Performance Evaluation Of Lighting Simulation Tools By Comparing Physical And Digital Measurements

Lighting simulation tools have been used for many years as the base of research projecton different areas such as aeronautics, automotive and architecture. Literature shows that inrecent years research done using light simulation tools is increasing, as the potentialapplications of such tools is said to become a standard that could overcome the usage ofphysical layout. The Recently acquired capability of lighting simulation design toolsallowsindustries and researchers to recreate a physical world (such as a Mock-Up) withoutthe actual need of building the real scene model, that represented the greatest characteristic ofsuch simulation tools.Then Physical mock-up (PMU) is the most widely applied method used to validate thelighting design in transportation systems. However, Digital Mock-up (DMU) is been tested asa possible replacement to the more expensive and more time consuming Physical Mock-Up.Therefore the key issue in such a context is, are the lighting simulation tools really reliable?Therefore, this thesis was proposed aiming to validatethe performances of differentlighting simulation tools and to provide a configuration for integrating those tools intolighting design with the following methodology.First, PMUs are constructed to represent real objects existing in a transportation system,and they are used as basic references to take the real measurements and real images. Thesame lighting configurations used on the PMU, will be set in different simulation tools.Four different cases have been designed in order to evaluate the performance of lightingsimulation tools in the most relevant way. Every aspect of the software will be tackled basedon those different scenarios, which include different environments and different kind of light.The simulation tools will provide three different results: luminance distribution,illuminance distribution, and photorealistic rendering. For luminance and illuminancedistributions, the simulated results are compared with the measurements taken from a reallighting environment; they are analyzed quantitatively. Regarding photorealistic renderingresults, the simulated images are compared with the images taken by digital camera on thePMU, and the degree of similarity is evaluated by using questionnaire to get thesubjective opinions qualitatively. Besides the quantitative evaluation (objective) and qualitative evaluation (subjective),the other aspects of the simulation tools are evaluated as well, such as CAD integration, andease of use.After software survey, three simulation tools were used to be further assessed in thepresent research. These are: DIALux, Radiance and SPEOS OPTIS. All the software toolswere deeply analyzed by the following two categories: feature comparison and ease ofuse.Three different lighting design applications software were been selected differ greatly.DIALux is a freeware and was found to be user-friendly, whereas Radiance needs someprograming skills. SPEOS OPTIS is the leader of lighting design field and because of thisfact its cost is high, furthermore it requires CAD knowledge.Main problems considered on lighting simulation tools are: time consumed to organizeand set-up parameters defining CAD model, and the optical characteristics in the digitalenvironment; time processing for simulation.After comparing the quantitative performance and qualitative performance of thesimulation tools, it was easy to draw the conclusion that SPEOS has greater performance inboth aspects.Then, it should be pointed out that the other two software tools also had greatperformance under certain cases. Radiance and DIALux have relative good performance forilluminance simulation in the cabin scene and under LED lamp. That means both softwaretools might be available for evaluating illuminance at first. Some significant differences werefound also during the comparison, and those points could generate great influences on thefinal comparison result. However, the absolute errors have not been shown in this thesis, butwe have found that the relative difference is high, but the absolute difference is lower than10for illuminance or10-cd. for certain points. These differences were normally found inrelative dark regions or close to the light source.There are lots of error sources that influence the final accuracy achieved by thesimulation, and they are geometrical errors, luminous intensity errors, BRDF errors,positioning errors. Those errors should be limited or controlled strictly in order to achievehighly accurate results.