In Minor Inspection, Our Feed Water Pump (FWP#13)'s all 6 Coupling Bolts was found rupture from shear force and insulation plate was crack. Please see the picture in the attached file.
Do any plant have a same problem? and any suggestion for root cause analysis?


Photos (1)
Original Post
You are sure all these photos are the 6 bolts pictures on the very end of the hub (not the 8 bolts that connect the hub to the intermediate plate), correct ?

We see the failure occurs roughly halfway along the length of the bolt. The bolt itself has no inherent stress concentration there and no obvious reason that stress would be highest there. Maybe that tells us something about how this bolt is used .. we don’t quite know.

The photo posted 16 May 2012 11:11 PM shows plastic deformation. My eyeball says the deformation begins or occurs halfway through the length of the bolt, where there is also a noticeable change in finish (the change in finish may provide some clue as to what this bolt actually mates up to inside the hub... we don’t quite know). Also this location halfway along where the bend begins and the finish changes matches the location that the failure occurred in the other photos. It suggests the bend is telling us about the same stresses that ultimately led to the failure.

The fact that it bends does not seem consistent with brittle fracture.

I would say also the bend suggests there is some stress other than tensile (overtightening). Yet if all that the bolts did was hold on that little tiny external insulation plate, we wouldn’t expect any other significant stress other than tensile stress from preload.

I assume the nuts in the very first photo remain in the position they were found (it was not necessary to loosen the nut since the bolt was broken). I notice the nut is in a different position on the bent bolt photo (slightly further our). I presume this nut was loosened during but removal but not restored to origianl position but it’s original position would have been similar to those of the first photo. Also the threads are discolored just to the left of the nut of the bent bolt.. probably the discolored area is where the nut originally sat.

If the nuts were butted up against the inside of the hub (so only thing between bolt head and nut was the thin external insulator and the end of the hub), then I can see no explanation for the bend and change in finish. Is there something else captured by these bolts inside the hub? Got any photo’s looking inside the hub... or additional info about the coupling and how these particular bolts are used?

I suspect if you piece together how the bolt is actually used, a clearer picture will emerge of why there is a large bending stress at that location half way along the bolt.
If the pump rotor did not seize, then you probably have a torsional vibration problem. I suggest measuring torsional vibrations after pump and coupling are restored to original condition. Does the crew that installed coupling have a torque wrench, have the correct bolt torque values, and were checked on the procedure? One can only learn so much from a bucket of broken bolts!!

I think I figured what I was missing. I was thinking (like Danny?) that the two flat smaller faces with 6 boltholes faced outwards from the center. That was silly. The two flat faces with the 6 holes face each other and are bolted together by those 6 bolts. The purpose of two pieces within this spacer vs one may simply be to insulate the halves. Then of course the larger flange end faces outward to a flex plane at each end.

So I have a new proposal.

The bolts were loose initially or loosened at some point. Otherwise I cannot think of any reason for the bend shown.

How could it loosen:

*** Insulation has a lot shorter lifetime than steel.. over time it ages and loses its mechanical properties. One possibility to consider is that the bolts were initially tight, but then the insulation degraded under compression which lessened the tension. Especially if that insulation is 10 or 20 years old.

It could be other factors. Lack of initial torquing or lack of loctite.
The purpose of two pieces within this spacer vs one may simply be to insulate the halves.

The only "spacer" I have seen in a feed water pump coupling was for control of end float (axial movement of motor shaft) and not thermal insulation;certainly not to transfer torque. I do agree that the bolts may have been loose and bent before breaking. The $64.00 question would be how did they get loose?

The only "spacer" I have seen in a feed water pump coupling was for control of end float (axial movement of motor shaft) and not thermal insulation;certainly not to transfer torque.

Let's clarify the terminology. I am talking about the spool-shaped thing between the two flex planes that creates the spacing between them. Mancuso (author of "Couplings and Joints") also calls it a spacer. Others may reserve the term for only very long spacers. I am aware other usages of the term spacer as a button spacer that is takes up only a small portion of the space between flex planes and is not attached at both ends, but instead bumps against one or both ends to limit end-movement... the usage that you mentioned. I thought from the context it was clear which usage of the term "spacer" applied. But feel free to substitute the term "spool" if you don't like the term "spacer".

I was talking about electrical insulation, not thermal insulation. I should have been more specific. It applies throughout all my comments in this thread - I'm talking about electrical insulation.

This spool/spacer does transfer torque.
Looking at pieces in original picture: we can detect three elements of the insulation between the two halves of the spool/spacer.
1 - First is the flat insulating washer laying underneat the bolt head and steel washer sitting on left.
2 - Second is the large flat insulating piece between halves, stuck on the left half, damaged.
3 – Third is insulation on inside of the bolthole as shown on the right half.
In order for the insulating system to work, the bolt has to be fully insulated from one half of the spool/spacer or the other. Since the half on the right has the insulation inside the boltholes, the washer would have to be on the right in order to provide insulation. From the positioning of bolts after removal, all bolts are positioned with bolt-head and insulating washer to the left. I don’t know if that’s the way they came out, but normallly you’d try to preserve the orientation in a photo. If this is the case (washer was installed on wrong side), it would have defeated the electrical insulating funciton. I'm pretty sure that electrical current would not have anything to do with the failure, but just wanted to mention it so it can be corrected during future maintenance if applicable.

It does bring to mind related item which may be more relevant to the failure. Perhaps there should be steel washers on both halves. That nut may not be much larger than the bolt hole. Small overlap between nut and steel could create weaken the bolted joint (actually EPRI mentions that without steel washer, this type is susceptible to developing indentions which may rob the preload after initial torquing.. so they recommend hardened steel washers for critical joints). This could be a contributor to loosening also.

Also going back to earlier discussion of possible causes of looseness attributable to insulating materials..... I said it could be degradation of insulating materials where they lose their mechanical properties. That applies not only to the large insulating circular separator, but also more importantly to those pesky insulating washers (if they fail mechanically, the connection loosens). It leads to a very tricky thing: insulating washers may have a relatively poor mechanical strength properties. So while you'd like to torque the bolts based on bolt properties to give a heavy preload in order to prevent loosening, heavy torquing might crack those insulating washers which could also be the beginning of failure (if failure of the insulating washer allows it to change it's geometry which lowers preload on the joint). Insulating hardware in mechanical connections can be problematic for this reason. You can inspect those washers to see if they are failed in as-found condition. If they are failed, inspect closely but failure alone doesn't prove anything (could have happened after loosening for other reasons). If they didn't fail, it may still be the case that low bolt torque was used to protect those insulating washers and that low torque was contributor to the failure.
Attached Slide 1 is a figure of how I think this coupling goes together. Slide 2 is labeling of the photograph.

Slide 1 shows the spacer/spool in the middle (two halves, with electrical insulation between them). The electrical insulation is shown in green. The failed bolts are the ones that hold the two halves of the spacer/spool together.

Also if you note if you flip those 6 bolts so the head and insulating washer goes on left side in my figure, it would defeat the insulation (because only the boltholes on right are insulated).

My drawing of the "coupling bolts" (8) could be improved. Each coupling bolt clamps between the flex element and EITHER the hub or the spacer, alternating. No bolt clamps all the way from hub to spacer. Since these are NOT the bolts of interest (my assumption), I'm not going to spend too much time making that look perfect.

Otherwise, is this roughly correct?


The non-metallic insert was obviously originally intended as electrical insulation. We can tell that because it has all three elements required for electrical insulation (insulating washers, cylindrical insulating insert for boltholes on one half only, and round insulating separator). There is no other logical reason to bring these three elements together. For example if you wanted to change torsional characteristics, there's no reason you would need the insulating washer (it doesn't come into play for torsional characteristics, but certainly does for electrical insulating characteristics).

I am in full agreement that the insulating separator along with the insulating washers are likely weak points in this mechanical system (which is otherwise made of steel components). And it's possible (if not likely) that they're causing a heckuva lot more problems with bolt loosening than they will ever solve by preventing pump bearing currents. Maybe we can comment further on the need or lack-of-need for insulating coupling if one of the guys familiar with this particular installation can tell us if it's motor driven or turbine driven, vfd or non-vfd motor. If non-vfd motor, is outboard motor bearing insulated?
We can tell that because it has all three elements required for electrical insulation (insulating washers, cylindrical insulating insert for boltholes on one half only, and round insulating separator). There is no other logical reason to bring these three elements together. For example if you wanted to change torsional characteristics, there's no reason you would need the insulating washer (it doesn't come into play for torsional characteristics, but certainly does for electrical insulating characteristics).

The geometry configuration can indeed be used for vibration isolation or torsional stiffness control. I don't know how you can tell what the material is made of from the photos. You have better eyes than me! I smell torsional resonance on this machine, but how can I smell from the photos!

I don't know how you can tell what the material is made of from the photos. You have better eyes than me!

Hi Walt. My conclusion was not based on determining the exact material by sight. My conclusion was based on the shape/form of the material which proves it’s function. Again we can see in the photo three insulating elements;
1 – Round shaped insulating separator
2 – Insulating Washer under the bolt heads (on 2 of the 5 bolts in the photo, I would presume the other insulating washsers fell off either during operation or between disassembly and photo, and the 6th bolt hiding somewhere else).
3 – Insulated bolt-holes covering at least one of the two halves of the spacer/spool. (insulating sleeve can be longer such as extending past the joint interface, but it is not required)

Now look how these parts are arranged in slide 1 of my powerpoint. You can see they provide effective electrical isolation. Now remove any one of the three and electrical isolation is lost:
1 – Remove round insulating separator and the spacer halves contact each other... insulation is lost.
2 – Removing insulating washer under the bolt head allows the right half of the spacer to electrically contact with the bolt. Insulation is lost because that same bolt was already in contact with the left spacer.
3 – Removing the cylindrical insulating sleeve from a bolt-hole on the right side allows the right half of the spacer to come in electrical contact with the the bolt. Insulation is lost because the that same bolt was already in contact with the left half of the spacer. (Actually if you have perfect centering of the bolt within the hole to assure no contact between bolt and bolthole, insulating sleeve would not be required, but assuring such clearance would require an oversized bolthole which increases stress on the insulating washer which is already mechanically challenged. )

You can see that these three elements are required to accomplish electrical isolation of a bolted joint. Exactly these three are what is required for electrical insulation. Any more would be superfluous. Any less would not accomplish the task.

Now let us try to imagine that these are used for vibration isolation:
1 - Why do I need a flexible washer under the bolthead? If we wanted to alter axial vibration we can do so using the separating piece. More importantly, regardless of the intent of the flexible washer, it inhibits us from clamping the bolts tightly, which is required to prevent the halves of the spacer from slipping relatively to each other. It is not a design choice we would take if we could accomplish the same thing other ways, like using the separting piece. It is a design choice that is forced upon us for electrical insulation because the bolt must be completely insulated from one half.
2 – Why should we have flexible materials in boltholes of only one half of the spacer? It is asymmetrical.. no logical reason for vibration isolation (very logical for electrical isoltion).
3 – Is this mysterious vibration isolating device for axial or torsional purposes?.... the bolthole insulation can serve no axial thrust isolation purposes. The other two serve not much purpose for torsional isolation.

Hopefully the logic is clear. Now let’s go to experience:

Every electrically-insulated bolted connection I have ever seen has exactly these three elements (no more, no less). Off the top of my head, that includes upper bearing support ring bolting to upper flange on our circulating water pump motors, upper bearing suport ring bolting to upper flange on our condensate pump motors, lower bracket to stator flange on our essential cooling water motors, oil cooler pipe insulated flange for coolant pump motor, and generator stepup transformer mounting flange for low voltage bushing enclosure (doghouse). Construction details for the Circulating Water Pump motor are shown in the attachments to this thread:
(which coincidentally describes failure of insulating materials used in a mechanical joint).

In contrast, I have never seen a mechanical isolation device built anything like this (including all three elements). The few isolation couplings I have seen were all forum links, but none of them looked anything like this. One torsional coupling I remember seeing had removable elastomeric elements enclosed between jaws. Another I vaguely recall had flexible sleeve around pin similar to bolthole insulation, but I’m sure those pins were not used for axial clamping as these bolts are. I have seen nothing resembling all three elements discussed above. As always I could be wrong (we don't know what we don't know), but now you know the basis of my opinion and it remains strong.

The geometry configuration can indeed be used for vibration isolation or torsional stiffness control.

If you can post an example of vibration isolating or dampening coupling that includes all three elements discussed above, I will find a way to send you a beer ;-)
The USBR does not rotate, so its function is not the same as a shaft coupling.

A quick search for bolt vibration isolators:

If you have a wet-mechanical cooling tower at your plant, then check out the drive shaft couplings. They typically look like the photo at the start of this posting. The coupling is fine for low speed, but not so much for high speed feed water service.

Originally posted by Walt Strong:
The USBR does not rotate, so its function is not the same as a shaft coupling.

The examples I posted were to support my statement: "Every electrically-insulated bolted connection I have ever seen has exactly these three elements"., which in turn supported the logical discussion that these three are necessary/sufficient for electrical insulation. The conclusion about how many types of elements are required to electrically insulate a bolted connection does not change regardless of whether that connection is stationary or rotating (still three). The function of the insulated joint shown in my powerpoint is simply a rotating insulated joint (the flex planes are elsewhere).

At our plant, we don't have any cooling tower fans (we have a cooling reservoir instead). As you say, that application is low speed but still interesting. Do these have all three types of non-metallic elements which are clearly shown in the photo, and in similar geometry? If so, maybe I'd owe you a beer on my challenge (I'll await proof). However, even if that is he case, there's still a very clear indicators about the electrical function of these pieces, namely the asymmetry:

  • We have insulating sleeve shown on one half the boltholes. Insulating sleeves on the other half would serve no purpose because once you interrupt an insulating path, any further impedance in series has no effect. In contrast, if the sleeve were intended to carry mechanical forces, the forces from both half would also be relevant and there is no reason to provide this function on only one half of the coupling.
  • We can apply similar logic to the insulating washers. They are shown on only the head end of the bolts. Interrupting the electrical path once is all that is required and another one accomplishes nothing further. There is no logical mechanical reason to include on one end and not the other (if bolt is too short, get a longer one). You might argue: "but they're only present on 2 out of 5 shown." The counter-argument: 5 out of 5 show that large steel washer also on the bolthead end. And on the 2 of 5 that have an insulating washer, the steel washer is located between the insulating washer and the bolthead, which is exactly where you need it to distribute the force over the insulating washer in order to decrease stresses in the insulating washer. (By the way steel washer in similar position occurs in all the insulated bolted connections I mentioned previously, including the linked one). So the steel washer which occurs on only one end of 5/5 has a very logical link to the insulating washer. In contrast, if that steel washer was used for mechanical purposes unrelated to the insulating washer, there is no reason in this geometry (bolt with through-hole/nut) to put it on one end and not the other. 5/5 steel washers on one end - perfect sense for electrical insulation. No sense to me otherwise.
Hi every poster, first I have to say SO SORRY to late response and thank you to every valuable suggestion that work under less information form us. The original poster is my boss and Adisorn is one of my team.
I'd to tell you more information and what we take action for this and what it's going on.
We replaced new 6 bolts and three insulated elements with new material(Bolt material change form 10.9 to 12.9 and insulated element change form fiberglass to Bakelite). It's just temporary action to continuous operation plant.
This coupling transmitted power from motor(non vdf motor) to fluid coupling set. I not sure what it's actually purpose, but it's original designed form OEM, that coupling's company told us, it's the requirement form they customer to put insulated elements. I've never seen coupling designed like this. In other plant have similar motor and fluid coupling, but they use gear coupling to transmitted power and no have any electrical effect problem after 25 operation. my new questions , can we change coupling type for example gear type. If it's use for insulated how can we check current on shaft.
From my standpoint, Fiberglass and Bakelite are hard materials which are electrically insulating and it is safe to assume from everything we now know that the intended use was as an electrical insulating coupling. Others are welcome to chime in if they have other ideas.

If you have no vfd, I don't see any need for insulating coupling. So I agree with your general approach to get rid of that insulation.

Replacing the coupling with a completely different style as you mentioned is one option to explore. (gear coupling brings lubrication requirement with it).

In the interest of a less extensive modification, presumably with less possibility to introduce new unintended consequences, I would suggest to explore re-using the existing coupling with these modifications:
* remove the insulating separator and insulating washers
* put hardened steel washers under both the head and the nut, not just under the bolt head.
* torque your high-strength bolts to 80% yield and use Loctite threadlocker (**) on the threads. Make sure you have good thread conditions and seating surfaces. (I think these measures along with getting rid of the insulation will will get rid of the main cause which I believe is self-loosening of your bolts.)
** There may be other vendors but we typically use Loctite brand. From memory Loctite 242 is good for most and can easily be removed. Loctite 271 (high strength) is even better locking, but requires heat to remove. You might want to review this further or talk it over with Loctite to pick the best one.
* don't forget to consider the effect of change in axial alignment resulting from removal of the insert (re-check sleeve bearing motor centering, at least a reasonable distance from it's limits of endplay).

It would be nice if you could touch base with OEM's, at least the coupling manufacturer to see if he they have comments/concerns about this use of their coupling, but I personally don't see any problems with it.

It would be nice to verify the outboard motor bearing is electrically insulated if possible (sometimes you can tell by external inspection).

It would obviously be nice to know how this style coupling originally came to be installed (did someone specify it and why). It sounds like you have investigated that already and not much info available.
For the sake of completeness, it's worth mentioning something you probably already know.

Any time you change or modify the coupling, the torsional natural frequency(s) will change. That applies whether you put in a new coupling or remove the separator from existing coupling. If it should change to become very close to a torsional exciting frequency, that creates torsional resonance.

Luckily, you don't have many torsional exciting frequencies. The only ones of concern are twice line frequency (for example 2*50hz = 100hz) which comes from motor in presence of slightly unbalanced voltage, and maybe pump blade pass frequency. So, certainly the risk of this type modification is much lower on this machine than it would be on recip machine or machine with gearbox or vfd. And considering that you are getting rid of a known problem, creating a small risk of a potential new problem may be a risk you judge to be acceptable (your decision to make, not ours).

I am not sure if there is much practical that can be done to evaluate the new torsional resonant frequency, other than a big calculational study. I'm not familiar with torsional testing... I'm under the impression it's not easy. Maybe Walt or someone has comments on that.
As we agreed, the bolts have clearly loosened. The reason we come to this conclusion: When properly clamped (bolts tight), the torque should be carried by friction between the halves of this bolted connection. In that case, there is only tensile stress on the bolts. There is no shear stress or bending stress on the bolts until and unless they loosen.

I think we are in agreement that far, correct?

The next question as you asked before: why did the bolts loosen?

My answer: We have a very likely culprit in the presence of these insulating materials. We know that:
1 – If the insulating materials contract/compress over time, the connection loosens.
2 – We are limited in how much torque we can apply to the bolts by those insulating washers (they would surely crack if you applied standard bolt torques). Low bolt torque leads to bolt loosening.
We know for sure that we have an insulating coupling and that insulating couplings of this design are susceptible to bolt loosening (for reasons above).

There are other factors we might hypothesize that can be influences towards bolt loosening.
Torsional resonance is one that could clearly contribute.
Do we have any reason at all to suspect the presence of torsional resonance in this scenario? I didn’t hear anything in the post which would lead us to suspect torsional resonance (other than the bolt loosening itself, which I would argue has more direct logical cause in the insulating coupling design which we KNOW exists, not something we speculated to exist).

Is torsional resonance a common condition in non-vfd motor, fixed speed centrifugal pumps? I wouldn’t think so, since the sources of torsional excitation in this type machine are relatively few and relatively weak (compared to recip machine). I don’t recall a single case like that posted on the forum or posted on the INPO operating experience network. I’m sure it is remotely possible, but surely not so common that we would predict its existence without any other evidence.

I certainly respect your opinion Walt. You have far more knoweldge and experience of vibration than me. Maybe there is something I’m missing. I would be interested to hear the thoughts behind your conclusion that this case is caused by torsional resonance.

The good news (for the people who posted the info) – The proposed fix to eliminate the insulation will most likely provide satisfactory resolution, regardless of who is right about whether torsional resonance played a role in these initial failures. Removing an insulating insert from an otherwise steel coupling will surely alter the torsional natural resonant frequency of the machine substantially. If we were unlucky enough to have a torsional natural frequency sitting right on top of twice line frequency or blade pass frequency, then this same torsional resonance condition will surely not exist after the fix.
The options for designing an insulating coupling are limited, and it's not suprising that to design an electrically insulating coupling will require some compromises on the mechanical design.

Yes, the OEM is usually aware of the weaknesses and potential failure modes of his design. For this design, he is surely aware that:
1 – insulating materials age. This is usually accompanied by changes in volume and always accompanied by changes in strength.
2 – insulating materials have lower strength than steel so the bolt torque is limited and bolt will be more susceptible to loosening (and by the way they will also tend to loosen if the insulation shrinks during aging).

How does the OEM cope with these realities? He probably gave some recommendations:
1 – Replace the insulating parts periodically. Maybe every 2 years or 5 years or 10 years.
2 – He probably specifies an exact torque value for those bolts. It is as high as possible without cracking the washers. But still not as high as can be accomplished with all-steel coupling and therefore more susceptible to loosening.
3 – To address bolt loosening concerns (due to lower bolt torque and potential insulation shrinkage), OEM may recommend to retorque the fasteners periodically. Or at any rate they get retorqued when you replace the insulating parts per recommendation #1.

What happens to all those OEM recommendations? They go in a document somewhere and get forgotten. And then this type of failure can occur.

You may ask: but what about an all-steel shim pack coupling... doesn’t that OEM have recommendations and might those be forgotten also?

The answer is the all-steel coupling has fewer maintenance requirements and they are more standard/familiar. It is therefore more forgiving of people who don’t read/follow the entire OEM recommendations. Torque it during assembly using standard thrumbrules based on bolt material/size and you are most likely not going to have problems (assuming you also follow standard alignment practices). The same cannot be said for this insulated coupling.
I'm not clear if your asking a question for me or for the original poster(s).

If it's a question for me, my response is that it is certainly not unusual for an OEM to provide PM recommendations and for these recommendations to vary among OEM's (although user's may not be tuned in to these differences).

I'm not sure what is meant by "minimal maintenance readiness before first production". Perhaps it means the equipment should be ready to go when purchased. I don't know how long this equipment has been installed before these photos were taken, but that is an interesting question for the original posters.
I'm so sorry too late for All.
- lap test - result failure mode is Brittle Overload
- OEM - this is the first design for insulation bolt ,it's recommended from Pump OEM. They don't have experience for this case.
- Maintenance Program - We create new program follow All advice, Inspection and re-tighten Bolt every CI(1 year). Use CBM program for visual inspect coupling slip every 1 month, Replace the insulating parts every MI (3 year)
thank you and best regards

Add Reply

Likes (0)