Flakt ID Fans Model HABS125-220 Fatigue Problem

I'm not finding the number here.

Would you care to point it out?

Here are a couple possibilities.

What is passing thru the fan and what are the impellers made of? Hydrogen embrittlement is common in many alloys especially ones that may be subject to fatigue. Things like acidic atmosphere can contribute.

Also if the fan is not properly balanced, the stress (possibly even impact loading) can be passed on to the impellers. Just doing a single axis balance job is not sufficient in many fan applications. The balance must be at least what is called dynamic. You might read up on this. It involves resolving thrust as well as singe axis balance. The impeller may also need to be taken out and balanced separately, then the rotating assembly, then the fan checked when running.

In addition, if you are able to determine that the original design would be better except for the low fatigue strengh behaviour, you could do some back of the envelope calculations and verify if the original impeller associated with some technique like shot peening could do the job. Stress relieving welds mechanically or by thermal process also could make difference. But I agree with you: replacing the impeller by a heavier one without anti-thrust features may not be the right way. But you should also check how much would it cost to replace the impeller and its performance over time (not to mention the amount of energy such replacement could draw). Maybe keep replacing bearings is already the cheaper solution (I am talking of course without any idea of what such stopped fan costs), maybe a new impeller design could save the company plant.

Fatigue usually results form high vibration. Pl submit previous actions taken. What are the present vibration levels? Balancing, alignment, natural frequency determination (Bump test) may help if the cause is high vibration. Increasing stiffness in axial direction if the fan is running near critical speed , will be useful. Axial thrust balancing- how it has been achieved? The fan is running near its design capacity or away from it?

Irshad

are the impellers made of Zinc? Which can fail over time.

Zinc Creep

Henry,

You will need to tell us the fan details to even has it a guess at the problem and solution. From the fan designation, the only people who probably know its type is Flakt and yourself. Is it an axial flow - variable pitch or fixed blade impeller, mixed flow, or centrifugal. What is its duty - mine ventilation, boiler, etc, what system configuration is it in & method of flow control. Where is it operating on its performance curve.

I have seen fatigue blade failures on axial flow fans caused by stresses induced from unusual operating conditions. In one case, 3 fans operating in parallel (each 300kw electric motor drive) on boiler induced draft duties, where one fan unknowingly was operating in a stall condition for an extended period of time. The ultimate failure modes are mechanical but given that they were designed to appropriate design standards extraordinary aerodynamic and system conditions are usually the initiating cause leading to failure mode.

If you can provide some fan/system details you will probably get some constructive comment here but my first call would be to Flakt to get some of the 20 year retrofit history of the fan.

There have been times of the last 40 years that I have ran up on these types of problems that everyone just lives with and passes on to the newbie. Seems that after a few months or so, another newbies comes along. In digging into these issue I have found some of what has been suggest by others as well as the fact that the design was wrong from the start. There is where the biggest problem is. Right materials wrong job.

Do you have a comprehensive record of the vibration analyst history on these units. Just guessing I would say no. Start there and run some testing for a few months and see what is happening before you get to far into it. Or, wait until and newbie shows up!

I've worked on these before. I have a few questions. 1st, is this handling air for a typical office building application, or is the air carrying any type of industrial vapors or particles? 2nd, these monsters are really maintenance heavy. Have annual or hour meter times been closely followed? 3rd, are you confident the people doing maintenance are qualified to work on this complex design? Lastly, have you done a cost projection to convert to VFD?

If your original diagnosis of fatigue is correct, I would look at fatigue resistant materials, static and dynamic balancing, and other modifications to minimize the vibration modes that are causing the fatigue. Instrumentation such as accelerometers and high speed movies can be used to detect and quantify the vibration modes.

do you mean this one ?

http://www.flaktwoods.com/242efcb1-5b25-44ca-8b44-ef1ac00fd769

How much simulation - finite element analysis, vibration analysis has been done?

do all impellers fail or only a percentage?

perhaps manufacturing tolerances - dimensional and balance.

Are they dynamically balanced? at full operational rpm and load?

Does some aspect of airflow result in uneven loading of the impeller?

Are there any resonances in the design at the operating rpm?

Perhaps an analysis by an independent expert in the field would be helpful.

Thanks for all your input.

Further information these fans are centrifugal fans in a furnace waste gas venturi scrubbing system and these gasses are slightly acidic. Impeller material is Sandvic SAF 2205 stainless steel . the latest drawings of the impeller have no Flakt logo on them so they are probably not "approved" impellers. Flow rate 38.7 m³ /s, ΔP 11.4 kpa, operating speed 1480 – 1500 rpm driven by direct drive 1.3 Mw ac motor. Inlet gas temp 67^oc and inlet flow dampers are generally about 40% most of the time.

New fans with VSD's have been costed but have failed to make it to the capital budget for the last 5 years and not likely to get on in the current economic climate.

Calculations on the shaft with the current impeller show that is running a 0.83 of a critical speed.

Vibration velocity measurements are monitored in the control system with alarm levels set at 10 mm/s and trip levels at 13 mm/s, normal running levels are 2 -4 mm/s. Acceleration spectrums are collected at least every month, though as the bearings start to wear this can increase to daily. Bearing temps range from 85^oc to 105^oc and alarm levels have been reset over the years from 70 ^oc in 1990 to the current alarm level of 110 ^oc in 2008.

There have been comments made that welding procedures may play an important role in the impeller failures but there is no documented evidence.

The inlet flow dampers were installed about 1990, after the impellers were modified but again documentation as to why this was done has not survived.

I will have an impeller that we will be changing out at the next shut in about a month's time so I can have a good look at it, and will post photos if there is interest, I would like to hear the Flakt side of the story but again comments suggest that relations between us may not be on the best of terms, though as the last communication between us was 10 years ago it may be worth getting in touch with them again.

I guess that having learned my reliability craft at the professional end of the spectrum, in the aerospace industry where "beefing things up" was not an option, that I am biased toward redesigning the impeller to eliminate or reduce the axial load on the bearings and cope with flow conditions inside the scrubber system.

Thanks once again for the suggestions and please accept my apologies for the tardiness at providing feedback. I will let you all know what solution is implemented but please no "breath holding".

What is the bearing configuration?

Deep groove, angular contact, other?

Reason for asking is you can't really reduce the axial load due to 'suction' area, unless you reduce the impellers performance. I.e. the 38.7 & 11.4 kPa.

The 11.4 of course acts on the disc area of a 'single sided' centrifugal fan - I.e it would be helpful to see the impeller and scroll design.

Neutral thrust/low axial stress, designs do exist

It does sound a bit like Flakt were not too happy about the fitting of a 'home made' impeller 10 years ago. "Cap in hand" might be required.

34point5,

Your question about bearing configuration just reminded me of another bearing story I have recently become aware of:

The early models of the gearbox in my car had tapered roller bearings with a specified pre-load applied during the life of the gearbox design this was changed to ball races with no pre-load with I believe a resultant improvement in the durability and longevity of the gearboxes.

It may be unrelated, then again it may be relevant

Bill

No - I think suspect it's very relevant - in an impeller design 'miss match' way - but also I'm not getting enough info to know 'who did what', to help much. I.e. as of post #1?

I guess it's the usual 'tease out the data' process with OP questions we have to go through.

-Thermally induced unbalance/misalignment.

-Open the damper 100% and watch for change in vibration.

Carry out bump test with vibration analyserand find natural frequency of the rotor.

Impeller showing stiffining ring on inlet side

Impeller showing stiffening ring on back side

impeller crack

Failed bearing

DE Bearing

Looking down on fans 2 & 3 (the red bits are the motors)

RHP 22228 spherical roller bearing

This bearing was in service for around 14 mths before failing. Gas cutting of inner race created the heat marks visible on the surface. All damage visible on the inner race, other than the pitted inner race damage, was caused during removal of the bearing.

Flakt claim that properly designed antithrust vanes will reduce axial thrust by 70% - 80%.

Current axial load to radial load ratio is 1.63

We have 4 of these fans and all get insitu crack repairs when a bearing failure occurrs as we wear 30 hrs downtime to change a drive end bearing. They are dynamically balanced when they are changed out at about 4 years. I don't know how much FEA and flow modeling was done by Flakt but 20 years ago you needed access to a supercomputer for anything meaningfull, though todays laptops must be getting close to a 20 year old cray.

I have vibration spectrum data for the last 4 years for the Fan DE & NDE bearings and the Motor DE & NDE bearings, should i be looking at blade pass frequencies for fatigue cracks in the impeller?

Regards

Henry

Haven't got a proper handle on it yet. What's the fan OD and end plate metal thickness?

My first reaction on actually sighting your fan and hazarding a guess at the size/s is involved is that the tensile stresses at the outer edge of the inlet disc could well be high as a result of centrifugal force, and made worse by running at a hotter temperature than when fabricated AND the MASSIVE stress raisers that arise out of what appears to be the blades being welded to the end plate.

And if the issue is not straight tensile stress with some miscellaneous vibration added, then the next issue is that the end plates look quite thin and there is the question of cyclic air pressure variation "oil canning" the end plates and that the stress from this (added to the centrifugally induced tensile stress) at the blade/plate junction is too much.

Solution - take Flakt's word on the reduction in axle load with proper thrust blading, and that ought to get rid of the obvious bearing issue, then somehow reduce the stress and or stress concentration at the blade/disc interface.

Thinking out loud - maybe welding tag/s to the blade outer ends and bolting these to the disc inboard of the outer edge of the disc, maybe a strengthening ring and attention to surface finish on the outer edge of the inlet disc and not taking the blade weld all the way to its outer end. There will be a number of variations on this theme with a bit more thinking.

A couple of thoughts:

On the impeller - perhaps fillets between the vanes and the end/side plates would spread the load, stiffen the structure against vibration and reduce the incidence of damage in this location.

On the bearing - the damage appears to be spalling damage or something similar due to excessive axial loading, possibly including some vibration.

I conclude this because the bearing face in one direction is pristine and in the other direction destroyed

Here are a few references which you may find helpful:
<http://www.machinerylubrication.com/Read/664/wear-bearings-gears>
<http://www.lubeng.com.au/assets/files/technitips/096_How%20Bearings%20Fail.pdf>
<http://www.timken.com/en-us/products/maintdiag/Documents/6347.pdf>
<http://www.vibanalysis.co.uk/technical/contents.html>

I searched on 'Spalling of bearings'

Perhaps the inclusion of a thrust bearing in your assembly to take the thrust load off the roller bearing assembly may be the quickest and cheapest solution.

A thrust bearing could be installed almost anywhere along the shaft, and the inclusion of suitable instrumentation (a load plate) could also measure the axial loading, thus enabling the appropriate preload to be applied to the thrust bearing.

Regards

Bill

My thinking on the impeller fracture is it's vibration fatigue.

And bearing failure is a component it that

My reaction to the fan in jig is; it's under-shafted in diameter terms.

I.e shaft is too flexible for the duty.

Which will lead to bearing failure.

Seems a circular argument?

Perhaps consider a spider bearing carrier in the intake duct, carrying additional thrust capacity?

Shaft end support is - I think - worth looking at.

May be 'complicated' by the thermal aspects in this scale of gear, if you are not running 24/7

P.s. yes you can reduce the thrust with 'intake' vanes, but I'm not convinced it is thrust alone that is causing the bearing or impeller fractures.

I am thinking that that the roller bearings look like spherical rollers so that flexing of the shaft will not be so critical, nor critical enough - and in any case given the total absence of spalling on one of the dual races, that shaft flex is not a strong candidate.

And I can't see vibration of the bearing being an issue unless, and it is quite possible, that a natural frequency of the "sheet" in the fan matches one of the major harmonics.

All speculation of course in the absence of some solid maths.