Sodium Hypochlorite PID Control Help

Hello Everyone,
I currently have a job with a customer that wants to operate an injection pump for Sodium Hypochlorite injection using a PID as follow.
"The control should depend on the finished water storage tank free chlorine concetration and the permeate flow rate. A setpoint for free chlorine concentration in the finished water storage tank should be made adjustable for the opetators, a PID control shall be used to adjust the pump speed so that the desired chlorine concentration is maintained"

From the controls description I'm thinking a cascade PID, but from what I been reading a PID is usually does not work because of the locations of the injection and the location of the analyzer.

I have not done a chemical dosing system does anyone have any samples logic or point me to the right direction.

[RO] --> [FLOW Transmitter] --> [Injection Pump] --> [Finished Storage Tank and CL2 Analyzer]

Thanks for the Help.

A lot depends on the physical configuration. For example, where is the chlorine injected? Is the storage tank mixed? What is the demand of the chlorine in the permeate - for example, is it nearly clean water or are there organics or compounds that will use the chlorine before a residual is created? How many chlorine sensors are available and where are they located. And so on ....

In a RO water system the chlorine demand is usually very low. Rather than use PID I'd suggest a feed forward control, with the chlorine flow rate based on the water flow rate. This is a straight algebraic calculation. Most potable water systems allow for a range of residual in the storage tank and distribution system, not a specific dead on value.

Many engineers specify PID as a reflex any time they see analog control. Sometimes that is the best way, sometimes not.

Tom Jenkins said:

A lot depends on the physical configuration. For example, where is the chlorine injected? Is the storage tank mixed? What is the demand of the chlorine in the permeate - for example, is it nearly clean water or are there organics or compounds that will use the chlorine before a residual is created? How many chlorine sensors are available and where are they located. And so on ....

In a RO water system the chlorine demand is usually very low. Rather than use PID I'd suggest a feed forward control, with the chlorine flow rate based on the water flow rate. This is a straight algebraic calculation. Most potable water systems allow for a range of residual in the storage tank and distribution system, not a specific dead on value.

Many engineers specify PID as a reflex any time they see analog control. Sometimes that is the best way, sometimes not.

Click to expand...

Thanks for the quick reply, this application if for a water treatment plant.
[Well], [Iron/Manganese Removal Filter], [Two Stage RO System], [Finished Water Storage Tank], [Distribution System].

The injection points are as follow.

[Well] --> [Filters] ---> [RO] --> [FLOW Transmitter] --> [Injection Points] --> [Finished Storage Tank with CL2 Analyzer]--->[Distribution System]

The permeate water will be injected with Sodium Hypochlorite and Sodium Hydroxide. The CL2 Analyzer is measuring the Free Chlorine inside the storage tank.

what I have seen done a lot in smaller plants is to have the sampling pump speed based off of incoming flow, like tom had mentioned. I think pid would not work very well at all, based on chlorine systems I have seen

In fact a system i just installed a new analyser in has the incoming raw water mag meter 4-20 signal go right to the injection pump, no plc in this system.

curious what brand of analyzer are you using and are you running an sample pump or just using storage head pressure.

In the configuration you outlined above, best scenario would be a long pipe-run after the injection point and leading to the storage tank. Then, the analyzer should be near the tank sampling the water in the pipe. If the analyzer samples the entire tank, this system will not work well (in which case you are better-off using the flow-pacing method already discussed).

I am a big believe in feed forward control BUT, it you know the feed forward gains then you have the inverse model for the plant or system. If you have model then you can apply a PID to correct errors between your plant and model. If you can do the math then you can control many things thought to be impossible to control.

Like you, I suspect you are going to have a lot of trouble with PID because the chlorine measurement is in the storage tank. That means a lot of lag time between a change in feed rate and a change in measurement. If the tank isn't well mixed it will exacerbate the problem.

I really think feed forward flow pacing is the way to go. I've done that successfully on potable water systems. The water will have almost no chlorine demand, so it is essentially a straight mass balance algebraic calculation. You add a specific weight (volume) of hypochlorite for each pound of water flowing into the tank. If the residual in the tank is off a little bit you simply tweak the ratio a little.

You may need to have a discussion with the engineer and get approval before you do the programming. The last thing you want is the engineer rejecting a working system because it doesn't meet the spec. It is also possible that he knew something we don't when he wrote the spec (although that is highly improbable).

I agree with everything Tom J. and others said.... just wanted to have you make sure there are controls on the injection pump (hi/low limits) so that a malfunction of the flowmeter or residual analyzer does not overchlorinate the water in the tank. If either the high or low limits are hit that should be set up to alarm the operators.

FLAengineer said:

I agree with everything Tom J. and others said.... just wanted to have you make sure there are controls on the injection pump (hi/low limits) so that a malfunction of the flowmeter or residual analyzer does not overchlorinate the water in the tank. If either the high or low limits are hit that should be set up to alarm the operators.

Click to expand...

The pumps don't have High and Low limits but we do have instrumentation failure alarms as well as high and low alarms. I will be onsite tomorrow to actually verify that the analyzer is reading the whole tank as per the P&ID.

Thanks for everyone's help.

Why not do both PID and Feedforward? The Feedforward would be a calculation based off of the fresh water flow. The PID integral would take care of long term errors.

The thread has come back from the dead.

g.mccormick said:

Why not do both PID and Feedforward? The Feedforward would be a calculation based off of the fresh water flow. The PID integral would take care of long term errors.

Click to expand...

Yes, both feed forwards and PIDs are used together.
I mentioned having a model above. Normally one is controlling by pH which is non-linear so non-linear gains are required if pH is going to change over a range. If the pH is supposed to be constant then one can use what the controlguru called the DLO or design level of operation and linearize around that set point.


The goal of feed forwards is to get close within 5% or better. The PID then has less work to do to close the last 5%.

Peter Nachtwey said:

Yes, both feed forwards and PIDs are used together.
I mentioned having a model above. Normally one is controlling by pH which is non-linear so non-linear gains are required if pH is going to change over a range. If the pH is supposed to be constant then one can use what the controlguru called the DLO or design level of operation and linearize around that set point.


The goal of feed forwards is to get close within 5% or better. The PID then has less work to do to close the last 5%.

Click to expand...

:site::site::site: