PID control for dosing pump?


Thread Starter

Ralph Reagan

Large water treatment plant(20Meglitres/day), has inlet flowmeter inputing 4-20mA to PLC, need to control output of PLC (4-20mA) to small chemical dosing pump(via VSD)with pump speed tracking inlet flow ("Flow Paced Dosing"). Is PID ok?

am writing functional spec for client approval which will also be used by my PLC/SCADA programmer (so it needs to actually work). need to control the speed of the new pump so it tracks the inlet flow of the plant using inlet flowmeter, and delivers the right amount of chemical back into the inlet flow (only a few l/h). am using allen bradley SLC505 networked to CITECT SCADA. not sure if i should use proprotional only,prop & integral or full PID (Derivative control not my strong point back in college)or directly link signal from flowmeter to the 4-20 input to the pump motor VSD (pump does have 0-100% adjustable mechanical proportional output)

the dosing rate (mg/L)will entered by an operator to SCADA, an algorithm in PLC will calculate a setpoint based on:
Dosing Rate - mg/L
Plant inlet flow - ML/Day
Chemical concentration - mg/L

have already worked out algorithm to determine the setpoint, i think that i should have some proportional control justin case pump needs some speedin up, but would integral or derivative help?

inlet flow to plant does not change very rapidly except for stormy weather when it can quickly increase by 50%.

Bill Clemons

Your control requirement for a dosing control, using 'flow pacing' can be effected by running "P-only" control. Since the dosing rate only adjusts to changes in plant throughput, the controller will continually update at a rate proportional to the flow signal.

Should you add a 'chemical concentration' input signal to the application, assuming this measurement is a residual of the injection rate, you could run a separate "I-only" control loop and sum the output from this loop and the flow pacing loop for adjustment to the dosing output to the vsd.

These applications, flow pacing, residual, and compound (combination), are fairly routine to water treatment operations, and also find applicability to pH control. For further reading see Moore Industries, 555 Chlorinator, with published examples and algorithmic considerations, like fixed and variable lag times, multiple gain settings, etc.

Robert Scott

Of the parameters that you mentioned:

> Dosing Rate - mg/L
> Plant inlet flow - ML/Day
> Chemical concentration - mg/L

only the chemical concentration will be subjected to PID control. The other parameters will influence the dosing pump in an open-loop manner.

I would suggest that you design your PID control so that the chemical concentration controls the output through a factor which gets multiplied by the measured plant inlet flow. The advantage of using plant inlet flow is that it reveals a need for a change in the output faster than the chemical concentration, which probably has a significant delay. You could almost run this as an open-loop control without PID, using manually trimmed factor on the plant inlet flow.

In answer to your question about integral or derivative control, derivative control could help, particularly if there is a lot of lag time between a change in the dosing output and the measurement of the resulting effect in the chemical concentration. And integral control will always help. Just make sure that the output of your PID calculation is applied to the factor that relates plant inlet flow and dosing output, not to the dosing output itself. You want dosing output to respond most directly to plant inlet flow.

Robert Scott
Real-Time Specialties
Embedded Systems Consulting
Feedfoward dosing based on flow with PI trim, if the variable is available, will provide and good solid control sceheme as long as injection and sampling are in the proper places. Derivitive would not serve here as the lag time in this loop is long.

Ralph Reagan

>can be effected by running "P-only" control.
Thankyou thats what i was hoping

> assuming this measurement is a residual of the injection rate, you could run a separate "I-only" control loop and sum the output from this loop...........

I should have explained more in my original query.
Concentration shall be entered to SCADA by an operator as the strength of the chemical used to treat the effluent.(based on info from delivery documents etc). If a batch of chemical is delivered to site 1/2 strength for example, then algorithm shall double the amount of flow from dosing pump. algorthm as follows

Req.Flow(PID setpoint l/h)=Dose Rate(mg/l)X Plant Flow(Ml/d) X 1000000(l/Ml) divided by:
Concentration of chemical(mg/l) X 1000(mg/l) X 24(h/d)
The chemical flow shall be closed loop control, (feedback from chemical flowmeter imediately after dosing pump), but actual treatment of the effluent shall be open loop as there will be no feed back from treated effluent.
Operator will choose a dose rate based on local knowledge of plant and various display only analysers downstream.(sounds dangerous huh?)this is how the client does it at other existing plants. this method should work fine up until stormy weather dilutes incomming effluent.

> For further reading see Moore Industries, 555 Chlorinator,

Thanks bill will check website