In this study we develop Proper Orthogonal Decomposition (POD) models for combined steady and transitional (quasi-periodic) convective air flows in a grooved channel with periodic boundary conditions. The grooved channel is appropriately mounted with constant heat flux sources. The constructed POD models for the flow and temperature fields are used in order to predict the off-design flow conditions. Considering the grooved channel geometry (with the height of the heat source H = 2.3 cm and the distance between the heat sources L = 3.0 cm), we form a database using a combination of three distinct databases. Thus we could compare the ability of a POD model applied to the combined database consisting of various flow regimes. The first database consists of steady flow fields with varying Reynolds number up to Re = 230. The second and third databases include instantaneous snapshots of a periodic flow field at Re = 300 and snapshots of a quasi-periodic (chaotic) flow field at Re = 1000 respectively. Then, the combined database is used to perform the POD analysis and extract the POD modes. The modes of the flow and temperature fields are used for off-design reconstructions. The performance of POD model for off-design conditions is evaluated. The mode coefficients are interpolated by kriging to predict off-design conditions. We show that POD model can be used as a useful tool to predict the flow and heat transfer at off-design conditions and to solve inverse design problems in thermo-fluids.

This content is only available via PDF.
You do not currently have access to this content.