# Thread: General PI Control Question for Lab

1. Professional Member
Join Date
Dec 2007
Location
Connecticut
Posts
1,162
Post Likes
Originally Posted by apprentice3
Thanks for the reply. I will keep the minimum percentages in mind. The sequence called for a minimum of 20% which I already have in the logic, but what can I tell the engineer the reason is for the higher minimum? Another thing I was thinking about in regards to the suction pressure is, since the PID for the suction pressure loop is direct acting. The suction pressure drops which in turn reduces the compressor capacity. There are certain abnormal events like low airflow, low charge, etc. that can cause the suction pressure to run at lets say 100 psi. This would in turn keep the compressor running continuously at minimum capacity. Do you incorporate logic that states if the suction pressure is below X pressure for X amount of time, then turn off the compressor on a low suction pressure fault?
The minimum was what the AAON startup guy wanted so that's what I set it to. As far as logic for low suction pressure, the entire dehumidification sequence is controlled from 1 PID. As the pressure drops, the demand goes down and controls the staging up/down. Every tome there is a stage up/down request, the PID is held at it's current level for the stage up/down delay so as to prevent unnecessary cycling.

2. Regular Guest
Join Date
Feb 2011
Posts
22
Post Likes
May I suggest that you Google "cascade control"? What you're being asked to do is SOP for many people in the process control world. Think of it this way. You create a PID based on space temp and setpoint. The PID (PI) generates an output of 0-100%. Take that output and use it to reset the discharge temp setpoint of your unit from say 55 to 105. That discharge setpoint becomes the setpoint for the discharge temp control PI.

A typical "reset schedule" that we use commonly for determining the setpoint of a steam to hot water heat exchanger [say 10 to 60 OAT = 160 to 110 Supply water Temp] is nothing more than the "P" in PID. It is straight proportional control. Response is determined purely by deviation from setpoint. Doing what is outlined above is nothing more than adding the integral element of a PID. Now the output of the "reset schedule" isn't impacted only by the deviation from setpoint but also the length of time there has been a deviation.

The technique is particularly useful for systems where there is significant time lag between a deviation from setpoint and the mechanical systems ability to control it.

Example: You're asked to control a unit mounted humidifier based on return air humidity. If the unit has been off and the space humidity has drifted downward from setpoint direct control of the humidifier from return air will probably cause the valve to open fully. "But the discharge humidity hi limit will prevent it from opening too far!!" you say. Hopefully it will but the result is ugly as the high limit PI and return air PI fight it out.

But suppose the output of the return air PI didn't control the valve. Suppose it output a value from 0-100 that caused the setpoint of the discharge humidity PI to vary from say 35 to 90%? A more elegant solution, I think.

Once you've used this a few times you'll be able to amaze friends, co-workers, and engineers with the results you can achieve!

3. Originally Posted by CraftyNRingwise
May I suggest that you Google "cascade control"? What you're being asked to do is SOP for many people in the process control world. Think of it this way. You create a PID based on space temp and setpoint. The PID (PI) generates an output of 0-100%. Take that output and use it to reset the discharge temp setpoint of your unit from say 55 to 105. That discharge setpoint becomes the setpoint for the discharge temp control PI.

A typical "reset schedule" that we use commonly for determining the setpoint of a steam to hot water heat exchanger [say 10 to 60 OAT = 160 to 110 Supply water Temp] is nothing more than the "P" in PID. It is straight proportional control. Response is determined purely by deviation from setpoint. Doing what is outlined above is nothing more than adding the integral element of a PID. Now the output of the "reset schedule" isn't impacted only by the deviation from setpoint but also the length of time there has been a deviation.

The technique is particularly useful for systems where there is significant time lag between a deviation from setpoint and the mechanical systems ability to control it.

Example: You're asked to control a unit mounted humidifier based on return air humidity. If the unit has been off and the space humidity has drifted downward from setpoint direct control of the humidifier from return air will probably cause the valve to open fully. "But the discharge humidity hi limit will prevent it from opening too far!!" you say. Hopefully it will but the result is ugly as the high limit PI and return air PI fight it out.

But suppose the output of the return air PI didn't control the valve. Suppose it output a value from 0-100 that caused the setpoint of the discharge humidity PI to vary from say 35 to 90%? A more elegant solution, I think.

Once you've used this a few times you'll be able to amaze friends, co-workers, and engineers with the results you can achieve!
Thanks for your reply. I understand what your saying, and I am working on implementing PI control over the reset setpoint as we speak.

Thank you everyone for your feedback on this! It is much appreciated.

4. I was doing some additional research on cascade PI loops, and found some recommendations on how to tune the loops. The information that I found stated that I should tune the secondary (DAT) control loop first. It also stated that it should be setup faster than the primary (Room Temp) control loop. Any other recommendations for tuning these loops? Any good starting parameters I should consider?

Thanks for the feedback.

5. Professional Member
Join Date
Dec 2007
Location
Connecticut
Posts
1,162
Post Likes
What product are you using?

6. Originally Posted by The Reaper
What product are you using?
I am using an American Auto-Matrix/ Cylon Auto-Matrix product called a CBX controller. It is their new expandable programmable controller. For PID constants it is using Proportional Gain, Integral/ Integration Time, and Derivative Time.

http://www.aamatrix.com/products/cbx-8r8

I assume most people haven't worked with it much, since this particular controller just came out a few months ago.

7. Professional Member
Join Date
Dec 2007
Location
Connecticut
Posts
1,162
Post Likes
I did alot of the AAM PUP controllers back in the day but haven't touched the Cylon stuff. As far as cascaded control, instead of using 2 PI loops, I use 1 PI that is fed a setpoint by a reset block. It's way less time in tuning and it works just as well. It works exactly like a pneumatic reset controller, very stable and fast responding.

8. Originally Posted by The Reaper
I did alot of the AAM PUP controllers back in the day but haven't touched the Cylon stuff. As far as cascaded control, instead of using 2 PI loops, I use 1 PI that is fed a setpoint by a reset block. It's way less time in tuning and it works just as well. It works exactly like a pneumatic reset controller, very stable and fast responding.
I have a reset block in there as a backup if I don’t like the results of PI Control. Typically I have always done a fixed reset when I am doing hot water reset and supply air reset based on a average VAV temperature. With this being such a high profile space, I thought I would give PI Reset a try. I will see what happens, and if I don’t like the result I will go to a reset just like you stated. Thanks for your reply.

Sent from my iPhone using Tapatalk

9. Originally Posted by apprentice3
I was doing some additional research on cascade PI loops, and found some recommendations on how to tune the loops. The information that I found stated that I should tune the secondary (DAT) control loop first. It also stated that it should be setup faster than the primary (Room Temp) control loop. Any other recommendations for tuning these loops? Any good starting parameters I should consider?

Thanks for the feedback.

For actual constants, they are going to be all over the shop as no 2 PIDs work the same. Giving that controller, your probably one of a handful using it currently.

10. Originally Posted by orion242

For actual constants, they are going to be all over the shop as no 2 PIDs work the same. Giving that controller, your probably one of a handful using it currently.
Yeah this is the first job we are doing with them. Probably not the right job to be experimenting on, but the expandable IO with HOA switches was a necessary thing on this job.

Thanks for the feedback.

Sent from my iPhone using Tapatalk

Page 2 of 2 First 12

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•