Glycol VFD issues

[Npanch]

We have just installed a VFD connected to an inline pressure sensor on our glycol processing pump. The sensor is inline on the supply header going to the tanks. The problem is that we have set the pressure on the VFD to stay at 15PSI, but when we turn the system on the pump cycles up and down continuously. You can hear the pump rev up and down about every 3 seconds. It stays in the 15 psi range but my understanding is that the pump would hold a steady flow at 15psi and power up or down to compensate when takes open and close. Is this normal?

[dick murton]

Sounds as though you don't have a deadband on the control loop set up, so that the system has a chance to settle down and operate between upper an lower limits instead of around a single setpoint. You may also want to start the pump ramp up from a slow speed and even give it a forced controlled ramp up until it hits the "minimum" limit of the deadband - or, simply slow the control loop down a bit - I should know, but not sure if this is P, or I, or D - but a proper suitable engineer / programmer will know.

[gitchegumee]

Totally normal.

But not what you want. PID controllers are very tricky to set properly. Especially when you have "step functions". Imagine your glycol header immediately spiking pressure because you just had a solenoid slam shut. So a normally very slow-response system will suddenly decrease in pressure when the solenoid opens. Hard to control instant changes to an otherwise slow-response system. Normally you would NOT use all P + I + D in a control algorithm; but instead P + I more often. Long story short, you should try to slow down the pressure spikes. Maybe by opening a bypass loop a bit, or by using slower-moving quarter-turn ball valve. Then you should find someone who would trade a bit of programming expertise for a bit of beer. It's difficult to muddle through programming yourself if you don't have any experience doing it. Another great option to controlling pressure is to add a bypass solenoid that opens only when all other solenoids are closed. Helps with those pressure spikes. Good luck!

[rdcpro]

Short answer:

If you're using a PID control for this, then it's out of tune. Configure the drive to run in proportional control only. So if setpoint is 30psi, but 26psi is acceptable, configure the proportional band to go from 30-26psi from low flow to high (high flow is at lower pressure). How you do this is dependent on the controller. Sometimes it's a percent, other times you set the proportional band, with minimums or maximums.

Long answer:

The "P" is for Proportional control. The idea is you set your output to some percent of the measured variable. It can be positive or negative, depending on what you're controlling. In your case, proportional control means the pump increases in speed as the pressure goes down, but only to a limited extent. It doesn't try to get it back to setpoint. If the proportional band is 4psi, the pump will be at minimum speed at setpoint (which is essentially zero flow), and max speed when the pressure is 4psi below that. So at 50% of the PB, your pump output is 50%. If pressure starts to go up, the pump output goes down.

The "D" (or derivative) part of PID is there to prevent overshoot of the setpoint. If the output is changing quickly, it's supposed to dial things back so you don't overshoot. It's based on the rate of change (in calculus, it's the differential). So the faster your output is going up, the more "D" will turn it down. And the closer to the setpoint you are, the more the derivative tunes the output back. Derivative control generally works against the output.

The "I" (or integral) part is what moves the output up to meet setpoint, when it wants to stay low. Without the "I" component, the pressure will never reach setpoint. There is also something related called "anti-reset windup" that is supposed to keep overshoot under control. This happens when your "I" says you're too low, and it keeps increasing the output, but you're still too low, so it increases output even more. Then it finally shoots up too high. Anti-reset windup dials things back if it spends too much time below setpoint, so the output doesn't go too high. You use it in systems that don't respond quickly, like heating a volume of liquid to a specific temperature, when the system is not "stable," meaning other things are contributing, like changing temperature of incoming water.

Tuning PID controls are kind of an art, but you do this intuitively all the time when you accelerate to freeway speed from an onramp. As you get closer to the speed limit, you ease up on the throttle (derivative). When you go up a hill, you have to push the pedal a bit farther than if you were on the flat (integral).

One of the first things I look at is whether or not PID control is called for in the first place. If the system has a lot of upsets like glycol solenoids opening and closing, its going to be harder to tune a PID controller to keep things at setpoint, if it's even possible to do so. So, the question is, do you actually need to keep it at setpoint, or is "close" good enough?

Example 1: PID control of mash temp through various steps. Assuming you wanted to do this, it's important to ramp at a particular rate, and maintain a specific target temperature when you get there, without overshooting. So, PID control would be used to prevent overshoot, and keep temp to setpoint with no deadband.

Example 2: Glycol flow in a loop. Assuming the differential pressure across the FVs isn't critical, you don't need to control to a specific flow rate or pressure. It would be OK if the pressure at full flow was, say, 10% lower than at low flow. In this case, you only use the "P" in PID, which stands for Proportional. Set your proportional band for the range you want to be in.

Regards,
Mike Sharp

[Npanch]

Thanks guys