GPU-based radiation calculation methods
GPU-based calculation methods for radiation exchange use the graphics processing unit (GPU) to perform accelerated radiation computation.
In GPU ray tracing, each element is assigned a unique ID, which serves as the random seed
for Monte Carlo simulations. This ensures that each element generates consistent and
reproducible random numbers, which is crucial for maintaining accuracy and reliability
in the simulation results. The thermal solver supports the following GPU-based methods:
- GPU-computed view factors
- Computes radiative view factors on the GPU, based on the modified Monte Carlo ray tracing method. This method supports diffuse emissivity and diffuse reflectivity thermo-optical properties. Direct view factor calculation is much faster than GPU computed ray tracing. However, this method cannot model the effects of real surfaces with specular reflection, transmission, and solar properties. If the model contains specular or transmissive properties, the thermal solver treats them as fully diffuse.
- GPU-computed ray tracing
- Computes ray-tracing on the GPU based on the modified Monte Carlo ray
tracing method. This method supports advanced thermo-optical properties such
as specular reflectivity, transmissivity, and refraction. However, tracing
each ray path until the energy is below a predefined threshold makes this
process more computationally expensive than the GPU-computed view factors
method. This method does not support temperature or time-dependent
emissivity.
For radiation boundary condition, you can decide to compute only view factors using the ray tracing method. Once the view factors are available, the thermal solver then uses them to calculate radiative fluxes and radiative conductances as temperatures evolve without retracing rays at every iteration.
