Open Collections

Amedu, Ojogbane

University of British Columbia

Thermal management challenges in modern avionics systems are increasing due to rising power densities, compact designs, and complex integration requirements. Avionics systems are at risk of overheating, being throttled, and eventually shrinking those systems operational lifespan. Industry standards such as DO-160G impose stringent thermal performance requirements, making it essential to develop accurate predictive models and efficient optimization strategies for avionics bay layouts. This thesis presents a systematic evaluation of numerical modeling simplifications in avionics thermal analysis, assessing the impact of geometric approximations, airflow blockage, and system interactions on predictive accuracy. Geometric simplifications were analyzed by comparing detailed and simplified representations of avionics units in computational models. The results demonstrate that while component detail increases the accuracy of a model, simplified geometries provide an adequate representation for practical applications. Using simple rectangular blocks to model airflow blockage was found to be an ineffective solution, decreasing the relative accuracy of predictions. Heat flux wall boundary conditions were used to represent the heat transfer between adjacent units, showing that models which ignore surrounding avionics present inaccurate solutions that underpredict the thermal risk of an avionics unit in a real avionics bay. These findings were validated through experimental analysis using a controlled avionics test chamber. Notably, validated modeling revealed simpler, more cost-effective cooling designs for avionics units. In addition to numerical model evaluation, this research leverages the Thermal Risk Approach (TRA) for avionics bay optimization. TRA is incorporated into four optimization methodologies; Score Minimization Method (SMM), Min-Max Method, Multi-Objective Analysis (MOA), and Weighted Risk Scoring (WRS) that use separate criteria to determine the ideal placement of units inside an avionics bay. A Multi-Level Selection Genetic Algorithm (MLSGA) utilizing these methodologies was developed, providing a robust, scalable tool for optimizing complex avionics bays. Comparative analyses were conducted between optimized and randomly generated layouts using SMM, Min-Max, MOA, WRS and MLSGA methodologies. The MLSGA approach generated substantial reductions in the number of high-risk units and the average surface temperature for units in a complex avionics bay environment. This work provides a foundation for more efficient and reliable thermal management strategies in modern avionics systems.

Text

2025-04-25

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/90803

Mechanical Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/