the DELPHI method, it is important to employ an optimization algorithm that can produce accurate results. The DELPHI method also requires a network solver. The network solver solves for the compact model parameters (junction temperature and heat flux through the boundaries) based on the network topology. For this initial phase, the experimental results from the conduction cooled application of a chip presented in Aranyosi et al.  are used for validation. This problem is chosen because the authors had used a Design of Experiments approach to reduce the number of boundary condition as opposed to a detailed boundary condition set employed by DELPHI. Such a reduced boundary condition set allows for faster execution of the optimization algorithm and for quick validations. Unconstrained minimization using gradient-based Quasi-Newton method, constrained minimization using Sequential Quadratic Method, and a Genetic Algorithm are the three different optimization algorithms that are tested for this purpose. MATLAB is used for implementation and testing of these algorithms. The results of the performance of each algorithm in predicting the experimentally measured temperature values will be presented in the final paper/presentation. References 1. Lasance, C. J. M., Vinke, H, and Rosten H, “Thermal Characterization of Electronic Devices with Boundary Condition Independent Compact Models”, IEEE Transactions on Components, Packaging and Manufacturing Technology – Part A, Vol. 18, No.4, December 1995 2. Aranyosi, A., Ortega, A., Evans, J., Tarter, T., Pursel, J., Radhakrishnan, J, “ Development of Compact Thermal Models for Advanced Electronic Packaging: Methodology and Experimental Validation for a Single-Chip CPGA Package”, IEEE Inter Society Conference on Thermal Phenomena, 2000|
|Arun P. Raghupathy, Doctoral Student
Electronic Cooling Solutions Inc.
Mountain View, CA