The Temporary Operation Your Program Says You Don't Have
Most of us learned to write procedures from other procedures. You inherit a binder, or someone hands you one, and the next procedure you write looks like the last one because that's what an approved procedure looks like. It's a fast way to fill a binder. It also carries every gap forward, and nobody notices a gap until a job shows up that the procedure never planned for.
The pumpout is one of those jobs.
A service tech shows up to change coalescers on a screw package at a cold storage plant. Before he opens the separator, the package has to come down. A proper pumpout gets the compressor as close to zero as the system safely allows. There are many ways to achieve this. How it gets done depends on the plant and the package.
Then he's handed the plant's procedures. The pumpout procedure says what most pumpout procedures in this industry say: stop the machine, isolate the compressor and blow it down into a bucket. Written broad enough to cover every plant; and sometimes right by accident. I've signed and reviewed plenty like it. I learned pumpouts as a service tech years before I ever wrote a procedure. I just matched the binder like everyone else.
So what is he supposed to do? He can follow the procedure, or he can do the job the way he knows it should be done, off the page. Plenty of good techs take the second option.
Some plants see this coming and tell techs not to touch setpoints at all. Those plants understand what a setpoint change sets in motion: an MOC, a revised procedure, training on the revision. So they haven't changed the procedure, and it's hard to blame them. They already paid to have their procedures written, and what came back looks like every other plant's.
1910.119(f)(1)(i)(C) lists temporary operations among the phases operating procedures have to cover. Most programs answer it with a sentence saying none are expected and anything that comes up will go through MOC.
Any plant that has pumped out a compressor this way has a temporary operation. The sentence is wrong on its face.
It also asks MOC to do a job it wasn't built for. MOC is for changes, and a temporary change has an end: (l)(2)(iv) requires the necessary time period to be addressed before the change is made. A setpoint move that comes back on every pumpout doesn't end. It recurs, and a recurring deviation is an operation. Operations get procedures.
Using MOC in place of that procedure leaves two options. Run a fresh MOC every pumpout for the same change, or skip it and move the setpoints without a record. It's usually the second one.
The pumpout procedure should say up front that it's a temporary operation. It takes protective setpoints outside their normal range on purpose, under control. If the pumpout is the only job at your plant that does that, the procedure can carry the temporary operation itself. If other jobs move the same setpoints, pull it into its own procedure and have each job reference it. Either way it's written down.
It needs to cover which setpoints move, how far they can go, who authorizes it, what gets watched while they're moved, how long they can stay there, and the step that puts them back.
"As close to zero as the system safely allows" needs a number, and zero isn't always it. The floor comes from the MDMT on your suction vessels. A common rating is -20°F at 150 psig, and -20°F on the PT chart is 3.5 psig. Your data plates give you the actual number, your PSI should carry it as the safe lower limit, and the procedure is written against it.
That changes the sequence. Pull the package down to the floor, then blow down what's left. The bucket stays in the procedure. It just moves to second.
The bucket step needs two details most procedures leave out. Secure the hose to the bucket; a pipe wrench lightly clamped does the job. And pull the hose out of the water as soon as the package is down, or water migrates back into the system.
While you're writing, define normal too. Write normal setpoints as ranges wherever the process tolerates it, 28 to 32 psig rather than 30, so an operator trimming inside the band isn't making an unwritten change. The safe upper and lower limits stay hard numbers. A temporary operation can take a setting outside the range but never past the limit.
The service tech knows the right method. The limits are yours — they live on your data plates and your PSI, and mid-job, your procedure is where he should find them. A procedure that says stop, isolate and blow down tells him nothing he can use, so he falls back on what he knows and you find out afterward, if at all.
A procedure that matches the job gives him your floor and gives you a record of what moved and when it went back.
Writing it is a change to your operating procedures, so it goes through MOC. Once. The people who run pumpouts get trained on it, once, and the training gets documented. After that, a pumpout inside the written limits is running the procedure. MOC comes back only when a job needs to go past what the procedure allows, or when the procedure itself changes.
That's the chain the no-touch plants are avoiding. Walked once, it's short, and it ends with the written procedure and the right way to do the job being the same thing.
Monday, pull your pumpout procedure. If it says stop, isolate and blow down, and your program says no temporary operations are expected, you just found your first one to write. Whoever copies your procedures next gets that one too.
Next: a simple setpoint change walked across the program map.