Inadvertent valve opening PSV case - also include max normal operating flow from other sources?

[Justice100]

Hi there,

I have a vessel that normally receives fluids from various sources. For the case where one of the sources has a control valve that can fail open would you consider the relief valve to be sized based on the maximum Cv of that control valve only (with the control valve pressure at the upstream trip pressure or design pressure) or should it also consider the maximum normal operating flow from the other sources?

Thanks in advance for any comments.

 

[Latexman]

The PSV should handle ALL credible scenarios of liquid, gas/vapor, and two-phase flow during normal operations, start-up, shutdown, maintenance, etc. Fire exposure, tube rupture, reactivity, reactivity and fire, heat imbalance, power/utility loss, dust deflagration, feeds with venting vapors, feeds with venting liquids, ambient heating, vacuum, etc.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[Justice100]

Ok, thanks Latexman for responding.

I understand there are various cases that need to be considered as you have detailed but I am trying to define the case I outlined.

 

[danschwind]

This vessel receives fluids from different sources simultaneously?

Daniel
Rio de Janeiro - Brazil

 

[RVAmeche]

The PSV should consider whichever is the worst case for that failed open valve scenario.

Typically you use the max Cv and max upstream pressure to determine the highest flow rate for that valve. If this scenario would occur while other sources are supplying product, they will also build pressure in the vessel.

 

[Latexman]

Assume 1 failure at a time. If the CV fails wide open, use the normal supply pressure and the destination vessel at it’s MAWP + allowed accumulation (usually MAWP x 1.1), plus whatever else would be going into the vessel at the time of the scenario.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[Justice100]

danschwind - yes, it is fed from other sources simultaneously although only one of the other sources has any significant vapour flow.

To add detail the vessel is pressure is controlled by a pressure controller and the level by a level controller. The control valve in question is electrically actuated but controlled manually.

My understanding is you normally only size the PSV for control valve failure as the normal inputs are being dealt with by the vessel pressure/level controller. By including the normal inputs as well I am effectively saying I also have a vapour closed outlet case at the same time.

 

[TiCl4]

Justice,

To add detail the vessel is pressure is controlled by a pressure controller and the level by a level controller. The control valve in question is electrically actuated but controlled manually.

In the failure mode you mentioned - failure of an inlet control valve that controls vapor flow to your vessel - you did not mention if the pressure control valve on the vessel is capable of handling the increased flow rate. If it is capable of handling the increased flow rate, then there is no overpressure scenario. If the vessel PCV at 100% cannot handle the entire incoming vapor flow rate, then the PSV should be sized to handle the incoming failed stream PLUS maximum normal inlet flow from other sources, MINUS the flow the vessel PCV is able to handle. This would be for a single failure event with the vessel PCV functioning normally (and with operators responding properly to the rise in pressure, as you said it is manually controlled).

Since this vessel PCV is manually controlled, you need to determine the operator response time. In the event above, how many seconds are there between the "High" pressure alarm and the time it would reach MAWP? How many seconds between "High-High" and MAWP? Does the operator have sufficient time to notice the alarm, view and judge the process conditions, and take appropriate action with the vessel PCV? If the scenario results in a 10-second overpressure case, there is very likely insufficient process safety time and you need to consider the vessel PCV as failed-in-place.

If rupture of this vessel is a single/multiple fatality event, do not rule out multiple failures. The initiating failure may be a BPCS failure on the inlet stream control valve, with an enabling failure of mis-operation of the vessel control valve to 0% open(you said the position was set manually, and is subject to operator error). This would direct ALL incoming vapor flow to the PSV. The initiating probability is lower due to the enabling event, but I bet you would still need a PSV sized for full incoming flow to get to the 1 in 100,000 mitigation level.

 

[danschwind]

Dunno the opinion of the other colleagues, but IMO if all the valves on this vessel (inflow and outflow) are actuated by the same utility (electricity or compressed air), in the event of utility failure you would consider the worst case scenario (inflow valves wide open, outflow valves closed), which would probably be the design case for this PSV (excluding the possibility of fire).

If I recall correctly, if a valve has a fail-in-position failure mode, you would consider it either wide open or totally closed to size the PSV, whichever permutation is the worst. I'm away from my API 521 right now to check if my mind is playing tricks on me.

Daniel
Rio de Janeiro - Brazil

 

[Justice100]

TiCl4 - The upstream control valve is manual electrically actuated. The downstream vessel (which has the PSVs I am assessing) has air actuated automatic pressure control and level control. The upstream control valve could easily exceed what the vessel pressure control valve could handle (it is a new tie-in to an existing vessel). I am placing a mechanical stop on the upstream control valve to limit the opening.

 

[TiCl4]

Justice100,

I misunderstood which actuator had what type of control, sorry. As long as your upstream equipment is fine with having a reduced bleed capability, then adding the mechanical stop seems to eliminate the over-pressure case in your opening post, no?

If it is a new (not yet installed?) tie-in, any reason to not go with a smaller valve instead of a mechanical stop on a larger valve? I'm not against mechanical stops (I've seen them used with good efficacy), but I've seen them used to prevent full valve closure on a control valve to prevent deadhead. I've not seen them used the other way around.

 

[georgeverghese]

If the other major feed source has no means of auto closing on detection of high pressure at this downstream vessel, and also has no PSHH to act as penultimate protection, then its normal max flow (reduced appropriately to account for higher pressure downstream ) should be included in the feed rate at the PSV, assuming its max normal source pressure is at least equal to relief pressure at this PSV.

 

[Justice100]

Thanks for your comments TiCl4 and george.

TiCl4 - The upstream valve is existing (we are providing a new cross connection to this vessel) otherwise yes I would have tried to pick a valve that had the appropriate maximum open Cv.

Looking at it from the blocked discharge perspective - Blocked discharge case is considered based on maximum normal operating flow into the vessel. I know the maximum normal inflow from the other sources. The question is what is the maximum normal inflow from the newly tied in manual valve. Since it is manually controlled it could potentially be left open up to any point up to the point at which it overwhelms the vessel PCV. On that basis the PSV should accommodate maximum normal inflow from the other sources and the manual valve up to its mechanical stop (which I will set based on the PSV capacity).

The manual valve should not normally be passing anywhere near enough gas to overwhelm the vessel PCV if being operated as intended (operating procedures) so I suppose you could argue this is double jeopardy when combined with closed outlet but at the same time it is possible to operate at this point if the operator wanted to so is credible but conservative.