This paper presents an approach for reducing detailed numerical models of electronic equipment into compact thermal-fluid models. These compact models have been created using network analogies representing mass, momentum and energy transport to reduce computational demand, preserve manufacturer intellectual property, and enable software independent exchange of information between supplier and integrator. A strategic approach is demonstrated for a steady state case from reduction to model integration within a global environment. The compact model is robust to boundary condition variation by developing a boundary condition response matrix for the network layout. A practical example of electronic equipment cooled naturally in air is presented. Solution times were reduced from ∼100 to ∼10−3 CPU hours when using the compact model. Nodal information was predicted with O(10%) accuracy compared to detailed solutions. For parametric design studies, the reduced model can provide 1800 solutions in the same time required to run a single detailed numerical simulation. The information generated by the reduction process also enhances collaborative design by providing the equipment integrator with ordered initial conditions for the equipment in the optimization of the global design. Sensitivity of the compact model to spatial variations on the boundary node faces has also been assessed. Overall, the compact modeling approach developed extends the use of compact models beyond preliminary design and into detailed phases of the product design lifecycle.
Compact Modeling Approach using GA for Accurate Thermal Simulation.
