Comparison of view factor computation modules
This page compares VFRTGPU to the HEMIVIEW and VUFAC modules.
- VUFAC module
- Computes the ray-traced, non-ray-traced, and other types of view factors, using machine's central processing unit (CPU). The VUFAC module calculates view factors with a combination of the contour integration and Nusselt sphere methods, using automatic elemental subdivision algorithm when shadowing occurs. View factors between elements are calculated one at a time. If shadowing occurs, the number of calculations required to compute the view factors varies roughly as the cube of the number of elements, which is very CPU intensive. Depending on the type you select, and number of view factors calculated by the solver, the view factor computations can be CPU intensive using the VUFAC module.
- HEMIVIEW module
- Uses the machine's graphics card in conjunction with the open graphics library (OGL) to compute the blackbody view factors using an imaginary hemicube. The HEMIVIEW module is much faster. It uses computer graphics techniques to graphically render the radiating elements onto the five sides of a hemicube (a half-cube) placed at the CG of an element, called the emitting element. The view factors from the emitting element to the other elements, called the receiving elements, are determined by post-processing the graphical images of the receiving elements projected onto each of the hemicube sides. The hemicube method is fast for two reasons: the number of renderings is linearly proportional to the number of elements, and the rendering calculations are performed with specialized fast graphics cards. The accuracy and computation time of the hemicube method is limited by the pixel resolution of the drawing area. However, it only supports blackbody view factors.
- VFRTGPU module
- Uses the Monte Carlo method to compute both the blackbody view factor and ray-traced radiative conductances using the machine's graphics processing unit (GPU). Depending on the number of radiative conductances that are computed in a radiation simulation, the GPU-based Monte Carlo method can perform approximately an order of magnitude faster on a supported GPU hardware compared to the other modules that compute the same amount of radiative conductance. In ray-tracing, many rays are randomly cast between elements participating in the radiative exchange. The rays follow the path taken by the radiation. The software launches several rays from each element using the Monte Carlo method.