Engineers use lots of computer modelling tools. It’s our professional responsibility to understand those tools well enough to understand their limitations and determine when they are sufficiently accurate.
This accuracy (ie computer prediction matching reality) is essential to providing good competent advice. At the core of understanding are validation studies like the ones below: ASHRAE RP-1117 and the more recent BIM models for load calculations.
The core result from these studies was that using BIM models (such as the ones we build for our designPH studies) are sufficiently accurate. For the lightweight buildings we typically construct in New Zealand, the current heat balance methodologies are more accurate than the older RTSM method often used when predicting cooling loads. These results are not surprising but it is good to have the model prediction and measurements directly compared and validated for accuracy.
Do note that even with expert level models, and precisely known construction and corrected for weather, cooling load predictions within +/- 12% of the measured value are considered accurate.
Validating BIM Load Calculations, Kevin Lawson, P.E,. Member, ASHRAE, ASHRAE Journal, July 2022
Looking at validation to measured real world load results “Fisher, D., J. Spitler. 2002. “Experimental Validation of Heat Balance/RTS Cooling Load Calculation Procedure.” ASHRAE Research Project RP-1117, Final Report.”
The software predictions are conservative compared to measured loads but that’s true for most of them (with Heat Balance methods being the most accurate). In any case the point being made is using BIM for the geometry and then checking the thermal properties closely results in BIM models imported by gbXML or other methods very suitable for loads calculations.
“The results of the calculations can be viewed in Table 2. Software 1’s built-in load calculation tool, which uses an RTSM method, overestimated the cooling loads in the heavy building by 35.3% and the light building by 39.4%. This is approximately the amount of overestimation from the RTSM methods used in RP-1117 conducted by Dr. Fisher and Dr. Spitler. Software 2’s default engine overestimated the true tested loads in the heavy building by 28% and the light building by 37.5%. Software 2’s RTS (Heat Balance) engine was the most accurate engine, overestimating the heavy building by only 0.1% and the light building by 14.2%.
While these may seem like overly conservative results, remember that the test was set up to exaggerate the inaccuracies of the RTSM. All results generally agree with the results of the RP-1117 calculations.”