Flow Control Valves - Parallel Operation

[99loki99]

I'm looking at an oil export system that has two valves in parallel. Only one at a time should be controlling the flowrate in a cascade level onto flow control.

Due to piping anomalies, I'm being asked whether the two valves can be operated with one in manual (@35% open) and the other in auto. I think this won't work because as the one valve closes, the flow will increase through the other one. (Which is now behaving like a fixed orifice)

However, I can't find any formulae to describe the interaction so I'm struggling to convince anyone. Upstream of the two valves are a couple of BIG (2x7MW) centrifugal pumps so I don't want things to go wrong...

 

[steamdog]

The fixed open valve will only change flow if the inlet or outlet pressure changes when the other valve is closed or open. If it does and you can calculate the flow through the valve at this maximum condition, and this is not more than the required minimum flow, then it should be OK. When the flow through the fixed valve is reduced, the other control valve should pick up the slack.

However, since these are both control valves, why not change the control technique to allow both valves to open (like split range?)

 

[JLSeagull]

I see no problem as long as one valve is in automatic. More interaction would occur if both were in automatic. If you already have two loop controllers, consider setting one in automatic and slowly open the second valve in manual and observe the affect for each 5% (or smaller) output change on the manual valve.

 

[JimCasey]

The change in the pressure at the "fixed" valve would be caused by backing up the flowrate on the pump curve and also by lower line/system losses at lower flowrates.

Traditionally for an application like this the small valve is sized to take 1/3 of the system capacity, and the larger valve is twice as big (1/3:2/3 split)
COntrol the valves with split range* and the large valve will transition to control the flow once the smaller valve exceeds its capacity.

* 4 milliamps: Both valves closed. 12 ma, smallest valve fully open, larger valve just begins to crack off the seat. 20 milliamps: both valves fully open.

 

[roydm]

If you have two different size valves as JimCasey suggests, another method is to have the smaller valve as a flow control loop and the larger valve as a position control loop.
Assuming you have one flow meter common to both valves.
The position control loop uses the small valves position (PID output) as it's CV and holds the small valve within a good control range by Open/Closing the large valve. This loop is set to react slower than the flow control loop.
To reiterate.
As the small flow control valve opens the large valve opens to bring the small valve back.

 

[99loki99]

Thanks to all for your responses - very much appreciated.

roydm - I have made this recommendation already. This kind of control is also known as 'gap' control and uses two valves; one with a Cv << than the other - maybe a fifth of the size. We have a massive valve (Cv = 1140) and a huge valve (Cv = 870) in place so a new trim would be required (Cv = 200?). This means spending money and shutting down production... You can imagine how well that went down!

JLseagull - I didn't mention that there is also a maximum overall flowrate limit. Leaving one valve in manual exposes it to downstream pressure changes. If someone else down the pipe shuts down, the pressure drops at my end. Flow is a function of dP at a particular Cv so the flow through the massive valve goes up. The FT is common to both valves and measures total flow so it shuts in the huge valve running the pumps up their curves (as noted by JimCasey) which puts even more through the massive valve.

Steamdog's idea is great, but I also neglected to include the fact that they run with one valve at max 35% open because of pipework vibration. D'oh!

Honestly, I really don't get paid enough for this crap :-(

If we use the analogy of two resitors in parallel, what is the equivalent formula for valves?

Thanks again folks.

 

[steamdog]

Here in NYC steam market we use a switching technique with two valves in parallel. One valve is sized at 100% of load and the other at about 25%. First the small valve opens. When this valve is close to its maximum capacity, it closes and the large valve takes over control. When the large valve is about 20% open or less, we switch back.
We do this so the big valve is never controlling close to its seat, giving good control and long lasting service.

 

[JLSeagull]

Configure an override selector on the DCS output using another controller set at the maximum flow rate.

 

[gerhardl]

Maybe I am missing some vital information here, reading the original question and the responses.

In my mind:

A) The answer to the original question is yes, under certain conditions.

B) The answer to additional questions are: it depends on the flow conditions.

As any flow and pressure regulating valve has to be dimensioned on the actual flow and pressure conditions before and after the valve, and vice versa: the actual performance can only be calculated when flow, Cvs and restrictions and conditions and losses upstream and downstream are known, and you in addition have a parallell/split range regulation, there is simply not enough data available to give a sensible answer.

If you gave all possible variations of upstream input flow from pumps (including worst case and performance on PRVs), and all details on valves and downstream pipe dimensions and restriction/losses you could perhaps come a step nearer, but only if you are for safety reasons sure you have taken into consideration all possible scenarios.

... and yes, you are probably not paid enough ;-)

 

[99loki99]

Cheers. There are loads of good ideas here.

I was trawling through 'Liptak' (Chapter 6 of process control and optimization) and found examples of 'Gap' or 'Floating' control, 'Small/Large' control and split ranging techniques with hints from Shinskey about how to make them work. Trouble is, they all require a re-trim to work which, as I mentioned, is not a popular solution. A quote for a Cv 200 valve re-trim is 58,000GBP BEFORE the production hit is taken into consideration (1 - 2 days @ 230,000 BOPD!).

I've done a 'toy' valve simulator using excel which shows that, as the valve controlling the flow closes - the pump moves up its curve slightly which increases the DP which increases the flow through the fixed valve. The flow controller reacts by closing in the controlled valve even more and so on. It does reach a steady state eventually, but there is the potential for big flow rate oscillations which would probably cause a production shutdown. Not good. Not good at all...

Studies using Unisim (HiSys) back this theory up too. I only have Matlab left to try.

I guess, at the end of the day, it's their valves and they're entilited to run them any way they like. My A**e is covered. And hey, if it works, I'll post it here for reference!

ATB.

Loki

 

[BigInch]

That model is showing a good similation. That's exactly what will happen. The flow variations will depend on the gain and derivative time settings of the controller. Modeling the time accurately would have to consider the rate at which the system curve changes and possbily how fast the motor and pump respond to that system curve change.

**********************
"Pumping accounts for 20% of the world’s energy used by electric motors and 25-50% of the total electrical energy usage in certain industrial facilities."-DOE statistic (Note: Make that 99% for pipeline companies)

http://virtualpipeline.spaces.live.com/