A main transformer for a large battery site costs several million dollars and takes more than a year to arrive. Once it is ordered, its details are fixed. If you find out later that you cannot test something about it, you live with that for the life of the site.
So I have one habit that people find odd. Before we order any large piece of equipment, I write the test we will use to accept it. I do not mean the specification. I mean the test.
A specification says what you want. A test says how you will know.
On a project in 2021 the specification asked for a transformer with temperature sensors in its windings. It arrived with them. On site we found there was no way to read two of the sensors without opening a sealed cabinet, and opening it would have cancelled the warranty. The specification was met. The transformer could not be tested.
If we had written the site test first, someone would have asked a simple question. Where do I plug in to read this? That question costs nothing in a meeting. It cost us nine days on site and a difficult conversation with the vendor.
If you cannot describe the test, you do not yet know what you are buying.
Writing the test first also changes the conversation with suppliers. A vendor who sees the acceptance test before signing knows exactly what finished looks like. A young engineer on my team, Ines, sent a test sheet to a switchgear supplier last year. They came back with three corrections to their own drawings before we had paid anything.
What goes in the test, before the order
What will be measured, and with which instrument.
The number that counts as a pass, with its tolerance.
Where a person stands and connects to do it.
Who witnesses it, and who signs.
None of this is clever. It takes about a day per major item. On Ostrand we wrote 23 of these sheets before the first purchase order went out, and I believe they are a large part of why the site energised early. Most delays I have seen were not caused by broken equipment. They were caused by working equipment that nobody could prove was working.

Linnea Voss
Energy systems engineer, Rotterdam