Simplified description of the system:
We have a VFD driven raw water pump puping water into a filtration unit. There is a flowmeter and modulating valve at the intake of the filtration unit.

Current control strategy in the specifications is such that the operator enters a flow setpoint and the PLC adjusts the speed of the VFD to meet the flow setpoint into the filter. However, the PLC also modulates the valve at the filter intake to maintain the same entered flow setpoint.

So now we have one flow setpoint that controls the VFD and also modulates the intake valve.

This is not practical in the real world as the VFD and valve will fight each to maintain the setpoint.

But my question is will this current setup work?

We want to avoid making changes in the code as the project has turned political and we want to make as less change in the control narrative as possible.

What are the possible "quick and easy" changes that we could implement in the logic to make this work?

1. Ones that I can think of is make the valve full open/close and allow the VFD to control flow.

OR
2. introduce some large deadband in the program. (don't know how to do it and some direction will be useful).

Looking forward to your help.

Thanks

 

One approach which I have seen used with boiler feedwater pumping systems is to use the valve to control the flow in the short term, then adjust the pump speed to maintain the valve at a large opening to minimise pressure drop. If the variable speed control reacts quickly enough, then simply run the valve at 100% - after all, the justification for the VSD is to maintain a high system efficiency.

 

Well, you need to pick what you want to control. the reality is that the VFD is really controlling the pump pressure. As long as that is working for you I would be inclined to use it and just run the valve open enough that it does not restrict flow on the pump.

--
Bob
http://ilbob.blogspot.com/

 

The basic rule is ... One measurement = one PID loop

But of course you have two actuators - each with a different transfer function (i.e. gain and time constant). There are a couple of standard ways of dealing with this ....

1. Split range control. (probably not appropriate here)

2. Midranging control

3. Use 2 PID loops but with different setpoints so that the second loop only comes into play when the first loop is saturated and cannot maintain its setpoint. (works in theory, but in practice is often a mess. Operators don't understand it and the loops end up fighting because setpoints and tuning are wrong)

4. Lock one actuator in place and use the other.

In your case you say you have a VFD and a Valve controlling the same flow in a pipe. (forget the filter unit, its basically a pipe)

The simplest solution in your case is probably to fix the speed of the pump and control only the valve. You can probably let the operators do this for you - which means you have a perfect solution to your "political" problem. Once you have shown everyone the system works much better this way, then you can probably go ahead and implement the better fix ...

Mid ranging control: basically - run the pump at a fixed speed and control the flow with the valve. If the valve exceeds 80% (or so) open, then speed up the pump slightly. If the valve is below 20% (or so) then slow down the pump slightly. Be carefully with making changes in the pump speed as this causes a bump in flow. Use small steps, make the changes slowly, and be patient. Wait plenty of time for the flow control to catch up and settle on a new valve position.

Finally, if you absolutely must use 2 PID loops - tune them properly for RRT (Relative Response Time). As a rough guide one loop must be 5 to 10 times (yes ten) slower than the other. This way, the faster loop can keep up with the disturbances introduced by the slow loop. So if your closed loop response time for the flow control with valve is say 3 seconds, the closed loop response time for the second loop should be 30 seconds.

http://www.expertune.com/PIDApp.aspx

Rob
www[.]lymac.co.nz

 

If you want to do the job right, add a differential pressure transmitter across the control valve and use that as process variable for the VFD PID controller, so you control the speed of the pump to maintain a fixed differential pressure across the control valve.

 

Power in a pressure system = volumetric flow x pressure drop. If the system has a high flow, then minimising the pressure drop across the valve will help you save money - which may be a carrot to dangle in front of the political animals.

 

A key principle of control is: 1 controlled variable (CV) – 1 manipulated variable (MV).

Thus: Flow rate is your CV, the VFD is your MV. Add some logic to ensure that the valve is fully open whenever the drive is operating – thus minimizing the energy loss over the valve – and (if necessary) closed when the pump speed is zero.

The old problem with fixed speed pumps and valves: It is comparable to driving a car always with full engine power and controlling the speed with the brake. The VFD enables us to do things right.

Throttling the valve as well will kill the fundamental motivation for using VFD’s.

 

I would suggest split range control.

Output of the PID controller:
0 - 10 % Open valve from 0-100%
10-100% VFD speed control from 10-100%.

(For more successful control you might want to calculate the speed of the VFD for which the 10% flow is provided and set this speed as minimum)

Assuming you are using a centrifugal pump.

This will work, and will only require very small changes to your control narrative.

Best Regards,
Heinrich Baumann