PID Lead-lag "feedforward" ?

[Ben]

Hi all!

I am looking for advice or recomendations re: if a feedforward solution will help me and if so recomendations to implelent same.

Challenge:

My customer has a chilled water system that has what apperas to be on-/off control for its temperature. If I watch the temperature of the water coming out of the chiller I can see about a four degree swing in the temperature.

My customer whats me to add a PID to operate a control valve that will mix the chilled water with warm water to further regulate the temperature.

Issue:

It seems that at best, the new PID will be constantly disturbed by the chilled water variations and a stable temperature will be a fleeting dream.

Question:

If I monitored the temperature of the chilled water coming out of the chiller, it seems to me that I should be able to anticipate the change and start my correction before the chilled water makes it to the control valve.

So....

I would greatly appreaciate hearing what Y'all have to say and any advice!

Thank you!

Ben

[johnsold]

Ben,

Where is the setpoint in comparison to the range of temperatures of the chilled water and the warm water? How stable does the customer need (as opposed to want) the temperature to be? And how much variation is there in the heat load generated by whatever is being cooled?

I think all of those may be factors in designing a control system which has a chance of actually working.

Lynn

[muks]

---

My customer whats me to add a PID to operate a control valve that will mix the chilled water with warm water to further regulate the temperature.

---

In addition to Lynn's question,After regulation will there still be some degree swing?If so what is acceptable limit?

[Ben]

"Where is the setpoint in comparison to the range of temperatures of the chilled water and the warm water? "

If the chiller was stable I would not have to implement the PID. When using my new PID the set point of the chiller will be below the set point of my PID.

The warm side... hmmmm....

From the P&ID it looks like it will vary at the same rate as the Chilled water but it will be warmed by some heat exchangers prior to me mixing it. It will probably be out of phase with the chilled water variation but of the same frequency with an average value above my set point.

"How stable does the customer need (as opposed to want) the temperature to be?"

As is the case with most cutting edge research, "as stable as possible". The last time I put together a control system for this customer he insisted on plus/minus 1 degree C. But in that case i was using a peltier (sp?) to regulate the temp. I don't think this group knows enough about the science they are studying (super-critical CO2), to be able to quote a number.

"And how much variation is there in the heat load generated by whatever is being cooled?"

As it stands they are try to achieve steady state condtions so the heat load should be constant once everything has stablized. There is one other factor that will intorduce some variation to the heat load and that is the air-conditioning system in the lab.

Here is the GUI Overview screen

A - Control valve pair

B - Chiller

C - Heat exchanger to warm up warm side of mix

D - Temp out of chiller

E - The temperature I have to regulate

What do you think?

Ben 

Message Edited by Ben on 07-25-2008 10:54 AM

[muks]

Wow Ben,
                   That defenitely is a colorful GUI.

[Ben]

"

Wow Ben,
                   That defenitely is a colorful GUI.

"

You should see it in action. All of the colors ramps update live to illusttrate the temperature gradiants.

Ben 

[johnsold]

Ben,

A reservoir between C and E with sufficient capacity to provide the needed flow for 3-5 times the chiller cycle time might do the trick. Let the average temperature be your friend. Reservoir = Low pass filter. Regulate the temperature in the reservoir rather than trying to regulate instantaneous temperature in the flow. Keep the water mixed (baffles?) in the tank. The chiller alone might be sufficient.

Lynn

[Ben]

Lynn wrote

A reservoir between C and E with sufficient capacity to provide the needed flow for 3-5 times the chiller cycle time might do the trick. Let the average temperature be your friend. Reservoir = Low pass filter. Regulate the temperature in the reservoir rather than trying to regulate instantaneous temperature in the flow. Keep the water mixed (baffles?) in the tank. The chiller alone might be sufficient.

Lynn, YOU are a genius*!

Ben

* Feel free to quote me.

[johnsold]

Ben,

Sometimes it helps to have a different set of eyes look at the situation.

I often find myself working with people who do not really understand how their processes work. The result is that they may ask for something which is somewhat different from what is really necessary. What I try to do is to dig down enough to determine the real problem and then work on that.

Lynn