Building simulation
By the time it reaches a solver, it is a different building.
No analysis tool can read a building as it was drawn. Energy needs thermal zones, fire needs compartments, structure needs a load path, acoustics needs surfaces — none of which is in the drawing. So each field rebuilds the building in its own tool, and the copies drift: the same wall sits in two places, the same room holds two volumes, and the results stop describing one building. Perrette does that translation once, from the geometry that already exists, and gives the same building to all eight.
The software is not released. Until it is, this site publishes the methods underneath it and the reference material it is built on.
Eight answers
A building has no single performance.
They all start from the same building now — and they still disagree. That is not a defect waiting to be reconciled: acoustics and energy are asking different questions of the same walls, and a light partition changes everything for one and nothing for the other. What one shared model removes is the drift between eight copies of the geometry. What it cannot remove, and should not, is that eight questions have eight answers, each true at its own resolution and under its own assumptions. The mistake is expecting one number.
Optimization is not on this list. It is not a reading of the geometry but a loop that replays the eight, so it sits on Tools, after them.
On this site
Four places to read.
Tools
What each tool is for, the problem it solves, and what we require of it — what we refuse to deliver, and what we guarantee.
Articles
One a month, in a scientific register. What was found and what it is worth, never a recipe for reproducing it.
Building science
Reference documents on how a building actually works, written from the plant room up: the physical thing first, the job it does second, the model last.
Perrette
Where the name comes from — Auguste Perret, and the rebuilding of Le Havre — and what we took from it.