Hello,

I'm trying to control the suction pressure of a reciprocating compressor with a flow of 12 MMSCFD, 160 psig, the problem is that the pressure upstream the valve change in morning there are 300 psig,and 600 psig in the afternoon.

How can i tuning this process? How can i estimate the proportional band?
Thanks

 

Hi there,

You are not really telling us what your problem is. You say the upstream pressure changes, but so what? On any pressure control application the upstream pressure will vary. That is what we do, to control and stabilize this variance of upstream pressure with the pressure control loop. Have you already tried various PID settings and still cannot stabilize the system or do you want someone to give you a estimate PID setting to start with?

If the later, the easiest is to go to one of your other pressure control PID controllers and use those settings and see how it goes. Based on the response you can then make small adjustments up or down until you get to the optimum setting. The change in upstream pressure should not be to much of a problem since the PID controller will compensate for that by opening or closing the control valve accordingly.

You just need to tell it how fast to compensate to make sure the control loop doesn't go into a swing or cause a bump in the process the moment the upstream pressure changes.

So start with a average PID setting and increase the response time until the system goes into a swing or becomes unstable, and then reduce it a little bit. Make a change and leave it for 24 hours, draw some trends and then check the response during low and high upstream pressure changes. This will give you a clear picture on what to do next to improve the system's response time.

 

1. Are you trying to control the discharge pressure to 160psi, and you are overshooting to 300-600psi?

2. You are stating that you are observing a change in pressure from morning to afternoon. Do you have a large ambient temperature difference between morning and afternoon? To this effect is the compressor air cooled or water cooled?

3. Where exactly is your pressure controller process variable signal coming from, before or after the compressor?

4. Are you controlling the discharge pressure by limiting the suction flow or by bypassing the discharge back to the suction?

 

Ok, yes you are right,

My principal problem is i need to control the pressure down stream of the valve to 160 psig. the overshoot maxim 10 psig, but when i put a proportional band (example 70%) and reset on 0,01 min/rep with the pressure up stream of the valve is 300 psig. when this pressure increase to 550 psig the valve begin to oscillate. if i increase the band to 140% stop the oscillate, but i have a big over shoot example 30 psig. i´m controlling the flow of the compressor with the suction pressure for regulating this.

What type of control or tuning i can to do for minimal overshoot, but no oscillations?

 

Well, it sounds like you have several control problems all rolled into one.

First, there is a tuning problem. You have a very fast integral action. Not sure what sort of controller you are using, but 0.01 min/repeat is less than 1 second/repeat. The controller sampling rate establishes the minimum possible deadtime (probably 0.25 to 1 second). Even so, it may take the valve a full second to move and for the process to respond, so your apparent process deadtime would be at least 1 second. I ran a short simulation on our ExperTune software, and even the most aggressive suggested tuning would put the integral action at 0.05 min/repeat. My guess is you may even need to go a bit longer than that. True deadtime, process gain, and suggested controller gain can only be determined from some response data.

Second, you mention that you get different response at different pressures. This indicates that there is some non-linearity here. You can expect most valves to behave differently in different ranges of operation. An output characterizer can help to ensure consistent control performance over a range of operation. For more info on how to handle this, visit:
http://www.expertune.com/r2.asp?f=AList&l=PIDmodel.html#characterizer

Third, making step-wise setpoint changes may actually be quite a shock to your process. Another way to minimize overshoot is to make use of Setpoint Ramping. With this approach, there is a ramp to the new setpoint over time. You can accomplish this in a DCS or PLC with a little bit of programming. It may be worthwhile to do this to avoid "hammering" the process when you make a setpoint change.

You may also have some valve issues, but I'd start with these three things first.

Good Luck!

-George