STUDY OF BUILDING SURROUNDING LUMINOUS ENVIRONMENT USING HIGH DYNAMIC RANGE IMAGE-BASED LIGHTING MODEL
Open Access
- Author:
- Sadeghi Nahrkhalaji, Reza
- Graduate Program:
- Architectural Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- March 02, 2017
- Committee Members:
- Richard G. Mistrick, Dissertation Advisor/Co-Advisor
Richard G. Mistrick, Committee Chair/Co-Chair
Kevin W. Houser, Committee Member
Stephen Treado, Committee Member
James L. Rosenberger, Outside Member - Keywords:
- Architectural Engineering
Radiance
Daylighting
Simulation
Level of Detail
LOD
Exterior
Surround
Tree
Landscape
Lighting - Abstract:
- The accuracy of daylighting simulations for the built environment depends on the way its surround is represented in the simulation model. The common modeling practice of many designers is to create an exterior surround with a flat ground surface with little or no external architectural or landscape elements. In this dissertation, this abstraction and a number of different levels of detail for modeling the exterior surround were investigated by comparing real-life empirical measurements with their corresponding simulation results. A total of 13 different exterior views were each studied under 15 different sky/seasonal conditions to investigate the contribution of the surrounding features to the daylight delivered inside a small office. These features were classified into one of four different groups - sky, architectural structures, standing vegetation, and horizontal ground. Five of these locations were chosen for daylight simulation by applying six levels of detail to the exterior surround in a simulation model. The results reveal that ignoring exterior surround objects (common practice) yielded an average 67.9 percent error, while a high level of detail (which included high polygon vegetation representations) resulted in only an average 11.3% error. The LM-83 modeling recommendations resulted in an average of 22.8% error. The surround landscape highly influences the quantity and quality of daylight in an interior space, and is often excluded from daylighting simulation mostly due to the complexity of 3D geometry of trees and lack of efficient and validated modeling technics. In this study, different methods of representing trees in a daylighting simulation model are investigated by comparing the simulation output to real life empirical measurements and computing the associated errors. Eight species of tree were studied in a natural and relatively isolated setting by collecting photographic (HDR; High Dynamic Range) and photometric (luminance, illuminance and irradiance) data at 4 sky conditions and 8 view angles. A daylight simulation model was created in which trees were represented with high-poly and primitive geometry using a range of reflectance-transmittance material properties. The results revealed the average simulation error for representing trees as primitive geometry (LM-83 recommendation) was 120.7% (RMSE; Root Mean Square Error). Additionally Plastic and Trans material definitions were considered in the simulation model and a Response Surface Model (RSM) statistical design was chosen for data analysis. The findings showed the simulation error could be lowered to 52.8% with proper material definition for the primitive geometry and 32.9% if a high-poly 3D geometry was applied.