Pressure Control with Remote Sensing 1

[HEC]

Currently trying to get my head around the pressure control dynamics of the following situation.
The fluid is an acidic liquor (8% H2SO4, and dissolved minerals). Nominal flow rate of 2,100m3/hour in a PE lined steel pipe 590mmID. approximately 650m long, minor change in static head normally to a Storage Pond.
Flow is controlled by variable speed of the Pond supply pump.
There is a temporary by pass situation around the storage pond, which means the Pond supply pump/pipe described above, is piped into the suction of the Process pumps from the storage pond.
To ensure the down stream process is not upset we intend to use the suction pressure at the Process pumps to control the Pond supply pump speed.

The question is what lag (if any) will there be between pond supply pump speed and the pressure change detected 650m away at the Process pump suction. Is it related to the speed of sound in the fluid or the fluid line velocity?

Mark Hutton

 

[BigInch]

The speed of sound in the process fluid will do. Its a relative velocity, but the fluid speed should be a maximum of around 0.002 * sound. I'd ignore it.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[HEC]

Thanks for the response Biginch, I'm now curious, at what order of pipe length do you start to get significant lag between pressure and pump speed. Say we were looking at a situation where we had a control valve on the outlet to provide a constant upstream pressure for a variable flow.


Mark Hutton

 

[BigInch]

Can't fool me. That's a different question than the first one. Now you're adding a control valve at the outlet, but I'm not sure what outlet you mean. Where exactly are you putting that upstream pressure CV, in front of the process pumps? Are you removing the pressure control to VSD and substituting a flow meter with a flow transmitter signal to control that VSD now?

[Supply Pump]=(FT)===================]V[=[process pump]===>

If this is the case, and the valve senses a local pressure change and adjusts its % open, it would take just under a second for the pressure wave to travel back to the supply pump. The flow change is registered at the meter/transmitter and sent to the VSD, which signals the pump to adjust speed. That part is pretty fast. Now all you have to worry about is how long it takes the pump to spin up/down. In the case of speeding up the pump, that's a function of the total available power (motor rating) minus the power actually being used at that exact time the signal is received. The total available - used is the power available for acceleration of the pump and liquid in the pipeline at the time. The bigger the difference, the faster the pump will reach its new speed and start increasing pressure and filling the pipeline faster. In the case of spinning down, the slowdown is accomplished by reducing power and the resultant reduction of the pump's rotating inertia by the fluid in the pipeline wanting to backflow into the pump as local discharge pressure decreases, but is still a bit higher in the local pipe area. With the typical slight oversizing of electric centrifugal pump motors at about 15 to 20 percent of what's needed at normal operating flowrates, and normal pump rotating inertias, it doesn't take very long for the pump to adjust, but then it does takes a bit longer for the fluid in the pipeline to slowdown or speed up, figure 2 times the travel time of sound for one length of pipeline to catch the bulk of the change. So, if its a normaly sized electric pump.. I'd say about 2 seconds should be pretty close for your line. If its diesel driven, that's a whole 'nother story.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[hacksaw]

I understand your question well and good, but for control you have to limit the rate of flow change i.e. speed. Your motor controller has settings for this. This time lag is typically much larger than the propagation delay by several orders, and is more a function of the transport delay.

for what 'tis worth...

 

[HEC]

Thanks again BigInch, you are correct it is a different question, however your explaination is closer to the information I was after. The real situation is a bit more complex, along the lines of the attached. It is a mess but a temporary mess. The second set of pumps would normally draw from the Storage Pond. All pumps are electric motor driven and VSD's. Your 2 seconds seems reasonable for a system lag. The "supply" pressure control loop will be tuned to run slower than the second control loops.

Hacksaw can you please expand further how the transport delay would be quantified.

Part of what is driving this quest for knowledge is the proposed pressure ttransmitter at the pump suction is quite remote. A Fisher Rosemount/Emerson wireless unit is proposed for this cost of wiring etc. The crunch is this unit will only update at 15 second interval! If pipe system dynamic is longer than this we can live with the long update time if not...the strings will need to be put on.

Thanks for the info so far.

Mark Hutton

 

[BigInch]

Mark, there's lots of ways to skin a cat, and I don't know the pump, valve and system curves, but I would tend to keep control variables localized and move them to the suction headers rather than introduce time lags with remote transmitters.

The way you have it now, assuming the first PT is maintaining a minimum suction pressure at pump group 2, there may be a couple of control limits you need to look at.

Suction pressures at pump group 1 could go below minimum suction pressure (NPSHR) as P1 goes moves to a higher RPM to supply increased pressure to P2

Assuming the group 2 PTs are maintaining minimum pressures on the process, they will increase P2 speed and reduce suction pressure at P2. If P1 can't keep up with its setting as before, once again P1 and now P2 will go below min NPSHR.

If the FCV setting is too high for any given flow the pumps and upstream portion of the system can supply at any time, according to their current RPMs, the system pressures will reduce until possibly minimum process pressure is lost. So a maximum FCV flow setting should be determined such that minimum process pressure is always maintained. That may require that you find a max FCV setting that is permissible for each combination of RPMs at group 1 and 2 pumps.

If PTs are set to maintain minimum pressures, maximum pipe pressures might be exceeded, if discharge pressures at pump's maximum rpm can do higher than the downstream allowable pressures. You may need to set max RPM limits.

Now I'll offer a suggestion as to how I would do this. The same control limits are needed, but everything is local and interaction between controls that might cause unstable looping is minimized.

Make a chart of pump rpms and flowrate delivered to the process. It may be a bit complicated, since the pumps are in series as you may need to establish system flows vs all combinations of RPM at pump group 1 and RPM at pump group 2, but I think its the best way to do it.

Now you can set your VSDs to control pump RPMs based on the flowrate you want to deliver to the process.

If you have a well designed system where pump flowrate, rpm, discharge pressures and system flow curves all correspond to your process requirements, you might be able to stop here and remove all the controls except for the pump RPM-Flow setting. But maybe you don't.

If not,

Then include min suction pressure overrides. With both groups of pumps on a local minimum suction pressure control, a PT in front of each overriding the flowrate set via the VSD, acting to slow down the pump rpm if suction pressure approaches a minimum suction pressure (the higher of either a minimum process pressure plus pipe losses, or the pump's NPSHR). That should cause suction pressures and all downstream system pressure levels to rise, as the slowing pumps reduce flow. With all pressures on the rise, but flow slowing, the FCV will try to maintain the set flowrate. If the set flowrate is too high for the pumps to maintain as they slow down, the FCV will start opening and all pressures will begin to drop.

So the trick to tuning and operating this system is finding the maximum flow setting you can have at the FCV while the pumps are at minimum RPM and the minimum flow you can have at the FCV while the pumps are at maximum RPM, so pressures are maintained below maximum pipe allowables, because if the FCV is set too low, the pumps will move to maximum rpm and highest discharge pressures. Limit the maximum pump RPMs so max disch pressures don't go over pipe pressures.

OK, I don't know eveything about this system, so you do it how you want.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[HEC]

Thanks a star for you. I have done part of what you have recommended already in relation to looking at the first pump set (there are up to two pumps in parallel, exciting stuff) curve and relating that to flow/speed. The reason for placing the pressure transmitter at the Pump 2 set is indeed to maintain minimum suctin pressure at these pumps. I believe that what you propose is the best solution. We will reconfigure the control such that there is a flow control of the first set of pumps with a low suction pressure "bias" input to the control loop.
Thanks again

Mark Hutton