I have a Dynamo workflow established for visualization of Air Terminals. The end product is a visual representation in section view:
The logic isn't too crazy - use the Room Calculation Point coordinate to determine which face of the air terminal is 'exposed' to the room, filter whether I care about the bottom edge of the face or the top edge of the face for worst-case visibility (in this case, there's a 'shelf' that obscures the bottom, so I care the most about the top edge), generate 1 degree offset vectors from horizontal down to vertical, and then shoot raybounces and see what they hit.
For a while I've wanted to expand this idea towards roof-mounted equipment, but I'm running into a few general problems:
- The above method is inherently a single-plane analysis. Because of this, it's programmatically light (91 raybounces per top corner of the air terminal face). At the minimum, expanding this to a rooftop unit would require 4-direction analysis, assuming a rectangular profile building, so now there's at least 4 times as many geometry checks.
- For irregular building geometry (as in, not the most basic rectangular flat-roofed warehouse), I probably shouldn't just rely on a section cut against the normal of the wall face adjacent to the equipment, I should sweep horizontally as well. However, adding a 180 degree horizontal sweep to the 90 degree incremental downsweep ups the raybounces per point to 16,200.... meaning 1 degree increments become programmatically heavy.
- Not all rooftop units are conveniently shaped like rectangles, or have generally flat surfaces to pull from for vector analysis. Exhaust fans tend to be circular domes, upblast fans might have vent stacks for high discharge, kitchen untempered ventilation units have the largest tapered weatherhoods for some reason, some times equipment is L-shaped, etc....so if I use Dynamo to grab the surfaces of the family, I will guaranteed get more than 6 cubic surfaces, I'll get dozens (some of which can very easily be curved).
The method I've come up with so far is to convert equipment to their bounding boxes - it doesn't match the equipment size appropriately (I'll get false positives on visibility), but doing so means going back to a 6-faced box where the 4 top corner points are well-known and I can start shooting angles off those corners.
I don't have a good solution for the horizontal sweep, except to reduce the fidelity by widening the increments. But, for each 'slice' of the sweep I can pick the closest visible spot, and use the collection of closest points to build a filled region in plan view.
The least computationally-intense method would of course to be for me to just go into a section view, draw the sight line using the equipment + roof corner, and just call it there. But I'm trying to get this to a point where it's lightweight enough to implement as part of routine project QC (basically if it can finish while coffee brews)
Any suggestions on a way to pre-filter the number of checks needed, or if I'm missing an easier workflow?
