17-7PH Spring Surge Fatigue

[Honel]

Hello all, I have a question for which there seems to be little guidance. I have a spring in a nozzle check valve that is failing due to what appears to be fatigue in a very short time. A few calculations showed me that the PD pump upstream was pulsing within 6Hz of the surge frequency of the spring (250Hz vs 256Hz) - which puts me pretty firmly in the surge peak. Due to system constraints, I am not going to be able to replace the spring with an adequate one, and the valve will be changed - the valve is also inaccessible for the next year or so. What I would like to do is predict (within a range of course) how many minutes of pump time I can expect the spring to survive. Spring specs are as follows
Wire Dia - 0.063in
Free Length - 1.196in
Solid Height - 0.308in
Squared and Ground Ends,
4.89 total coils (2.89 active)
The spring is required to exert 6.48lbs at 0.821in long, and 12.96lbs at 0.446in long.
From these numbers spring rate is ~17.28 lb/in.
Matl. is 17-7ph tempered

I don't have information on the tempering time/temp, so all I really need here is the methodology and I can play with the rest.

The few calculations I have done using Goodman criteria give me a fatigue life of around 100k cycles - which when cycling at 250Hz only lasts ~7 minutes. I am not sure I believe this number, but there is pretty sparse (publicly available) information once you're in the range of resonance

Thanks for the help.
Wes

 

[israelkk]

What is the spring outside diameter?
Is the spring cycle between the 0.821in long to the 0.44in long?
Surge and resonance are not the same effects. Surge is due to impact force.
Tempered is not a term that is used in the heat treatment of precipitation hardened alloys such as 17-7PH.
Springs are usually heat treated by an aging process to CH900. The wire is preconditioned by the steel wire manufacturer to "C" (Cold drawn) condition. After coiling in is aged at 900F to CH900.

 

[israelkk]

I have quickly checked the spring life cycle depends on the spring OD and it seems that something is not quite right in the spring design. Before knowing the spring OD (outside diameter) it is not possible to give an answer.

Can you provide more information on the processes and heat treatment done to the spring?

 

[Honel]

Ah, sorry I didn't realize I had forgotten the outside diameter. I will post it Monday when I get back to work and have the data sheet in front of me. I believe it is somewhere in the 0.750in range though. As far as the temper comment - I am not exactly a spring expert, that is what is stated in the data sheet. Part of the reason I am digging so far into this issue is due to the fact that I have found several problems with the vendors math/this particular data sheet and am approaching the end of my depth on the subject.

The .821 to .44 is the stroke of the valve - so it will be stoked that distance any time the valve is actuated. What I have no idea about is how the poppet of the valve responds to the pressure pulses caused by the pump. Obviously if the spring is in a fixed-fixed end condition (on its stop) the primary frequencies change by a factor of 2. If the pulses are large enough to physically move or vibrate the poppet on its stop then it is a different story. Judging by the fact that I failed the springs on two different valves with less than 200 activations each - there has to be something going on. One poppet failed, as well as both springs, all with indication of fatigue being the culprit.

Thanks
Wes

 

[desertfox]

Hi
I found this on spring surging:-

http://www.engr.colostate.edu/~dga/mech324/handouts/spring_surge.pdf

Go down to the heading compression spring surge.

I estimated the mean diameter of the spring from the other information you gave to be 0.782" so that figure of .75" wasn't a bad memory recall.

 

[Honel]

Desert fox,
Those equations (the same as those ones found in shigleys mechanical design) are the ones I used to calculate the frequency above. Also, the number you have for mean diameter sounds correct. I will confirm tomorrow though. Thanks.

Israel,
Surge is when the coils are collapsed on top of one another due to actuation from one end that exceeds the springs ability to get out of the way of itself, essentially causing two coils to be solid or touching correct? Resonance is the frequency at which that wave propagates end to end thru the spring. In practicality though, when the forcing frequency is so close to that resonance peak do you not enter surge like conditions? If not what actually causes the cyclic fatigue that happens in resonant conditions? If the frequency is ~250Hz, does 250 Hz= 250 cycles of the spring?

This is where my question stems from. There is only fuzzy guidance about what actually is going on under these conditions in the literature I have found. Mostly just advice to steer clear of anywhere close to resonant frequencies.

I am going to replace these valves at the next opportunity so that this is not an issue. I would however like to have some idea of what margins we are currently working with as far as how many minutes of pump run we can withstand before fatigue becomes a concern.

 

[desertfox]

hi Honel
I'm inclined to agree with you that the spring is most likely surging and this condition can increase the stress range by 50%, so I think the springs are failing in fatigue due to the surging.
The actual spring stresses at the two loads and displacement figures you posted are 57,839lbs/in^2 and 115,679 lbs/in^2.
It would be interesting to get the stress limits for the material you have so we can compare with these I have calculated.

 

[israelkk]

Honel

Still waiting for more info about the spring manufacturing processes and the spring O.D.
From my quick check you have a life cycle issue without even resonance and surge.
From the info you already posted and checking for the O.D. that will give the spring properties you posted, it is clear that this spring is not designed properly for enough life cycles to last. More than that it will not even last two thousands compressions from 0.821in long to 0.446in long.

Therefore, to my analysis there is no need to go to the surge theory.

For surge theory and where it is applicable see Dr. Karl Mayer article at the book SPRING DESIGN AND APPLICATION from Nicholas P. Chironis, Pages 63 to 70.

 

[Honel]

All,
Mean Spring Diameter per data sheet is 0.757in.

I'm afraid I don't have more information on the material other that what the data sheet lists.
From the data sheet - "Stainless steel 17-7PH Wire Spring Tempered" is all I've got.

Israel, I look forward to reading through that reference (I'm a total reference junky) - also, thank you for confirming that the spring is inadequate. The thing that is driving my inquiries however is the fact that each of these valves have been cycled fewer than 200 times a piece. So even 2000 cycles should have given a longer lifespan than we have seen - which lead me down the surge/resonance path.

Outside the information in the given reference (haven't had time to look it up yet, if all these answers are contained, then I suppose these questions are unnecessary), One thing that would be helpful for me to know would be IF the spring is indeed resonating, does the frequency of resonance correspond to cycles of the spring? ie 1 cycle of the spring per Hz or is there no direct correlation?

 

[israelkk]

Honel

As I understand the check valve should have a plunger that is forced by the spring to seal the inlet/outlet orifice/s of the valve. Did you take the plunger mass in your spring-plunger mass resonance frequency? The plunger may weighs more than the spring.

The 2000 cycles is an estimation since fatigue analysis depends on surface conditions, material quality and many more parameters. As I mentioned previously a fatigue design should have a safety of factor of 10. Therefore, 200 cycles may be in agreement with the analysis. Any surface defect such as scratch, tooling marks, nicks, etc. on the spring wire surface is a cause for fatigue. The fact that the analysis gives a result of 2000 cycles means that the working stresses are too high.

Working in the resonance frequency of the spring can cause the spring an erratic behavior and unless there is a positive stop that limits the spring deflection from 0.821in long to 0.446in long, the spring can even deflects up to solid which will cause even higher stresses.

https://sites.google.com/site/israelkk/

 

[Honel]

Israel,
Due to the fact that this component must undergo seismic qualification in its current position, the spring/mass relationship has been evaluated and is around 39Hz. However, that is the free response of the mass spring system. If the free end is driven at a certain frequency (with the poppet being the driver as would be the case in a flow/pulsation perturbation) then the mass of the poppet (or plunger) becomes subject to the combination of the flow and the spring. So while the free response of the spring mass is appropriate for seismic analysis, it is insufficient for actual response of the spring mass system to in-use conditions. It is apparent from pictures of the failed springs that the coils were making frequent contact with one another.

Also, there are no mechanical stops for the spring in the compressed direction.

So it seems (unless someone has any counter ideas) that my original estimation of 3-7 minutes of pump time to failure is not far fetched at all?

Thanks for all the help so far guys.

 

[desertfox]

Hi Honel

I agree that the spring in question is failing due to fatigue and this is premature due to the fact the spring is being driven very close to its natural frequency also known as surging, the seven minutes life may or not be correct because it's hard to analyse the stresses during that phenomena due to the fact that deflection of the spring coils occurs across only one or two coils to start with and then progresses like a wave through the spring.
It is easier to avoid the the natural frequency during the initial spring design if you know the cyclic duty before hand.

What is clear even from my hand calculation is that even if the spring wasn't surging and this is exactly what Israelkk is alluding to, then your spring stress are still to high to avoid fatigue.

The best course of action would be to see if the spring can be redesigned and in conjunction with a spring maker who will probably have a wealth of knowledge that you can benefit from.

 

[dgallup]

I ran that spring through ASD7 and it says the natural frequency is 526 Hz and an estimated life of 9 million cycles. I really would look into the heat treatment, if that is not right then you won't get anything like the expected life.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[israelkk]

dgallup

To my best knowledge and experience I suspect the ASD7 program has an error in the life cycle calculation for a spring without preset (set remove, cragging). The 9 million cycles may be correct only for a spring with a heavy preset operation as the last manufacturing operation. For a spring without preset it should be around 2000 cycles.

https://sites.google.com/site/israelkk/

 

[Honel]

dgallup,
Thanks for that screenshot - that natural frequency would be for a spring between two fixed plates - for my purposes I believe a fixed plate/sinusoidal drive would be more appropriate which would be equivalent to a spring of twice the length resonance wise. (or would half the frequency).

In any case, I expect broken springs when I open these after this operating cycle