Hello all,
As you can see we have a major problem. this compressor has just 1400 hrs on it. The previous compressor failed in a similar manner after 100hrs. There were no oil alarms. vib analysis was carried out 2 days before the failure. nothing showed up on this. Can anyone hazard a guess on the root cause of this? lack of lubrication? i will try and get more pics up.

Stephen

P1010008.JPG (714 KB, 190 downloads)

My thoughts:

- Lack of lube would be the primary suspect. How is it lubricated?

- Is that an aluminum journal bearing? Though aluminum is often used as a journal bearing, it can have a tendency to gall if the lube is marginal. Lead infiltrated copper bearings are better.

- I'm curious why I don't see bluing on the parts, which I often see when journal bearings are wiped out. (Maybe it's because my experience is with automotive engines.)

- That looks like a cast ductile iron crankshaft. One little-known issue with ductile iron crankshafts is that if the crank journal surface isn't properly ground and polished in the proper directions, a phenomenon known as "ferrite caps" develops on the surface. Ferrite caps are basically tiny burrs that behave like a cheese grater and can wipe out a bearing in short order. The auto companies have paid me good money to study this phenomenon. I'm even half-way through sumbitting a patent along these lines (if my company can find the money to fund it further).

Thanks Kestas,
forgive me if i get the termonoligy mixed up, im new to these compressors!
It is lubricated through an oil pump attached to the crank. EMKERTE RL 32H food grade oil.
The compressor is being used to compress refrigerant in a chiller circuit (r507 refrigerant)
The journal in question is furthest away from the pump. notice the lub holes on the journal. picture 3 of 7 shows the remaining good journal.
You may also notice the bent conrod in picture 6 of 7. Could this have been the start of the issue. could some liquid refrigerant in the head caused this conrod to bend and start a chain reaction. apparently all this happened within a few minutes which is strange given the damage.

Now that I'm looking at it again, I notice that the block is a "W" design, with one journal driving three pistons. It appears that all three sectors of the crank pin journal are wiped out. The other pin is healthy - correct?

So there is something systematic about the journal that causes all three connecting rod bearings to wipe out. Again, this leads to oil starvation, or poor finish on the journal, or maybe somathing else, like fitment.

Note that the chamfered oil holes should not have any burrs after chamfering. A fingernail test should suffice to rule this out.

How the connecting rod broke at all is a mystery to me, unless there was enough rotational friction from the damaged bearing to bend and break the conn rod. Once broken, the bending (and other damage) could have come subsequently from the whipping action of the conn rod. Looking at witness marks on the conn rod and block can give a clue as to what contacted with what.

I've also seen contaminant from the oil holes not properly cleaned after manufacture to wipe out a bearing. A good 20X visual inspection of the bearing surface (what's left of them) may reveal this problem... or an inspection of the other good set of bearings may give evidence to suggest this.

With crankshaft designs, it is my experience that if anything is substandard with lubrication or grit contamination of the lube passages, this will result in a wipeout of the last bearing to see lubrication in the lube circuit.

Are you sure , the big end bearings are being assembled correctly so that it is in communication with oil coming out from the journal for lubrication of big end and gudgeon pin?Two common failure of similar nature and affecting all connecting rods tell the story of some common cause of oil starvation. Pl. check OEM manual for the correct bearing orientation. In case of big engines, I have found incorrect installation during assembly . Luckily, these were rectified before final installation.
Regards
Akhtar

Stephen,

I have done RCA on many failures of this magnitude or greater over the past 30 years. (just to give you my credentials). It is hard to say from one photograph and limited information as I am sure you are well aware. Your best clues lay in the parts analysis. If you see bearings which were hot (blue color) you can start investigating a lubrication problem. I can not say for sure because the photograph is not that revealing but it looks to me that a more likely cause may be a mechanical failure such as as a broken cap bolt from under-torque or loss of torque. Take a close look at the failed bolts and see identify their failure mechanism (shear, bending fatigue, torsion). Many times this can offer a clue as to the failure mode. Start at the end and work your way backwards. For example; the end event would be a connecting rod failure. Then investigate all things which could cause this failure. Eleminate the least likely causes and go to the most likely causes such as lack of lubrication (as you stated) or rod bolt failure. Then investigate the most likely cause of each of these events. Work your way back until you have a most likely cause(s). You may not be able to pin point one item and the root cause may be a combination of events but this is the RCA method that I use and it will help you recomend actions to prevent reaccurance. Hope this helps. Good luck with your investigation.

Thanks for all the comments guys.
The one thing that is leading me away from a lubriacation failure is that we had vibration analysis done on the compressor 2 days before the failure. surely any lubrication problem would have shown up then?
CGTCS: there are a number of photos in the other thread that may give you a better view.

Callins:

Do you have the process conditions of this compressor? Also the operating conditions(speed, %torque)? How are you doing vibration analysis on the compressor...since most recip compressors are better analyzed through a PV curve in addition to vibration. Just some additional information might help with determining the failure.

quote:

Originally posted by callins:
picture 3 of 7 shows the remaining good journal.
You may also notice the bent conrod in picture 6 of 7.

I only see one picture. Am I missing something?
Thanks

EPete,

See second thread by callins.
http://maintenanceforums.com/eve/forums/a/tpc/f/5951015103/m/5021056004

D

Aha. Thanks Dave.

How fast this compressor runs. Was it running hot? For the reciprocating compressors I would see oil VG68 and not 32 but is it specified by the manufacturer?
When compressor was overhauled the first time was it adequately cleaned? Does it have in line oil filter?

If it is a double acting compressor with lub oil pump mounted on crankshaft or getting drive from crankshaft, reversal of direction of rotation will have no effect on the performance of the compressor but oil will not flow to the bearings. Just a wild guess. Check DOR of the motor.
Regards
Akhtar

quote:

Originally posted by callins:
Thanks Kestas,
forgive me if i get the termonoligy mixed up, im new to these compressors!
It is lubricated through an oil pump attached to the crank. EMKERTE RL 32H food grade oil.
The compressor is being used to compress refrigerant in a chiller circuit (r507 refrigerant)
The journal in question is furthest away from the pump. notice the lub holes on the journal. picture 3 of 7 shows the remaining good journal.
You may also notice the bent conrod in picture 6 of 7. Could this have been the start of the issue. could some liquid refrigerant in the head caused this conrod to bend and start a chain reaction. apparently all this happened within a few minutes which is strange given the damage.

A hydraulic lock could have started this. Of the three pistons two appear to have suffered something more than impact damage. Do the piston crowns exhibit distorted (dished or fractured) in comparison to new ones?

I have seen this kind of failure before. I my opinion it was caused by liquid refrigerant "hydraulicking" the cylinder.

I would look at the expansion valve,it may be sticking, or the sensing bulb has a poor connection to the refrigerant piping. This would cause too much refrigerant to enter the evaporator, flooding over to the compressor suction line.

Is this a water chiller?

Rocky R.
I can't compare the old pistons to the new ones unfortunately. Yes my first opinion was hydraulic lock also.
Klaus K,
I assume by hydraulicking you mean hydraulic lock?
It's a chiller for a solvent recovery system.
The problem with refrigerant getting into the suction line is that there is a superheater between the evaporator and the suction line. The chiller was up to temperature at the time of the failure, so you would assume that any liquid refrigerant would get boiled off before reaching the compressor inlet?
Since this incident we have had several oil pressure alarms on the same compressor, and have discovered that the oil solenoid for this compressor was faulty. so maybe it could have bee an oil supply problem. we have replaced all oil components as a precautionary measure.

It would've been interesting to see an example of vibration signature which DID provide an advanced warning before failure occurred?

May be some one has one? Although, likely the impending failure would progress too fast to be captured.

Stephen, can you post your vibration data?

Thanks,
Dave

This message has been edited. Last edited by: David_G,

I'd love to Dave but unfortuantely the readings were taken by an outside contractor, and we have since parted ways!