Controller PID Tuning

It sounds like you have something similar to a few of our air heaters. Your description is a little too little though. How, exactly are they connected? Can you provide a diagram?

On a hunch, I think you're using a cascade control configuration where the output of the temperature controller provides the setpoint for the Fuel Gas Controller (which is probably a flow controller). I don't know how it connects to your Combustion Air Controller (which is probably an oxygen trim controller).

For starters, I would tune the Fuel Gas Controller (FGC) first. You do this by setting your Temperature Controller to manual (I'm still assuming you're using cascade control) and use it to set the Setpoint for your FGC. Once your FGC is tuned, you tune the Temperature Controller. When you're done with these two, you can tune the Combustion Air Controller.

I am working on a Gas Fired heater. Temperature controller output is fed directly to FGC and Combustion Air Controller. Hence the FGC and Combustion Air Controller receive exactly the same setpoint. Ratio Control of the Fuel and Air are fixed by ranges of the the respective flowmeters Gas 0-400m3/h and Air 0-1200m3/h. The setpoint to the controllers from the temperature controller is interpreted as %. Take for example setpoint 100% Gas flow will be 400 and Air 1200 which gives a ratio of 1:4 which is fixed.

Okay, things are a bit clearer in one sense and worse in another.

First, there seems to be something wrong with your ratio. If your fuel flow is 400 and your ratio is 1:4, your air flow should be 1600m³/h. I'm not saying that your required fuel ratio is wrong, just that 400:1200 is not the same as 1:4.

Now, as for tuning, I'm not sure if you can do this but it might be advisable to tune your fuel flow and air flow without firing the heater. What I mean is that you recirculate your fuel so that you can tune the fuel flow control without having to run it through the heater. As for the combustion air flow controller, you can probably just vent it to atmosphere through an orifice. I prefer to do it this way since tuning with the burner running is tricky.

Flow control is probably the easiest to tune from my experience. I usually set the gain to a low value, say 0.5 to 0.8. I then set the integral to a high value, say 100 repeats per minute (or 0.01 minutes per repeat if you use the inverse). The low gain means that the control valve is not reacting too much from the natural noise inherent in flow measurement. The fast integral ensures that the controller responds quickly to setpoint changes.

Once you have your fuel and air flow controls tuned, you can fire the burner and tune the temperature control. This is more difficult and, if you have a big heater, waiting for a cycle to finish (so that you can determine if you need to change any PID values) can be quite tiring. I can't give you any tuning values to use since it depends on the particular process or equipment.

Alternatively, you can purchase tuning software. Expertune and InTune come to mind. You can even hook it up to your PLC and it will automatically set the PID values itself.

Control Soft also has a nice PID Tuning Pocket Guide which you can get at their website. Select DS405 (PID Loop Tuning Tips Pocket Guide, Version 2.3).

Thanks Vulcan for the feedback. On my ratio, I made a typing error Gas is 300 and Air is 1200

This info may be dated, and subject to correction.First, I would make certain that the flow( Both Fuel and Air) controllers are properly calibrated.Use a Transmation unit or other substitute box.Make sure that your Thermocouple controller is properly calibrated.Input 8 milliamps, and check for 25% .Input 16 Miiliamps and check 80% span.This avoids going below zero and above 100%.Make certain that the ratio is correct by observing the output versus the input.If either input changes, the corrresponding output should change accordingly.

After all transmitters and receivers are calibrated, tune the thermocouple controller as suggested by Vulcan,and treat the other 2 controllers as one unit since they are cascaded from the same signal, and the ratio is fixed.They should work together no matter what the demand.The second two controllers only control the fuel air mixture, lean or rich'.to ensure proper efficiency.Miscalibration of either could cost lots of money in wasted fuel over the long term.

If your fuel/air numbers are accurate, your process is very large, and will not have fast oscillations due to high thermal mass.Be patient.

Hope this helps.

HTRN

If previous assumption is right then ur FG flow control which takes input from temp contr gives o/p to air by ratio contr. Im a chem engg so have less knowledge in tuning.... studied of complex methods like loop stability check etc but are there simple thumb rules to give good initial values to pid constants

Dear Brother,

It is rightly pointed out in the discussion that the temp cont o/p is fed to FG cont, which o/p inturn is used for combustion control by controlling the air. O2 trimming can be used in supplement for fine air control. Also one should first tune the FG controller and then come back to temperature control tunning.

For tunning, by keeping the loop in maunal, increase set point by some constant. Ensure proper care is taken. As setpoint increases, see the behaviour of controller. If slow increase the gain accordingly. Once gain is set, tune Ti accordingly.

The procedure takes time, being a cascade control.

Hope it answers to your querry.

Have a nice time.

Start at the primary controller somewhere between: 50-75%

1.P=50%

2. I= max

3. D=min

Observe process stability and set P to a value where oscillations are acceptable.

Then you can Halve I and double D.

Proceed to the secondary controller and follow same procedure.

A basic tuning theory goes like this:

Start with gain only.Upset the process temporarily(increase setpoint), then return setpoint to starting value.Increase gain until you get a sustained oscillation.Decrease the gain until an upset in the process results in a quarter-wave decay: that is, each succesive oscillation is 1/4 the amplitude of the previous one.Do not worry about whether it is centered around setpoint at this time.When you have achieved 1/4 wave decay, measure the cycle time of 1 period.Take the reciprocal of this time.This is your reset(Integral) value.Derivitive will be 1/8 of Integral value.

This usually gets it pretty close, with fine adjustments from this point onward.

Hope this helps.

HTRN

Without refuting your technique (I've used it also), but having the process oscillate while tuning is not always advisable. In some cases, it can be quite dangerous.

Our OP's particular application is probably safe enough to do the "sustained oscillation" technique. I would not want to do it in a nuclear power plant though .

As with many things, one size does not fit all.I do not recommend 1/4 wave decay tuning where large deviations from setpoint can be destructive.If the option is available on the controller, I suggest setting limits on the output until preliminary tuning can be done.This will prevent wild swings of the process varialbe.Your SOP sounds reasonable to me.More than one way to tune a cat.

The new self -tuning controllers are the way to go, much improved from their debut in the '80's.

My experience goes back to the '70's to pneumatics, and hybrids (I/P P/I) DDC controllers,Taylor/Sybron/Combustion Engineering /Foxboro/Honeywell/Fischer Porter/And just about every other controller of that time.I have been away for 20 years now, and I know things have changed, but I try to keep abreast of new developments.Smart transmitters, distributed control, etc.I realize my data base is dated, but some things do not change greatly.I wish I was a young man just getting started today.There is so much furthur to go than we have been.

Your knowledge is obviously more up to date than mine, so I will defer to your opinion on this matter.

Some of the new Omega controllers have both fuzzy logic and a learn mode . You can put theses controllers in a learn mode and put your system in manual, and run it thru out it's range and the let the controller self tune. After it gets a good PID set up it's self, you can fine tune it to what your application requires. I would also check on the lag time of sensors to correction time.