Relief Valve Failure Cause

[jgar1]

I am evaluating a failure of a relief valve for liquid application to lift at the required set pressure during surveillance testing. The valve lifted 6% high. It is an obsolete Crosby JB-36 valve.

I read some operating experience where an engineer attributed a failure to lift (lifting high) to spring relaxation and cold working. Spring relaxation has the effect of reducing the set pressure of the spring. The theory is that adjusting the spring over time because of spring relaxation can have an opposite effect - the effect of work hardening the spring, leading to a higher required pressure to lift the valve.

Is this a well-accepted phenomena / failure cause?

 

[dcasto]

Can't help on mechanical, but can you describr the fluid and where the relief valve is piped to? Is it piped to a flare/vent header to atmosphere?

How do you know it only moved 6%, that would be a few hundreds of an inch? The valve will start opening at set pressure and will not go full open until between 105% and and 110% of set pressure.

 

[Ashereng]

jgar1,

Just because something is well accepted and is true, it doesn't mean that it is the cause of your specific problem.

Have you also considered back pressure?

We usually use the Crosby JOS line, so I am assuming that the JB 36 is a conventional PSV.

It is entirely possible that the springs go weak over time (heat/cool cycles, fatigue, etc). However, you need proof of this (from testing, investigations).

I am also curious as to how you knew the valve only lifted 6% high.

"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?

 

[jgar1]

Sorry about the confusion. What I meant was the as-found set pressure when it was tested (on a test bench) was 6% higher than what it was set for (it's as-left pressure). Meaning the valve is supposed to lift at 600 psig and it lifted at 636 psig. Then again at 627 psig. The test medium and medium in service is water. It is piped to the pressurizer relief tank (at a pressurized water reactor)which is normally at 3-4 psi.

Unfortunately the valve was readjusted and placed back in service before we could do any internal inspections. So for now I'm forced to come up with a "best" explanation, based on valve history and age of course.

 

[tkdhwjd]

If the valve fully opened within 110% of the set pressure (6% in your case), why do you call it a "failure"? SRV generally starts simmering slightly below the set pressure and is supposed to fully open below 110% of the set pressure.

 

[jgar1]

It's supposed to crack open (initiate lift) at the set pressure +/- 3% or within a range specified by the owner. 10% accumulation is allowed but a verifiable lift beyond just simmering should initiate at the set pressure.

 

[gerhardl]

Jgar1,

OK. You are describing testing an object (the PSV) with two tests only 636 and 627 psi. The two tests are not more than 1,5% apart. This is indicating that the accuracy of repetitive action is fairly good, perhaps within acceptable variation for the PSV. (You need more tests to prove this to search for failure course - but should of course anyway change the spring before putting the valve back in service)

Then you have the 'systematic' variation, claiming that the valve is systematically about 6% off target. To prove this you would have to prove that all conditions are exactly equal when the valve was first set originally compared to the second tests, exactly same testrig, and that the valve was sealed and untampered with inbetween.

In a closed pipeline system (pressurized tank for outlet) and a manual operated test procedure, and with an 'unproven' (?) original set point for the valve, I am more leaning to believe that 'outside influence' rather than a suddenly occuring mechanical variation in spring material is the cause.

As given from others above, you could list a lot of possible explanations, the simplest being variatons between tests (if any?) when first setting the original set-point and the second pressure test. Even human error and calibrations of pressure reading instruments will contribute.

 

[Ashereng]

Or, maybe it was set wrong the first time?

"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?

 

[JLSeagull]

How often is the valve tested? If recorded, how much change in the setting has been logged over the last few tests? Is the test rig calibration current?

 

[jgar1]

The valve is tested every 2 or 3 years. Usually it lifts lower than the set pressure or not at all due to leakby.

We've already considered a lot of possibilities, test rig, pressure reading calibration, human error, procedural guidance, etc. The unfortunate truth is we may never know for sure because the valve was never taken apart for an inspection.

And I'm already convinced that my original curiosity has been satisfied, work-hardening of the spring is not what has happened.

Having said that, I am content with the outcome of this thread and I am greatful for everybody's input.

 

[sjrfc2]

The temperature of the test facility could have been a factor when it was originally set. I looked through some info on the NRC website several years ago and if the ambient temperature was different than the lab temp it could have an impact. I think I remember about 5% being noted. I think I remember as much as 4-8% setpoint shift occurring due to either fluid temperature effects or whether the valve is lagged vs unlagged. I think Information Notices 89-90 and 96-03 might have info. I'm sure there's more if you search through the NRC docs. I haven't looked at that for a long time.

 

[JAlton]

The Crosby JB style is a bellows design. This is an obsolete valve replaced in the 1980s by the JBS design.
The 36 indicates high temperature >450F to 800F. There may have been a temperature correction applied during the previous test/adjustment. If the bonnet vent was plugged, it could have caused pressure traped in the bonnet toprevent the bellows from flexing, thereby rendering the set pressure higher. Also, if you got 636 psi the first time and 627 psi the second, you should have made a third lift to determine if the Set Pressure was trending downward. ASME PTC 25 requires Set Pressure to be established and stabilized with no significant trend either upward or downward. This means that three (3) pops is the minimum in order to establish a trend. If the spring is trending, you need a new spring. I know you don't have a definitive answer, but those are just some things to consider.

JAC