AbstractThe aerospace industry is undergoing substantial growth and change which includes an increase usage in light weight composite materials coupled with an overall increase in the electrification of so called “fly by wire” aircraft. To cope with this projected growth and change there has been a renewed focus on increasing efficiency both in aircraft design and operations through sustained research and development efforts.
Aircraft compartments are packed with heat dissipating avionic cables and components, the thermal management of which is a primary concern for the reliable operation of aircraft systems. Therefore, there is a pressing need for fast thermo-fluid modelling capabilities, and moreover there is a desire to incorporate such practices early in design. Incorporating Reduced Order Models (ROMs) alongside existing Computational Fluid Dynamics (CFD) simulation has been touted as a promising means to achieve this. A significant portion of these ROMs are built using a numerical technique known as the Proper Orthogonal Decomposition (POD) method.
This work explores the possibility of utilizing control surfaces to develop POD ROMs that can bypass the need to exploit the underlying governing equations with a view to evaluating their potential for fast thermo-fluid modelling in a forced, natural or mixed convection regime. The thesis replicates an existing methodology in the modelling of forced convection, and adapts it to the natural and mixed convection regime. This resulting methodology uses a variant of the standard computation known as the POD with Orthogonal Complements (PODc) while also introducing a Profile Correlation Procedure (PCP) into a 3-part modelling methodology. The proposed framework is first tested on a typical case of natural convection in a square-enclosure with a heated cylinder at its centroid, next being tested on a case of mixed convection in an overhead aircraft crown compartment populated with two avionic components. The findings of the thesis strongly suggest that it is indeed possible to forego the need to exploit the underlying governing equations for cases of natural and mixed convection in compartment models. Fast and accurate reconstructions of thermal fields were achieved for all cases. Although the work recognizes that the PCP component to the methodology requires further refinements to allow for full automation in all cases, the underlying soundness of the method has been demonstrated and accurate reconstructions of velocity fields were therein achieved. The work provides a definite, albeit small contribution to the reduced-order-modelling community, and concludes with a comprehensive discussion which acts as a basis to provide strong recommendations for future work.
|Date of Award||2018|
|Supervisor||Trevor T Robinson (Supervisor) & Marco Geron (Supervisor)|