Thomas L. R. Davenport

Optimization for illumination systems: the next level of design

Computers are routinely used to design illumination systems. Automating the design process is enhanced through the use of optimization procedures. This paper describes some of the underlying illumination optimization fundamentals: parameterization, merit functions, and optimization algorithms. Numerous interesting examples of illumination design problems that benefit from optimization are shown. These examples illustrate illumination optimization through use of ray aiming, computing illuminance using flux tubes, and computing illuminance using Monte Carlo simulations.

Automated design of a uniform distribution using faceted reflectors

William J. CassarlyStuart R. DavidDavid G. JenkinsAndrew P. RiserOptical Research Associates3280 E. Foothill Blvd.Pasadena, California 91107E-mail: billc@opticalres.comThomas L. DavenportUniversity of ArizonaOptical Sciences Center1630 East University Blvd.Tucson, Arizona 85721Abstract. Faceted reflectors are a ubiquitous means for providing uni-form illumination in many commercial lighting products, examples beingnewer flashlights, department-store display lighting, and the faceted re-flectors found in overhead projectors. However, the design of facetedreflectors using software has often been more limited by the tools avail-able to design them than by the imagination of the designers. One of thekeys to enabling a broader range of design options has been to allowmore complex surfaces using constructive solid geometry (CSG). CSGuses Boolean operations on basic geometric primitives to define shapesto create individual facets. In this paper, we describe an improved fac-eted reflector design algorithm and use it to create a wide range of CSG-based reflectors. The performance of various reflectors is compared us-ing a Monte Carlo ray-trace method.

3D NURBS representation of surfaces for illumination

Use of NURBS surfaces to create facets on a reflecting surface will be considered. Specifically, the design of a reflector that generates a circular illuminance pattern will be investigated. Important considerations are: choice of variables used to represent a NURBS surface, total number of variables, parameterization and/or knot vector specification, and where to use algorithmic vs. optimization approaches.

Optimization for efficient angle-to-area conversion in illumination systems

Angle-to-area converters are a key topic of illumination design, and much work has been done in this area over the last 30 years. However, relatively little work exists in the literature in which these converters have been designed using optimization techniques. The present work takes a fresh look at some angle-to-area conversion problems using optimized, circularly and non-circularly symmetric surfaces.

Non-rotationally symmetric mixing rods

Uniformity remains a central topic in illumination system design and mixing rods provide an effective means to providing uniformity. Typically, flux enters one end of a mixing rod and the flux exiting the other end provides improved spatial and/or angular uniformity. We investigate the use of mixing rods with rippled surface structures to provide enhanced uniformity.

Optimizing angle-to-area-converting, light-piping systems using surface features

In this paper, we investigate how to improve the uniformity of the spatial distribution of the illuminance at the output plane for angle-to-area-converting, light-piping systems through the introduction of cyclical surface features. A superposition approach is used for studying uniformity. Improvements in uniformity for square-to-circle and rectangle-to-circle lightpipe configurations are demonstrated for a short package length.