Hey all, booster pump station, 50hp motor on a VFD. Fitted a loop powered vibration sensor on the motor's drive end bearing housing, 4-20 mA, velocity, 0 to 25 mm/s RMS. It's in the PLC and on the HMI and reads about 2.1 mm removed link I verified the loop at the panel, 4.0 mA with the pump off, scaling checked.
Now nobody will give me an alarm number and I'm a bit stuck. The sensor vendor says use the ISO 10816 table. Maintenance says alarm at double whatever it reads today. The ISO table for our motor size puts 2.1 in the middle band and the next band up starts at 2.8, which seems awfully close for a pump that's perfectly happy.
What do people actually set on these? Would you really alarm at 2.8 on a pump that's sat at 2.1 since the day I fitted it?
Loop's proved, good. Does the reading change with speed? A booster pump at two setpoints won't read the same. What do you get at normal speed and at full speed, both numbers?
It moves, yes. 2.1 at the normal 60 percent speed. 3.4 at full speed when the second zone calls. So the number depends on what the pump's doing at the time, I guess.
That's why one number fails. 3.4 at full speed is already the next ISO band and it's a healthy pump. Log it to the historian at 1 min with the speed reference beside it. Run two to four weeks. That's your baseline, per speed band. Alarm on a rise from the band's own baseline, something like 50 percent, and on rate of change, a climb over a few days. The ISO zones are a sanity check that your baseline isn't already a sick machine. Not the alarm.
Deadband it, and a start inhibit.
The baseline then alarm approach is in this one: https://plctr.com/utilizing-plc-for-condition-monitoring-and-predictive-maintenance-in-manufacturing/
And no alarm in the first ten seconds after a start.
Three weeks logged. The 60 percent band sits 1.9 to 2.3. Full speed band 3.2 to 3.6.
Set a high alarm per band at 1.5 times that band's average, a rate alarm if the 24 hour average climbs 20 percent inside a week, and the ten second start inhibit. Nothing has tripped, which is the point.
Landed on a baseline per speed band, alarms on the rise and on the rate, and the ISO table as a sanity check and nothing more.
What I don't have is any idea what a real bearing failure looks like on this loop, we've never had one, so 1.5 is a guess until we do. Thanks scott, marek.
The first real failure sets the number. Keep the log until then.