This is an 1800rpm horizontal double-suction single stage radial flow centrifugal pump with sleeve bearngs. Driven by a 2500hp horizontal sleeve bearing motor thru a Thomas Shim Pack coupling.

While inspecting the associated recirc valve, a triangle-shaped chip of the impeller was found (slide 2). We inspected the pump impeller and found the hole to match the triangle shaped chip on the inboard suction (slides 1 and 2)

Valve had never before been inspected. Pump has not been inspected in 15 or 20 years.

The pump has never been monitored by prox probes. Review of housing vibration shows a very steady pattern throughout available history almost 20 years. The 1x is approx 0.02 ips on all pump positions except pump outboard horizontal which is about 0.05 ips. This is similar to or lower than the sister pumps. Vibration was last measured in October 2007.

Review of pump performance shows no noteable abrupt changes.

People who inspect the failure site (lack of jagged edges) have the opinion that the piece must have broken off a long time ago. But why wouldn't it have shown on vibration. I would be interested in comments on whether the following scenario's are likely or credible:
1 - The piece broke off at sometime after our last vibration measurement October 2007
OR
2 - The piece broke off before then and was not detected by housing vibration.

FME_small.ppt (547 Kb, 117 downloads)

Pete,

If your chunk weigh 3 ounces and is on about a 6" radius it's only about 100 lb. of force due to imbalance.

How much does the rotor weigh?

This is a prime example that the closer to the cemterline of the shaft a weight or the absence of a weight is, the less effect it has on the imbalance of the rotor as far as the general run of the mill equipment goes.

Some teach that the wrong key length will cause a balance problem, but not on the majority of equipment running in most plants.

I have tried and have never been able to make a full key, half key, empty keyway or anything liken unto that have any effect on the balance of a rotor, especially motors uncoupled. At a speed of 10,000 rpm and running near a critical, now it might have a little effect but on normal day to day stuff, I do not think so, unless the weight is exceptionally heavy.

OMOAICBTW

Ralph,
I don't know the condition of the equipment you monitor, but it is totally misleading to tell folks that the key can't make a difference. It all depends on the speed, the size of the key, etc. It's just so easy to cut the key the correct length in the first place not to do it.
Especially when some customers are demanding 1x vibration of less than 0.04 inch/second.
I had a customer test a brand new, in the door, motor on his shop floor and claimed it didn't meet their vibration specifications with a 1x vibration of 0.3 inch/second. This motor was not bolted down.
I asked if he had the coupling on. No.
I asked if he had a 1/2 key in the keyway. No.
I asked him to please try that. His response was that it couldn't make that much of a difference, but he agreed to try it.
This turned out to be one of those times when someone actually calls back to thank you.
He stated that he had learned something. The vibration at 1x had dropped from 0.3 to 0.03.
Now I am not saying that this motor might not have had some resonance going on with whatever he had it sitting on, but the fact still remains that the final trim weights for today's balancing efforts can involve 5 grams or so. The balance radius on a 200 hp motor is 5 inches and the keyway is sitting out at an inch, so let's just say for argument sake that the tolerance at the shaft would then be 25 grams. That's less than an ounce. Now weigh up a 1/2 by 1/2 key stock and see what that calculates out to for a key that is 1 inch too long..........
0.5 x 0.5 x 1 = 0.25 cubic inches of steel
.2836 lbs/cu in. = 0.07 lbs. = 31.75 grams, or over 1 ounce. Just threw away my specification.
You are absolutely correct that it makes much more of a difference at high speed and with critical speed conditions, but then again that's all about balancing, isn't it?
I just want to say that making the key the right length is so easy, just do it.

Speed is the killer since the load increases by the square of the speed.

1.77 x ounce x inches x (speed/1000)^2

Speed kills, play it safe!

Epete
I tend to agree with the thoughts of Danny and Ralph.
I too had a very similar experience with a site discharge trash pump (noisy to begin with) and did not see the impeller failure (ie 1X change) until we inspected for cavitation concerns. I suspect it was due to stiff structural mounts that simply didnt react to the unbalance force.

Compare that to shroud break on a 10,000 RPM turbine (no real missing mass - just displaced) and the unit shook violently as we brought up to speed. Just needed to get to ~ 2000 RPM and it became obvious we had an unbalance problem in the turbine and shut down immediately. The main difference was in this case the structure and bearing supports were fairly compliant and the mass displacement much farther away from centerline.

Jim P

Ralph I have come across several instances where a long key resulted in high unbalance vibration. The good news is that it is easy to fix...

Pete I think you had a pretty heavy case, and a pretty small increase in unbalance, and as a result your casing transducers did not pick it up.

I have on several occasions come across on boiler feed pumps high vibration on the prox probes that was not picked up with casing vibration.

quote:

Ralph,
I don't know the condition of the equipment you monitor, but it is totally misleading to tell folks that the key can't make a difference. It all depends on the speed, the size of the key, etc. It's just so easy to cut the key the correct length in the first place not to do it.
Especially when some customers are demanding 1x vibration of less than 0.04 inch/second.
I had a customer test a brand new, in the door, motor on his shop floor and claimed it didn't meet their vibration specifications with a 1x vibration of 0.3 inch/second. This motor was not bolted down.

quote:

Ralph I have come across several instances where a long key resulted in high unbalance vibration. The good news is that it is easy to fix...

Ron and Steve and All,

I said in my opinion

quote:

OMOAICBTW means Only my opinion and I could be totally wrong

I have ran several test to see what happens on 3600 rpm motors with keys up to 5/8 x 5/8 inch and I have never seen any amplitude increase or decrease from the key.
I also said at higher rpms it might be a problem. It is just that I have never in all these years seen an actual case where the key solved or caused a balance problem. I am glad you guys shared your case histories with us. Thanks.

I never ever want to mislead anyone. It is hard enough doing this type work without someone giving misleading statements and not ending it with OMOAICBTW, because believe me, no one knows everything about anything, vibration included, if they think they do, the only one they are fooling are themselves. So gentlemen, always be leary of what someone says until you have tried it yourself. Sometimes good ideas come from strange places. Never disregard an idea as a total dud until it has been proven a dud.

Again, I apologize if anyone feels mislead. Remember what Danny siad "Speed kills".

"I feel the day is a total loss if I do not learn at least one new thing about vibration." quote from Ralph Stewart, 1988

If the key does not make a difference, it kinda shoots 4W/N in the head, huh?

quote:

If the key does not make a difference, it kinda shoots 4W/N in the head, huh?

Well Said. Guess we all have learned something today. What, I am not sure, but at least something.

This message has been edited. Last edited by: Ralph Stewart,

Pete,

Not only bearing mounts are relatively stiff but, considering sleeves, stiffness of the bearing itself is relatively low - a typical situation of a system with two degrees of freedom when two different stiffness springs connected in series. With the measurement taken on the stiffer one, sensitivity to imbalance will be low.

David

Thanks for the comments and insight.

I have another story about very high vibration during uncoupled motor run without half key. But I don't bring this up because I want to beat a dead horse... I bring it up for purposes of comparison with the chunkless-impeller.

This motor run without half key occurred on an 1800 rpm 2500 hp sleeve bearing motor driving an identical (sister unit) pump as the one discussed in my original post.

Slide 1 (see atttached) is the trend of overall vibration for this motor. I think you can pick out the point where we ran the motor uncoupled without a half-key: 0.6 ips housing vibration resulted (which is pretty scary for a big 2500hp sleeve bearing motor... and yes I know that even our baseline level 0.2 ips is high).

Slide 2 (see attached) is a photo of the rotor to give you a rough idea of the size of the keyway.

The keyway dimensions are approx 7.5" x 1.25" x 0.625".
This gives a volume of 5.859375 cubic inches.
Assuming density 0.3 lbm/inch^3, it is about 1.7578125 pounds.
The radius is about 3.5", resulting in about 6.15234375 inch-pounds.
Or approx 98.4375 inch-ounce.

Compare to our triangle-shaped chunk which came out of the pump impeller:
The approx dimensions are a triangle base 1", height 1" depth 0.375"
Volume is 0.5 * base * height * depth = 0.1875 cubic inches.
Assuming density 0.3 lbm/inch^3, it is about 0.05625 pounds.
The radius is about 7", resulting in about 0.39375 inch-pounds.
Or approx 6.3 inch-ounce.

Since the pump and motor have the same speed and similar bearings, we might make a giant leap and ASSUME that they have similar influence coefficients (although they have different support, different rotor, different geomtery for overhung key unbalance compared to the between-bearings chunk unbalance, and possibly the motor may have been beyond its linear range).

But assuming the influence coefficients are similar, the vibration is proportional to the imbalance (inch-ounce). If we take our 0.6 ips from the motor and estimate pump unbalance vibration based on ratio of inch-ounce, we get:
0.6 ips * (6.3inchounce/98inchounce) ~ 0.04 ips.

Considering uncertainty of the assumption of equal influence coefficients, I guess it does seem reasonable that the "calculated" effects of the pump chunk unbalance (0.04 ips +/-???) may have been contained within or masked within our measured pump housing vibration levels 0.02 - 0.05 ips as was suggested by others above.

I will post an overview photo of the outside of the assembled machine if I can find one. Maybe then based on that photo and the impeller photo above you guys have balanced similar pumps and can guesstimate the reasonable range of influence coefficents.

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

NoHalfKeySmall.ppt (200 Kb, 38 downloads)

Attached are some photo's and drawing of the pump.

Does anyone have a guess as to what might be the expected range of influence coefficients for unbalance applied to the impeller this pump?

(again it is 1800 rpm, single stage double suction centrifugal pump on sleeve bearings).

FWBP_PumpOverview_small.ppt (558 Kb, 33 downloads)

There was a time when we were getting new air handlers in the plant that just did not meet our spec for vibration let alone ISO.

We started looking very closely at key lengths and had to start double checking them from the factory. That was a major pain, but it paid off in the end. We were very surprised at how many came in that were wrong.

Balanced_Key_Way.pdf (110 Kb, 36 downloads) key calc

Interesting. I have always heard a thumbule that the length of the exposed key (d2 in my attachment)should equal the length of the unused keyway (d1 in my attachment). That way the volume of the extra chunk of key sticking out roughly matches the volume of the unused length of keyway (the hole) (A1 = A2 in my attachment). That is just approximate, but a good start and very easy to "eyeball" whether those two distances are roughly the same.

I'm sure your method is more exact.

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

Keylength.ppt (22 Kb, 23 downloads)

ISO 8821 deals with key fits and balancing. It is worth reading.

I assume all (with an exception) agree that one must include the key in the balance.

As to the vibration level expected from missing mass (or influence coefficients), perform a forced response analysis to determine what to expect.

I'm attaching the picture of similar pump impeller removed due to performance degradation. The vibration levels did not show any major vibration level increase. The impeller is not to heavy and running speed of 1800 rpm makes it a very robust pump.

Impeller.pdf (780 Kb, 31 downloads)

Alex - Wow. That didn't show on vibration? Amazing. Do you remember if it was sleeve bearing machine with vib monitoring on housing also?

It brings to mind another thought. I have heard the term "hydraulic unbalance" before. My understanding is that an asymmetry in the impeller (such as chunk out of impeller at inlet or discharge) can cause an asymmetry in the pressure and force distributions, rotating at 1x... resulting in 1x vibration. Is that true? Any way to guess the range of vibration magnitudes that might be expected? Would it be worse for a chunk on suction or discharge side or some location in particular?

Pete,
This picture was taken several years ago and as far as I remember these pumps were experiencing low flow recirculation (two pumps running and most time only one was required). These pumps have oil lubricated ball bearings. Due to hydraulic recirculation it alway exhibit some non harmonic components in spectrum but 1xrpm was not high. I always seen 1800 rpm pumps to be very forgiving. With so much impeller damage high speed pump would be walking on its own.