Predicting chattering

[vgarzani]

We are analyzing a system at a plant for a postulated accident and are predicting transient conditions. This is a water system and in the transient two phase conditions are predicted. Predicted seal leakage is an input to the system and the transients show peaks of high leakage flow into the system, but the leakage drops off after a minute or so and we are concerned both with predicting peak pressures as the peak flow likely overwhelms the valve capacity if mostly steam; and also the potential to start chattering. Can someone point me to references which address quanitifying conditions that lead to chatter? We will want to either show that chatter is not expected, or if we can't avoid it with procedures, then attempt to evaluated it. Understand evaluating chatter is not expected to be accurate... but there it is.

Any pointers or references would be appreciated.

VGARZANI

 

[Ashereng]

Chattering and/or simmering are both problems that cause seating/sealing to be damaged. Usually, this happens if the PSV is sized too big for the actual flow.

Have you tried talking to the PSV vendors to see if they have equations/tools to predict chattering/simmering?

Tyco, Farris are two that may have the information, or know where to get it.

"Do not worry about your problems with mathematics, I assure you mine are far greater."
Albert Einstein
Have you read FAQ731-376 to make the best use of Eng-Tips Forums?

 

[jrwchem]

I think this might contribute to chattering.

If the inlet frictional losses exceed the blowdown setting.

In the above case, once the PSV opens, the frictional losses will cause the PSV to close because the pressure at the PSV inlet will have dropped below the closing pressure (determined by the blowdown setting). The pressure will quickly rise again, and the PSV will open again, only to close once flow has started.

This is my understanding of chattering. It is fairly simple. I did see one reference to a paper that talked about chattering being a resonance frequency effect, but i dont have the reference handy.

Perhaps others can comment on my idea of chattering.

Regards,
Jonathan

 

[mark06]

(1) built up backpressure too high. 10% of set pressure for a conventional valve, 30% (i think) for a balanced valve.
(2) Inlet pressure drop too high. 3% of set pressure in hydraulic losses for a conventional or bellows valves. Inlet pressure drop does not affect remote sensed pilot valves.
(3) Valve capacity is too large. The rule of thumb is required flow / rated capacity should be greater than 25% for vapor and 10% for liquid.

Try API 520.

 

[JobDone]

Conventional safety valves will chatter for all the reasons that have been previously mentioned. These can all be solved using a Modulating Pilot Operated Relief Valve which can :
1. Tolerate high back pressure (well over 50%)
2. Can be set up with remote sense to overcome inlet losses.
3. Modulate to relieve only the capacity required, this means that oversized modulating pilot operated valves are not an issue.

Hope that this helps.

 

[dcasto]

Can I ask what you would do if you predicted chattering? I had a consultant that said the PSV would chatter and that we must change it out. I reality, every relief valve out there is subject to chattering. We assume blocked flow for example as a point where the relief valve goes off, what about PARTIALLY blocked flow? Partially blocked could be a more probable case, if so the relief valve will chatter, now what?

 

[vgarzani]

Thank you all for your posts. In response to the question what would we do if chattering would have been predicted: Well, we avoided the issue by making a procedure change which precluded having to wade into the what-if's that resulted in conditions that may have resulted in chattering or chugging. The scenario involved conditions resulting from a fire. We changed the fire response procedure. At the time of the initial post I was sure we would have to analyze our way through the mess. The whole thing certainly gave me an appreciation for challenges associated with pressure relief design. If relief capacity is not big enough pressure goes up; if it is too big unwanted dynamics may break things - so use more than one with staggered set points; or just avoid the issue in the first place, yeah that's it... Thanks again.