We had a VFD fail on a critical warm room application. The VFD does not have a bypass. I'm looking at ways to improve failure recovery time or reliability of the system. The AHU system runs continuously (24/7) and steadily at about 56 hz. I've set up some options in a report which I'll paste clips from. Rockwell Automation performed a failure analysis of the drive and listed possible root causes as overload or shock load heating of the IGBT or premature IGBT failure. One of my risk reduction options is to replace the drive with an across-the-line motor starter. I state that there are fewer risk-failure parts associated with motor starters than VFD and lower failure rates. I can't find any publication on failure rates or MTBF to compare motor starters to VFDs to substantiate this, just my experience and gut feeling.

Clips from my report attached.


Thanks for any input I get,
Jeffro

Clips_from_my_report.doc (28 Kb, 16 downloads) Clips from my report

I agree with you 100% that a simple electromechanical starter will be more robust than a solid state VFD as far as the component itself goes. I wish I had stats to prove it but I don't. There might be some loss of efficiency in going to a starter (depends on application) but it sounds like reliability is a bigger concern than efficiency.

Any modification will require a little bit of review. I would say there are two things to double-check along the way:
1 - Can this motor safely handle starting of this load on this power system direct-on-line? As you are probably well aware vfd makes the start easy on the motor but accross the line start can be tougher on the motor. High load inertia, high load torque during start, low voltage, high power system impedance can push the DOL motor time/current starting profile towards or beyond the motor manufacturer's safe time/current profile. If you are adding thermal overloads, the starting could be studied in conjunction with sizing the overloads.

2 - Can the mechanical system satisfactorily unload itself to handle possible long periods at low heat load? Some mechanical systems don't like low load without a vfd to slow the machine down to match the load. One example is centrifugal pump which doesn't like operating far below BEP. Another example (possibly more relevant) that we have at our plant is centrifugal air conditioning compressors (York 150 ton unit). When there is a low heat load, these things vibrate like crazy due to suction conditions (something similar to a "rotating stall").

Thanks. I wish there was a place to go to get general electrical componenet reliability stats for items like VFDs, motor starters, transformers and such.

This is an AHU system so starting torque is very low. We proved the system would handle the motor running at 60Hz with the unexpected failure. I'm thinking we'll just go with the bypass, but management challenged me for other options. We are good at thinking of redundant systems for critical pumps and such, it got by us to consider bypasses for these two AHUs.

Something on failure data came to mind with your reply.

We had a Liebert UPS fail some time ago. In this instance, the trigger board or output SCR failed catestrophically shutting down the UPS AND tripping the bypass breaker. No power was available until the on-call elctrician arrived to troubleshoot and reset the bypass. I tried to get failure history on the model from Liebert. I know they have the data because that's one industry where most of the maintenace is contracted to the manufacturer or mfgs reps. They had all the maintenance, callout history and parts replacement records for our plant. But they would only give me brouchure-type calculated MTBF which doesn't really help me predict upcoming years of service reliability. I told them to "think Consumer Reports". What can we expect for circuit board and power component failures based on past history of this model in its 8th thru 12th year of service. No help from them. They don't appear to want to let true failure data out to customers on a product that is supposed to be there when normal power fails. Or, perhaps they don't know what to do with all the data they do have. I guess if the electrical maintenance groups want failure history benchmarks we'll have to build it ourselves.

Thanks for the input,
J-