Abstract
This article introduces a methodology for designing the geometry of diffuse-walled radiant enclosures through nonlinear programming. In this application, the enclosure is represented parametrically using B-spline curves, while the radiosity distribution is solved by infinitesimal-area analysis. The enclosure geometry is repeatedly adjusted with a gradient-based minimization algorithm until a near-optimum solution is found. This approach requires far less design time than the forward "trial-and-error" methodology, and the quality of the final solution is usually much better. The methodology is demonstrated by optimizing the geometry of a 2-D radiant enclosure, with the objective of obtaining a desired radiosity distribution over a portion of the enclosure surface.