SLC 5/04 PID tuning

[shanamischelle]

Hi,

I have a system (see attached illustration) that is PID controlled. The pump is used for the injection of chemical from the tank to the boiler drum. The speed of the pump injection depends on the conductivity in the boiler drum (there is a 10mins. delay in the change in conductivity).

(There are 2 pumps, one pump is operational while the other one is on standby mode)


I have read lots of threads here about pid tuning and nothing seems to work when i tested it. My output is still unstable with my PV(conductivity) fluctuating.

I also attached my test program.

 

[Bob O]

How do you have your system connected to the SLC? I only see DI and DO where I would expect to see some analog.

 

[OkiePC]

How do you have your system connected to the SLC? I only see DI and DO where I would expect to see some analog.

There's analog in there. See I:3.1, O:4.0, O:4.1.

I would suggest setting up a trend to monitor your PID operation. If you look in the tree on the left and scroll down to find thte trends folder, right click and add a new trend, then populate it with your process variable and your control variable. You can change the trend properties to get the update rate and timebase you need, and configure it to store records, set up the axis limits, and more.

A graphical view of the process would probably help you see the effects of your PID adjustments.

Since the conductivity is probably not linear in relation to the pump speed, and there is quite a long lag time, this system will probably be time consuming to tune and a PID may not be the best approach. I haven't done anything like this, so this is just my suspicion.

 

[Bob O]

Interesting because my software at home this morning only showed the DI and DO.

 

[frmanpete]

I have had the same problem with sampling time delay before. The company I worked for bought a $150,000 chromatograph for gas scrubbing process control. Unfortunately it only sampled every five minutes. I ended up sitting down with the engineer and figuring out the non-linear functions involved in the process control. (Mine happened to be CO2 % in scrubbed gas) If you can get enough info on the chemical process; IE, temp, pressure, flow rate, sg, moles/L etc. You should be able to formulate what you flow rate of chemical should be. If you wanted to used a PID to control flow rater you could. Then you could use your delayed conductivity measurement as more of a set point than a process variable; IE have a high and low conductivity set point, or just use it for monitoring purposes.

 

[shanamischelle]

There's analog in there. See I:3.1, O:4.0, O:4.1.

I would suggest setting up a trend to monitor your PID operation. If you look in the tree on the left and scroll down to find thte trends folder, right click and add a new trend, then populate it with your process variable and your control variable. You can change the trend properties to get the update rate and timebase you need, and configure it to store records, set up the axis limits, and more.

A graphical view of the process would probably help you see the effects of your PID adjustments.

Since the conductivity is probably not linear in relation to the pump speed, and there is quite a long lag time, this system will probably be time consuming to tune and a PID may not be the best approach. I haven't done anything like this, so this is just my suspicion.

yeah, ive tried trending my PV and CV during tuning but i still get the same fluctuating output.

I have also thought that using PID would'nt work. Can anyone share any methods aside from using PID?

btw, thanks for the reply.