Viewfactor > Theory and Computational Limitations > 8.5 Computational Limitations
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX''">XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX''">   
8.5 Computational Limitations
There are three areas in which we know the computational powers of the Viewfactor program are limited. These are not severe limitations and most users would probably never discover them. However, in the interest of completeness, they are explained in the following sections.
Grazing Incidence of the Intersurface Ray with the Surface
The grazing incidence of an intersurface ray with a potentially obstructing surface may not be correctly determined. This is due to the finite precision of computer arithmetic. There is no hard and fast rule for when this problem will occur. Generally it is not a problem unless the incidence with the surface is somewhat less than 10-5 radian for computations on typical 32 bit computers.
Spatial Resolution
The obstruction checking algorithm in Viewfactor is unable to detect potential obstructions which are very near one of the surfaces for which the view between is being checked for obstructions. Here “very near” means that the distance from the potential obstruction to one of the surfaces in the viewing pair is less than about one five-thousandth of the distance between the pair of viewing surfaces. This situation is shown schematically in Figure 8‑1.
Figure 8‑1 Spatial Resolution in VIEWFACTOR
Extreme Scales
Models which have very large or very small numbers for their dimensions will cause an arithmetic overflow or underflow condition to occur in Viewfactor. The approximate limits at which this occurs depends on the dimensionality of the model. For the 2-D XY model, the limits are approximately the square root of the largest and smallest numbers representable in the default FORTRAN single precision variable. For the 2-D RZ axisymmetric model, these limits are approximately the cube root of the largest and smallest numbers representable in the default FORTRAN single precision variable. For the 3-D XYZ model, these limits are approximately the fourth root of the largest and smallest numbers representable in the default FORTRAN single precision variable. If the user’s model has dimensions which exceed these upper or lower limits it will be necessary to scale the model by a suitable factor in order to avoid arithmetic overflow or underflow in Viewfactor. Viewfactor does not check for overflow and underflow conditions, since for most users this will not be an issue.