The radiation enclosure concept is fundamental to the analysis of thermal radiation problems and also to the techniques used to model the problems in Viewfactor and Patran Thermal. Therefore, it is important to understand the concept, not only as it is classically applied to thermal radiation problems, but also as it is used in creating the computer model of the thermal radiation phenomena.

For our purposes, an enclosure is a collection of thermally radiating surfaces which have the potential to see each other (radiate to each other), along with open areas which can potentially be seen by the surfaces and participating media or ambient nodes associated with these surfaces. From this definition you may infer that there are a large number of enclosures possible in even a simple model. It is up to you to select appropriate enclosures for the particular thermal analysis problem at hand. In most cases, appropriate choices of enclosures are natural and obvious from the model geometry.

Surfaces in different enclosures do not have the potential to radiate to each other. In addition, a surface in one enclosure does not have the ability to obstruct the view between a pair of surfaces in another enclosure. These properties of enclosures are exploited in the Viewfactor program to reduce the CPU time required to analyze the viewfactor problem. Surfaces not in an enclosure need not be considered as potential obstructions for that enclosure. Surface pairs that are not in the same enclosure need not have calculations done for them.

The enclosures are also used for defining portions of the model over which the viewfactors from one surface to all other surfaces it sees are summed. These sums have two uses:

Enclosures are made distinct by giving each different one a unique identification number. This ID number is associated with all of the surfaces in its enclosure. The enclosure ID is assigned in the VFAC LBC form as described in Viewfactor (p. 126) in the Patran Thermal User’s Guide Volume 1: Thermal/Hydraulic Analysis.

Wavebands are used when the thermal radiative material properties depend on the wavelength of the radiation (they may also depend on time and temperature). The concept of wavebands and their proper use in modeling the thermal analysis problem are explained in Patran Thermal TEMPLATEDAT Files for Surface Property Description, 52.

If you do not need to use spectrally dependent material properties to adequately model the problem, then this subsection need not be understood. Note, however that you will need to understand this subsection in order to understand all of the examples in the next subsection, list2+s of the Use of Enclosures, 26.

Enclosures and wavebands (see page 52) have a special relationship. The wavebands associated with each surface in an enclosure must match exactly the wavebands of every other surface in that enclosure which the first surface can see.

This holds true when every surface in the enclosure has the same wavebands. It is recommended that all surfaces within an enclosure have the same wavebands. For two surfaces in an enclosure, if the surfaces can see each other and the wavebands are not identical, then a fatal error will occur when you attempt to make radiation resistors for the model. This error cannot be detected until after the viewfactors are calculated (the most CPU intensive part of the Viewfactor analysis). It is a mistake that you should avoid so that you do not have to redo the viewfactor calculations.

For the wavebands associated with two surfaces to be identical is meant:

Figure 3‑1 through Figure 3‑6 show schematically some examples of enclosures and their use in modeling thermal radiation problems.

Figure 3‑1 shows a rectangular cross section of either a hollow torus in axisymmetric space or of a long tube in Euclidean space. The section is hollow on the inside and the interior surfaces are thermally radiating, as indicated by the arrows attached to these surfaces. These surfaces are in the same physical enclosure. It is best to use the naturally occurring enclosure as the enclosure for the computer model. It does not make sense to divide these interior surfaces into two enclosures. Some pairs of surfaces would then be in different enclosures and no viewfactors would be calculated for them. This would not be correct, since in this model all of the interior surfaces can see each other.

Figure 3‑2 shows a cross section with two cavities. Each cavity is filled with a different participating media, and thus there is a different participating media node for each cavity. Two enclosures are needed to keep separate the two different media. Different enclosure ID numbers are assigned to each cavity.

If the media in each cavity had been the same (or if there had been no participating media), it would have been acceptable to give both cavities the same enclosure ID. This is not preferred, because the cavity surfaces naturally fall into two groups. Any surface in one group cannot see any other surface in the other group. By identifying these group through different enclosure IDs the CPU time required to analyze the viewfactors will be reduced.

Figure 3‑3 shows an object which is exposed to ambient radiation nodes above the object and below. Due to fortuitous geometric circumstances in the model, this object could actually be modeled as one enclosure. This is not recommended. A better approach would be to divide the model into two enclosures. One consists of the upward facing cavity and surfaces and the second consists of the downward facing cavity and surfaces. The astute reader may also observe that the model can be properly divided into even more enclosures. Recognize that the horizontal surfaces not in the cavities see nothing but an ambient node. Thus each of these surfaces could be identified as a unique enclosure. Doing this in a model where each of these surfaces was divided into many elements would result in substantial saving of CPU time to perform the viewfactor analysis.

Figure 3‑4 shows an example where three enclosures have been identified. Other groupings of the surfaces, media node, and ambient nodes are possible. The reader who wishes to master the art of identifying enclosures in the thermal analysis model is urged to devise some other groupings of the surfaces and nodes into enclosures and then ascertain their correctness for modeling this problem. For the correct enclosure groupings, what are the advantages and disadvantages?

Figure 3‑5 shows an object made of two different materials, each having different wavelength dependent surface emissivity properties. This cavity is correctly modeled as one enclosure, but the wavebands in the enclosure must be the composite of all the waveband transitions for the various materials’ wavebands. Thus, there are six wavebands in this enclosure and each material will have to be described in terms of these six wavebands, not in terms of the two wavebands of surface one, for example.

Figure 3‑6 is an introduction to the kinds of complex problems which may be modeled with enclosures and wavebands. It consists of a cavity surrounded by two different materials, each with different wavelength dependent emissivities, and a partition vertically through the center of the cavity and at the boundary between the different wall materials. This partition is transparent to some wavelengths and opaque to others as shown in the transmissivity graph. For the spectral region where the partition is transparent, the cavity may be modeled as one enclosure, since the radiant interchange is unimpeded by the partition which is transparent to radiation in this spectral region. For the spectral region where the partition is opaque, the right and left halves of the cavity cannot see each other, but the surfaces in each half can see the right and left faces of the partition, respectively. Thus we model the right and left half cavities along with their associated face of the partition each as an enclosure. The wavebands for these enclosures are shown at the lower left of the Figure 3‑6.

Note that the wavebands for enclosure 1 end at lambda sub 4, and the wavebands for enclosures 2 and 3 begin at lambda sub 4, the transition wavelength between transparency and opaquecy for the partition material.