We have several mag drive pumps in a project. I am looking for advice on how best to condition monitor these self-contained pumps.

Ted,
I haven't specifically worked on these pumps, but I have worked on a pump with a magnetic coupling (MagnaDrive). A motor-magnetic coupling-centrifugal pump is a basic machine train. Here is a link to mfg:
http://www.sundyne.com/index.cfm?go=/displayProduct.cfm&type=5&page=59

Are you expecting something unusual about these pumps?

Walt

Many are not spared and failure can shut us down. So we need early warning of problems. I understand how they work, but I was hoping to find someone with experience or tricks on how to detect early problems with them.

Hey Ted
I monitor several of these pumps in a petro-chem plant. The best set-up we use is trend data of the high frequency detection in acceleration. This has proven to be the best for catching failures.

Hey,"vibration guy" ... please tell me more! I had not heard that technique being used on this type of pump before. Are you trending data with on-line installed sensors and continuous monitoring, or are you using a walk-around method with portable analyzer and magnet-attached accelerometers? Like a CSI analyzer and "Peak Vue"? What is "high frequency?" What kind of data or amplitudes says "problem developing" on these pumps? Feel free to email me directly at S-Ted.Myrick@USA.dupont.com

Thanks in advance. Ted

Hello this is Keith Bluestein from Beran Instruments. We mfg condition monitoring equipment. Feel free to contact me to discuss your pump requirements. I may be able to steer you the right direction. 804-290-7871.

Thanks to you all. The note from Walt Strong led me to Paul Ritger, Sun-Air (215-657-9800),which led me to a low-cost "emotron" power monitor which looks like it will give me "eyes" inside the sealed pumps.

Ted,

I don't know if you've found what you're looking for yet, but the company I'm with created a technique that trends features of acceleration and current data to estimate health of the rotating machine. In one application, we identified a bearing problem in a gearbox 9 minutes before it destroyed the machine; when other techniques only ID'd it 30 seconds before the event.

We've used it in diagnosing mechanical and electrical faults, so the pump problem is a logical extension. We have stand-alone hardware that can be connected in an existing system, or deliver just the algorithms that perform the diagnostics. If you're interested, call me at (585) 424-1990 x129.

Jeff

An issue with mag pumps is, "what's in the medium being pumped?". If they are batch pumps you may want to monitor for particles that can be attracted by a magnet as those particles will collect in the pump and cause catastrophic failure very rapidly as a rule.

Else monitor as a more-or-less normal pump and include spectra @5kHz if AF brgs >3,600 RPM.

I agree with Sam and only want to add that the drive shaft is a short stub shaft with a large overhang. This is sensitive to the foundation stiffness. With some bad luck, the critical speed is too near the running and that will kill the pump soon. Alignment to the motor should be perfect and the rubber sleeves in the coupling in good condition.

The pump side should be looked after just like all canned pumps.

I have successfully used acceleration and velocity to monitor canned motor and mag drive pumps. For the mag drive, use standard bearing failure and unbalance spectral set up and time wave to watch for the mag carrier support bearing failure. For the driven bearing/bushing, remember it behaves like a journal bearing and use the journal bearing spectral and time wave set up with one exception, reduce all analysis and alarm/danger values by 1/10th. Look for time wave and spectra evidence of rubs (sub-harmonics) whirl and whip and unbalance on the driven bearings. Look for any evidence of low flow or cavitation in the process. For continuous monitoring add a "power meter" to meausure the KW load on the motor. Low KW compared to pump full load indicates run dry, high KW indicates overloading due to rubs, failing bearings or other process parameters that increase the load - viscosity, solids. For more detail see June 4-6, 1991 Vibration Institute Proceedings, or Texas A&M Turbo, 13th Pump Users Symposium, Short Course 2, March 4, 1996.