Two relief valves in parallel with set pressures not close to each other

[chemks2012]

Dear all

One more query please

Vessel operating pressure is 5barg
Vessel design pressure is 8barg

Our team has decided following

1) 1" relief valve set pressure 7barg. Assume, it's designed for 300kg/hr steam.
2) 1.5" relief valve set pressure 8barg. Assume it's designed for 600kg/hr steam flow rate.

Please note second valve was added to add extra layer of protection for a higher flow rate in case of rarely possible worst case scenario assuming first one does not lift.

Thanks
KS

 

[EmmanuelTop]

Whether it is rarely or likely possible, the PSV needs to be designed for the worst credible scenario. Hence I do not understand what is the purpose of the PSV No.1 because it will be useless in the worst-case overpressure event and it would probably cause the PSV No.2 to chatter, during the same event.

If you want to increase reliability of the system you could install two identical PSV's in parallel (one online, one standby), both designed for the worst case. That way you can perform inspection whenever you consider it suitable and make sure everything will work as intended.

Dejan IVANOVIC
Process Engineer, MSChE

 

[zdas04]

I can't support that conclusion. Say you have two credible scenarios. The most likely one has to protect against a flow rate of 300 kg/hr. Relieving at 600 kg/hr would cause this valve to lift multiple times as the drawdown is twice the inflow rate, you could easily damage something by slamming the valve open and shut repeatedly.

If the less likely event happens, then the small valve will lift at 7 barg and the pressure will continue to increase (more slowly) to the setpoint of the other valve. At that point the larger valve will be passing twice the required mass and will do the same cycling thing that the would happen if the first valve were twice the required capacity. My take on this is that if the small credible scenario is more likely then I prefer the OP design. If the large credible scenario is more likely then that is the only one I'm going to protect against.

I never sanction two PSV in parallel with the same setpoint. I've seen this scenario cause problems too many times. Each valve has a dead band and one will lift before the other one every time. But the second valve is close enough to its setpoint that it will tend to chatter, creating cyclical pressure transients that can do harm.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[chemks2012]

Thanks for your input Dejan and David

Please note that second valve is there just to protect the valve in case first valve doesn't open for some reason.

To simplify my query, please assume both valves are designed for the required release rate of 300kg/hr.

With these, is it not fair to consider that the first valve will open almost every time when required and say second valve will say never opens and when second valve opens only when first valve is blocked or doesn't activate?

Thanks in advance for your help.
KS

 

[zdas04]

It all depends on your credible scenarios. If you only have one then your strategy needs to address that, if you have two, then there are several ways to address that.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[chemks2012]

Thanks David,

Do you think 1barg pressure difference in the set pressure of both valves would cause any problem (chattering)?

Thanks
KS

 

[zdas04]

You have to find out what the dead band is on the valves you are considering, and you want to be at least twice that value between the setpoints. That number is usually around ±0.5% of setpoint, so for an 8 barg setpoint, the range is ±4 kPa so 1 barg between the setpoints would be more than enough to minimize the risk of interference.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[georgeverghese]

Dont understand why you have this first PSV at 7barg - why cant it be at 8barg.

And the second one could be at 8x1.05 barg. With relieving pressure for both at 8x1.1 barg

Are you sure that steam pressure is always at 5barg? What pressure can it get up to in normal operations when there is a big load swing?

If you have the first PSV at 7barg, and assuming it is a conventional PSV , it will start to chatter / simmer at 0.9x7barg. So that would limit your max operating pressure to say 0.85x7barg ?

If you set the first PSV at 8barg, then your max operating pressure could be as high as 0.85x8barg

 

[EmmanuelTop]

I had a similar discussion recently on the subject of oversized PSV orifices, and "how much less flow is still good enough" to prevent from chattering. The figure I found in the Fluor Daniels design manual says 25% of the rated flow for conventional, spring-loaded PSV's in gas/vapor service. For PSV's discharging liquid, the limit is around 40% of rated flow.

Having this in mind, the arrangement of 2x100% PSV's sized for 600 kg/hr flow should be OK also for the 300 kg/hr case. I have proposed two PSV's only because the OP wanted an increased degree of reliability. If there are two online PSV's - one for 300 kg/hr and the other one for 600 kg/hr - I don't see the point of such arrangement. With the set points staggered as proposed, the smaller PSV would release more than 300 kg/hr at the pressure point when the larger PSV lifts, and that could cause the larger PSV to starve of flow and ultimately get damaged.


Dejan IVANOVIC
Process Engineer, MSChE

 

[LittleInch]

This is all madness to me. A simple spring loaded relief valve is simply the most reliable piece of protection kit on the planet. It's not a "possible worst case scenario", it's not a credible event. If you want to somehow add more reliability, think different relief system (bursting disc for instance), but these are also introducing other issues. Your relief should be the last line of defence, not the first and adding more and more layers does not always mean greater and greater reliability....

All this is doing is solving a problem which doesn't exist, IMHO, and creating several more.

Dave, I would love to know where you can find spring relief valves with a 0.5% accuracy / dead band - did you mean 5%??. Most simple non pilot ones I have used are ~8-10% and pilot ones ~3%.

So 1 bar between these two is still too close as these bands overlap (one high, one low)

The only way you really want multiple reliefs is when you have credible scenarios with quite wide flows at different pressures. However as zdas 04 says, you need to make sure you have sufficiently high step points if you're going down that way.


Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[zdas04]

It was late at night and I just pulled the number from the air for an example to show the concept. It has been years since I actually looked at the dead band on a PSV. 8-10% sounds high, but I don't really have anything to base that on.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[chemks2012]

Hello all
Thanks very much for your valuable input.

David,
Thanks for your post regarding dead band. Please let me know if I can get information dead band from relief valve suppliers?

Littleink,
Agree. It's total madness to me too but what would you do if you are being challenged by higher management and asked what if relief valve doesn't open?

In past, I have seen following arrangement in parallel.

2"x3" relief valve , set pressure of 2barg, designed for vapor release rate of say 1kg/s

12" bursting disc , set pressure of 3barg, designed for two phase release rate of say 150kg/s

As you can see above two release rates are quite apart from each other however there is no guide in API on when to consider another relief device I.e. At what difference in flowrate should be considered apart enough and hence the confusion.

Thanks
KS

 

[LittleInch]

What would I do?

Find a different type of ultimate relief system - the bent pin idea on your other thread sounds good.

Also point out the layers of protection that you have before the relief valve lifts.

Find the reliability figures of the relief valve - i.e. probability of failure on demand

Point out that adding a second relief valve actually causes significant problems

Ask them where does it stop? If one relief doesn't open, what about the second or the third or fourth.... That's why people stop at double jeapordy - you can't look at multiple simultaneous failures because you never stop adding more layers of protection.

Ask then how they think a spring relief valve will fail to open. The problem is clearly their perception of failure. The normal issue with a relief valve is that it opens or starts to chatter below the set point.

Accuracy on set point is provided by the valve vendor, usually something like -8% + 5% from set point.

If one flow is less than 40% of the max flow then I think that is a good number to use for when you need different staggered relief systems





Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[MortenA]

I would also recommend something else as a back-up, e.g. instrumented relief (SIL rated if need be). You can add to the reliability of this by using, ESD system independent of the PCS etc.

Best regards, Morten

 

[e43u8]

Referring to following statement of OP in his/her second post:

"To simplify my query, please assume both valves are designed for the required release rate of 300kg/hr.

With these, is it not fair to consider that the first valve will open almost every time when required and say second valve will say never opens and when second valve opens only when first valve is blocked or doesn't activate?"


If every of two valves would handle the same worst case scenario but with different set points, although the configuration isn't common to be executed at all, it not only wo'nt lead to major peoblem in realistic safety/protection standpoint but also will create a back up protection against the rare case of the first valve not to be opened when is required to be opened but in the expense of an additional valve mostly isn't required to be installed.
In fact, if in rare case the first valve not to be opened to protect the vessel against overpressure, the second valve will open at higher pressure and will do the job. If in any case, during second valve relieving the first valve will suddenly open; the second valve will be closed due to higher set point and sudden pressure drop than its set point and the first opened valve will continue to do the the job.

Of course, the configuration isn't attractive in economical stand point and also norrowing the margin between operating pressure and design pressure by considering a set pressure below the vessel design pressure for the first valve...

 

[EmmanuelTop]

Rather than just listening to your manager wobbling around, you should demonstrate how effective are the available layers of protection and if any further design measures need to be taken. This is why HAZOP/LOPA are done in the first place.

Performing LOPA analysis within your team should give you all the answers you need. Consider how many over-pressure scenarios exist and choose the most severe consequence, then dig the information on what is the probability for each protective layer to fail on demand (PFD), and what is the required TMEL (Target Mitigated Event Likelihood) based on consequence severity. If you end up with IEL>TMEL, you need to do something. If you can prove that IEL<TMEL, your boss will shut up.


Dejan IVANOVIC
Process Engineer, MSChE

 

[chemks2012]

Thank you all for your input.

Will update you soon!

Thanks
KS