← Field notes · Sheet D-99
The archive
Everything, oldest kept alongside newest. Nothing here is deleted when it stops being current — corrections are added to the note instead, so the record of what I thought at the time survives.
2026 5 notes
The ΔT you assume is the plant you buy
Airflow is inversely proportional to delta-T, so the number you assume across a data hall decides how much cooling plant you buy. Most take it on faith.
Colebrook, and why I stopped using the explicit fits
Swamee-Jain is a good approximation and it is free. Iterating the real Colebrook equation is also free, converges fast, and removes an error term.
Rule #1 is already doing the work
Most drawings I review are covered in flatness and straightness callouts that the size limits already enforced. Here is what Rule #1 actually controls.
Sixty per cent out of the build time, and nothing given back
A redesign that took 60% of the manufacturing time out of a heat exchanger line, with none of the saving bought by accepting a worse machine.
Thermal stress does not care how long your pipe is
Length appears in the expansion equation and vanishes from the restrained stress equation. That cancellation is why shortening a hot run does not help.