Shaft Centrifugal Pump failed by torsion or fatigue?? 1

[ridwansunarya]

Dear All,
I'm new here, i've got a problem with this stuff. Anyone care to look at this failed shaft?
It was a centrifugal pump with double suction enclosed type impeller and shaft supported with plane bearing on both ends
I'd like to know if it is torsion or fatigue or even something else that might cause it to break like this?
Material : A276 316L
Fluids service : MDEA Semi-Lean Solution
RPM : 2980
Suction Pressure (Max/Rate) : 9.0/3.8 kg/cm2G
Discharge Pressure : 36.4 kg/cm2G

Thanks !

 

[metengr]

The fracture surface on first glance appears to be fatigue crack initiation and propagation versus torsional overload. This is based on the mostly flat fracture appearance. I would send this failed piece out for a proper metallurgical analysis. It also looks as though there is a transition or step in the shaft diameter at this location.

 

[EdStainless]

I agree with Metengr, when you look at the side of the shaft is there any sign of twisting? Generally if a shaft fails in torsion overload there will be some twisting distortion.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[ridwansunarya]

Thanks guys,i agree with you
But i think its not overload, but from torsional fatigue because final fracture was smaller than total area of the shaft. It means stress which applied on the shaft was below the UTS, isn't??
Besides, after i look at microstructure there was many presence of twinning grain, is there any relation between twinning grain and torsional stress (plastic deformation) ??
Could you help me again??

Thanks for your kindness

 

[EdStainless]

The twining is probably from the original cold work, I doubt that the shaft is annealed as soft as 316L is.
Is there a step in the shaft where it broke? The stress concentration at the step would really accelerate fatigue at that location.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[ridwansunarya]

There is step on shaft for placing sort of split ring, but it was a radius not taper,so i think the stress concentration on this place couldn't be a main problem, though that would accelerate crack propagation still.
So,torsional stress doesnt have effect to twinning grain??
Im suspicious about one thing, is it possible that microstructure like this because of heating while operation??
because recently i just know that treatment was conducted on the impeller . But from hardness test, there are no different value on center shaft and its surface.


Thank for your attention

 

[metengr]

You should send this failed section out for proper metallurgical analysis.

 

[blacksmith37]

Torsional fatigue will be at 45 to the axis; bending fatigue will be at 90 to axis.

 

[Tmoose]

Where on the shaft did the break(s) occur?

https://www.ksb.com/blob/189738/7fec819715ee3d7ea923ba96e4891148/pipelinepumpe-1-data.jpg

- If Direct coupled, type of coupling ?
- Belt driven?
- Service life and repair history?
- How many of these pumps do you have?


I believe shaft features like grooves or keys can modify the "torsion = 45 degrees " rule sometimes.
Fig 3.3 page 14 here -

http://www.rexnord.com/ContentItems...010_Failure-Analysis-Gears,-Shafts,-Bearings,

 

[EdStainless]

Even a radiused groove will greatly increase local stress.
If they heated the impeller they could have caused other problems in the structure.
Andy stress relief would make that section softer, and it would have lower fatigue strength.
If they got it hot enough they could have sensitized it, and this would have much lower ductility and lower fatigue strength.
Were there valves in this system that were opening and closing frequently? Pressure pulsations would greatly accelerate fatigue.
Was the impeller well balanced?

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[mrfailure]

ASM has a figure available in their handbooks that illustrates fatigue morphology under different conditions, including rotational versus torsional fatigue. It is worth looking up.

That said, I find it interesting that the fatigue at face value appears to initiate in the interior of the shaft and then propagate mostly to the left as if it were under unidirectional bending. Final fracture occurred to the right shortly after the crack reached the OD surface on the left. I would be very focused on why this appears so, especially since most fatigue cracks are expected in this environment to initiate on the surface.

 

[Guest102023]

The area of final overload (at approx. 1/2R) is extremely small, indicating low mean stress (unsurprising for a shaft).
Looks to me like the main fatigue crack started at left, and later a sympathetic fatigue crack started at the right side.
The morphology suggests a shaft straightness and/or alignment problem, again unsurprising for a shaft.
Enquire if the equipment was recently disassembled.

mrfailure, I keep a photocopy of the handy dandy ASM graphic aid taped to my office wall

"If you don't have time to do the job right the first time, when are you going to find time to repair it?"

 

[Maui]

The image of the fracture surface you provided is inconsistent with failure by torsional fatigue. It is far too flat. Here is a paper that discusses torsional failures in steel alloys:

http://www.sciencedirect.com/science/article/pii/016766369400039J

Maui

 

[tbuelna]

The photo provided in the OP appears to show some radial clearance between the hub and shaft, and only a key installed in the lower groove but not in the upper groove. These factors may have contributed to the shaft failure.