Question-and-Answer Resource for the Building Energy Modeling Community
 | 1 | initial version |
TLDR: I'm in complete agreement with Daric (he got my vote!), but don't see this as a binary question with only one right answer so I'm contributing some additional perspectives. I'm going to diverge a little from LEED / Appendix G and the 40% rule to tackle the topic of window distribution during model development more broadly as well, so I'm not expecting this to get upvoted as a best answer - just adding to the discussion:
I typically also take the latter approach, adjusting each window's area in-place thereby retaining the proportionality of the window distribution by any metric: facade/surface/orientation. Appendix G contains fuzzy phrasing (subject to interpretation) with regard to maintaining the proposed cases' distribution proportion, and LEED/GBCI have interpreted that to mean you should maintain WWR by major orientation (still a fuzzy line for complex floorplates)...
However my reading of Appendix G is that you CAN ultimately take either approach, and this decision is a good framework to consider for workflows both inside & outside of the confines of Appendix G. At times (early schematic/box-model studies come to mind), you might not know anything about the eventual window distribution (or what you do know is subject to major revisions), so uniform distribution could make the most sense when you don't have a better guess.
More often, you as the modeler can make the informed decision that a strategy of uniform distribution for geometries is a "fair enough" estimation for model development (where the actual distribution is or could be relatively uniform), to allow you to smartly invest more of your limited development time in more pressing/important areas of the simulation. Consider: Appendix G has nothing to say about a "distribution accuracy threshold" in discussing the baseline model's fenestration requirements - it is more broadly concerned with ensuring the baseline/proposed case are on equal footing (at least at/below the 40% mark).
In practice however, I have found aiming for "equal distribution" by facade/surface (by ensuring glass in on each opaque air-facing exterior surface) can become problematic, particularly in the proposed case.
One reason is this can introduce "disconnects" with details of the proposed case HVAC system(s) design: You can end up for example with zones/systems that weren't designed to handle the new skin/solar loads introduced by "spreading the windows around," and conversely systems/zones designed to handle a lot of glass are now "oversized" in your model's representation. This leads to concessions either with more autosizing and/or more unmet conditioning hours taken as an acceptable threshold (potentially butting heads with App. G's thresholds).
Minor reasons may include that the models can end up looking substantially different from the design (which is a big deal for some individuals), and also it can make for additional necessary documentation/explanation in a review process.
I hope this helps suggest a few new ideas - In my mind there's a time & place for "approximated" uniform window distributions, and it's better to keep such approaches in the back of your mind even when it isn't "standard practice" for the odd occasion where it could be "the better choice!"
