Computational method for optimizing segmented nonimaging concentrators

TL;DR: In this paper, a method for designing segmented non-imaging concentrators with planar or tubular absorbers is proposed, where concentrators are represented as concatenated linear segments, forming a faceted reflector.

Abstract: We propose a computational optimization method for designing segmented nonimaging concentrators with planar or tubular absorbers. Concentrators are represented as concatenated linear segments, forming a faceted reflector. Ray tracing is then employed to compute the optical concentration ratio equivalent for a given concentrator configuration. The user may chose to systematically optimize the locations of the concentrator vertices or the interior angles between adjacent segments. Since gradient-based solvers become easily trapped in local minima for this problem, generalized pattern search is used to obtain optimized concentrator configurations. For a small and fixed number of segments, we show that optimized segmented concentrators preform better than an equivalent number of uniformly spaced segments of the collocated ideal shape. We also show that concentrators with a lower number of segments are more robust to manufacturing errors.