Parallel valves and compressed air pressure control

[carlosfandango]

All,

I've an application coming up where we're doing some R&D testing on air compressors, this requires a pressure map to be completed at various speeds and back pressures. Back pressure on the compressor is normally controlled via a valve downstream of the compressor outlet and subsequently controlled by a PID controller of types.

The problem is as the compressor is reasonable high flow we have a 8" pipe on the outlet but require very accurate pressure control down to 10-20 mbar level, the compressor on test will be capable of 2 bar maximum.

I'm not sure with using such a large valve on the outlet we can get this type of control accuracy on the outlet of the compressor. What has been discussed is using two valves in parallel to control the pressure ?. One larger 8" valve to manually set at a fixed point roughly around the pressure required and a much smaller valve in parallel in a PID loop to automatically trim the pressure into the desired range. Haven't managed to find an application like this online either because nobody has really tried it or it just doesn't work !!!

I've had a few recommendations on single large valve but from experience we just don't seem to be able to get the accuracy on the pressure control. Our application has a large tank on the outlet of the compressor but before the valve to damp out any pressure oscillations.

Anybody any experience or knowledge of a parallel valve application to control pressure along with knowledge of the control side of it ?, any pointers or help gratefully received.

Thanks

 

[LittleInch]

Don't understand why you're not trying two large valves in series, either as a fixed orifice type or fixed percent opening valve or slow acting primary valve followed by a valve with a lower Cv to act as the "trimming" valve.

Hence say you have 2bar outlet then primary valve takes say 85% of the pressure drop (1.7bar) and then your trimming valve takes 15% (0.3 bar)?

Trying to do it with a parallel line will be difficult and ultimately you have the same thing.

The discrepancy in flow between a large and small diameter is enormous and hence from closed to full open it will be difficult to get as much control as you want.

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

 

[gerhardl]

Hello carlosfandango!

As known all accuracy on all pneumatic based systems is depending on the total balance of air in and out, reservoar sizes and control of openings for supply and outlets.

The problems with testing systems is that they often have a wide testing range, and it could be difficult to find a single valve covering all situation. I think you have had this in mind asking the question.

When deciding on total system solution you will have to chart all possible testing situations. Maybe it would be necessary to split in several system, or have available several single control valves in parallell, using the most suitable for the range you are tseting in.

 

[carlosfandango]

Thanks for the replies so far. The theory of controlling the two valves in parallel would be.

Smaller valve P.V comes from a pressure transducer mounted on the tank (in between compressor and valve/s), this is the pressure we wish to maintain.

Larger valve P.V comes from the position of the smaller valve or if not available the smaller valves i/p signal from the controller. The setpoint of the larger valve is always set on 50% on the controller so it always attempts to keep the smaller valve operating around the 50% area.

The larger valves response should hopefully be set reasonably slow to prevent large pressure oscillations and the smaller valve set reasonably fast. This will obviously need some time to get working, hopefully not too long

I've a few back up plans with this set up just in case it doesn't work at all !!!

Haven't even thought about any hysteresis settings yet during operation ?

LittleInch - My issue with two valves in series is the cost of an additional large valve is significantly more than just buying a smaller valve and putting it in parallel with our existing 8" one by several thousand uk GBP. The parallel idea just seems cheaper. I understand your theory though, thank you for your input.

gerhardl - you are correct with the wide rage of testing it is very difficult to find a single valve that will cover all the pressure and flow points we wish to test at, high pressure/low flow and low pressure/high flow I am sceptical that even a valve with a large turn down ratio will work to the pressure tolerances we would like. Having two valves whether in parallel or series just seems to be a different approach and an interesting solution (if not a little more complex) to our problem.

Any feedback or experience of something similar please let me know.

Thanks

 

[LittleInch]

Depending on the time you need to vary the flow you could do this with a set of orifice plates instead of the first valve then modulate using a valve with a smaller pressure drop range?

downside is you might need to keep stopping to replace the orifice plate, but it will be cheap.

I just can't see the smaller valve doing what you want to do unless your main valve is taking 80% of the flow or more.

Or set up a manifold with a load of 1" or 2" valves, maybe with orifices in each and then just open one at a time until you get close to your desired pressure / flow and then modulate on that with a 2" control valve?



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

 

[carlosfandango]

Ultimately I'd like an automated pressure control loop I can just leave, type in a pressure setpoint and then run a slow speed ramp on the compressor, ie 0-3000 rpm over say 45 secs and control the back pressure all the way up (once sufficient pressure has been built up), this is my ideal scenario. If not it may be a case of logging at steady state speeds and waiting the pressure control to eventually settle itself.....hopefully !!!

 

[georgeverghese]

Try doing this with 2 flow controllers instead of pressure control. The bigger FCV to get you to roughly say 90% of required flow, and the smaller one to give you trim control. When flows have settled down, record the discharge pressure at reasonable accuracy at the sum of the flows. Allow instrumentation for pressure / temperature compensation of flow readout.

Obviously, pressure readings will be more accurate when the calibrated span of the PT is as narrow as possible. Also, suction pressure readings will also need to be accurate, as will suction temp, and any other parameters that affect throughput ( humidity, polytropic eff for example).

You havent said much about the type of compressor you've got. If this is an oil flooded screw, then compressor flow may be affected somewhat by the heat soaked up by the oil for example, which in turn, is affected by air cooler performance??