Viewfactor > Theory and Computational Limitations > 8.4 Obstructions
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX''">XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX''">   
8.4 Obstructions
It is understood that the finite element discretization of the model geometry represents the finest geometric detail available in the model. Therefore, surfaces are not subdivided beyond the nodal subareas for obstruction checking. This restriction on obstruction checking may be viewed as a bit too severe. Different schemes have been tried. The error introduced by this discretization of the obstruction checking process was compared with the error introduced in the viewfactors by the finite element discretization. Considering the cost in computer execution time for a finer obstruction discretization, it was better to limit the obstruction checking to nodal subareas.
You have control over the accuracy of the obstruction test in the usual way that accuracy in a finite element model is controlled (i. e., by refining the finite element mesh). Thus, if more accuracy is needed in the obstruction checking, you should refine the finite element mesh in the regions where greater accuracy is desired.