Cooling tower Fan failures

[cedeng]

Hi
I'm working on the case of our cooling tower fan (Hamon-Sobelco fan) whom brake in the blade holder (good term?).

Blade holder is cracking at bolt hole (which retain blades) and advance in fatigue propagation. It is 40'' diameter and 1 inch thick.

What could be the cause of failure ? Vibration ?(sure) Flow and pressure of air sucked? I don't know where to start, I'm a beginner in fan modelization.

Material is NOT A POINT TO BE CONSIDER, after fracture analysis, we conclude that fatigue failure is occuring even in stronger material.

Assembly of the blades are as this picture on the Howden site and blades are FRP.

http://www.howdencoolingfans.com/files/upload/DSC09355 zonder roes_groot.jpg

 

[GregLocock]

Nice big toys. I don't really understand what you want, but I think you are unwise to eliminate fatigue so quickly.

Do you undertsand the theory of crack initiations?

What number of cycles (revolutions) are you getting failures at?

Is this a one-off unti or do you have many units taht all fail?



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[hacksaw]

I agree that you need to be more specific.

Have you performed a stress analysis.

 

[ivymike]

sounds like fatigue has NOT been eliminated ("fatigue failure is occuring") from consideration.

Is the crack initiating on the "saddle" side, or on the nut side? What is the direction of propagation? Is the saddle a very stiff component, or relatively flexible?

 

[GregLocock]

From the photo it looks soft, are there ribs underneath? Sorry, I meant don't eliminate material from the investigation.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[hacksaw]

Gregg,

I am surprized that their design is not using through-holes on the pillow blocks (aluminum blade supports on their web site).

They may be creating quite a stress riser between the edge of their Al support and the bolt hole.

 

[ivymike]

I was thinking the same thing. It sounds like there is significant fluctuation in the bolt load, though (seems undesireable), and I figured I'd wait for more info.

 

[vanstoja]

For airfoil blades of this huge diameter the blade support system appears to be about as flimsy as I have ever seen and presuming that bolts are holding down the two U-shaped brackets for each blade, it is no wonder that they might easily be overstressed by even ideal flow conditions over the blades. If they are operating at off-design flowrates then lift fluctuations on the blades could be atrocious and lead to early failure of the bracket holddown bolts. The blade support designer appears to warrant hanging from the most distressed blade.

 

[electricpete]

Good one Johm V!

I'm not a mechanical engineer but I agree with John... it sure looks flimsy to me. The separation distance of the two U-type clamps is about the same as the diameter of the blade stub. With this small separation distance, the force for a given moment on the blades is magnified. Longer blade stup with bigger separation distance between clamps would seem a better design. Even better continue to blade stubs to a center point and join them all together somehow. None of those are practical modifications after the machine is built, but it seems like the machine should have been designed that way to begin with.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[hacksaw]

Don't forget that it is the plate that is developing cracks, cracks that start at the bolt hole. The blades are FRP.

 

[electricpete]

OK, Diameter of the blade stub wasn't a relevant point of reference for me to mention.

The relevant comparison of moment arms I should have talked about is the small distance between clamps compared to the large length of the blade.

If forces act on average 2/3 way out the blade, my eyeball say the moment arm between those clamps is 1/10 of that distance and the clamps will see 10x the force that was on the blade. I agree with you guys 100% the stress-riser created by configuration of the U-bolt through the plate is very important.

To understand the forces, it would be interesting to know the speed of the machine, vibration magnitude, vibration measurement point, vibration frequency content. Also might be interesting to look at it with a strobe.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

Does this really need to be discussed?

The design LOOKS very weak. It fails. So, what's the problem? It seems to me to be a classic case of bad design.

Gunnar Englund

http://www.gke.org

 

[MikeHalloran]

A quote: "Working with reinforced composites and aerodynamics is our ‘core technology’ at Howden Cooling Fans." It may be time for them to hire some expertise in stress, and in materials selection.

Your lawyers should be strongly encouraging them to do so.



Mike Halloran
Pembroke Pines, FL, USA

 

[Tmoose]

Can you post pictures of some broken ones?
Do the failed hubs have cracks with multiple forks, kind of like lightning bolts? That would be stress corrosion, which can happen at pretty low constant stress levels to some aluminums if they are operating in the wrong atmosphere.

Although the hub does look pretty flimsy, Howden has been around forever, so I'd expect them to have worked things out pretty well. Similar fans apper in their literature here -

http://www.howden.com/library_section/pdf/Howden Cooling Fans.pdf

 

[cedeng]

Thanks for the tips.... Here are some more informations...

- Good idea about getting more distance between both u-bolts to give more stiffness to the assembly.
- Plate of blades holder was in carbon steel. Bad corrosion from sulfuric acid environment forced to change for stainless 316L. It was cracking in steel and worst in 316L. So we tried Ferralium 255 (Duplex 2205 steel). It is stronger and cracks as well. But slower.

- I'll post some picture on monday, I promise. You are so kind answering and trying to find a solution, I owe you some pictures, at least. A picture of the fracture surface will be posted, and you'll be able to see that crack propagates slowly...

- Also, I'd like to try to put some stress probe near bolts holes,to monitor the stress, on both side of the blades holder. I guess that stress is not the same on both side.

- What do you think, guys, about shotpeening the stressed surface of my plate?

- Supplier was Hammon-Sobelco,(I used Howden site because of nice picture on their site).

Waiting your suggestion. Project is on going...for year 2006.

 

[MikeHalloran]

I don't know what a stress probe will tell you that the cracks won't.

Doubling the thickness of the plate will help more than shotpeening will.

Which leaves you with the dilemma of getting longer u-bolts, or counterboring the plate for their nuts.

I'd toss them, and use through bolts, double the number of saddles, and two plates, sandwiching everything between them.







Mike Halloran
Pembroke Pines, FL, USA