Air pressure control loop

Pandiani

Lifetime Supporting Member
Join Date
Apr 2005
Location
Tz
Posts
718
Hello guys,
I have one not well defined question about one control loop. I'm not interested in details,
but general cause of the following behavior:
Let's say there is a control loop for controlling air pressure in some kind of pipe's ring.
There are three lines as inlet to the air ring, on each there is an air flap that is controled by the
controller. Controller is some kind of black box, but I know for sure that is PI. Three air flaps are
actuators, and air pressure is maintained by opening or closing these flaps.
If one flap is switched from auto to manual and maintained open, for example, 80% and other two is left
on automatic mode, interesting behavior is happening.
For example, if set point is set to 20 mbar, controller is acting in such way that PV (actual pressure)
is oscillating in range from 18,5 to 19.5 mbar and causing two air flaps constantly to open and close
alternatively by 5 %. Practicaly whole system is in slight oscillation state. Air flaps are constantly opening
and closing. I'm not sure why this is happening, since when all three flaps are on automatic mode, everything
is working fine, after some time flaps are open in such way to produce steady air pressure in the ring.
(without steady state error due to PI control).

What can cause this behavior, generally speaking? What are possible remedies to this problem? Maybe to
try to increase dead band on the controller, because 19,5 mbar if 20 is setpoint is not a big deal?

Can you give me any recommendations?
Thanks
 
Last edited:
If they were all setup and tuned together then thats how they need to stay. By removing one part you have changed the gains in the system.You could have different gain settings for the number of air lines running in auto.

Just my guess.
 
I agree, it is tuned and optimized to control three flaps, and is now overshooting and oscillating.

Can you plot the oscillation of the other two flaps position command vs the pressure signal?

I am curious if it would eventually stablize, and what is the rate that the flaps change direction? Are the two automatic flaps synchronized? Does the controller attempt position all three flaps with a single output command or to the same position in auto?

Maybe there is a way to adjust the tuning of the loop when one flap is removed from it's control. possibly add some deadband.

Is the output flow rate of the air ring constant?
Are there disturbances in the output flow rate that keep the flaps from stabilizing?

Just food for thought...
Good Luck!
Paul
 
Like I said, problem was not well defined and explained, but today, I managed to get more information about structure of control scheme. You can find it in the uploaded file. As you can see, designer is used three identical PI controllers, and number of controllers (flaps) that are in auto mode determines a constant value that is multiplied to the error signal.
I made a sketch and left out binaries that are used to test and indicate if analog signals are valid. It is very interesting how this control algorithm is implemented. You can see that measurement values of % of openess of flaps (throttles) is multiplied with constants (0.961, 0.99 and 1.053) which in sum are 1, and that delta (error) signal is made by subtracting this value from average value...
What do you think of this?
Have you ever seen this type of control algorithm?
 
Multiple PIDs controlling the same PV will fight each other. The integrators of the PID will wind up or down. What will prevent this? I will study the diagram some more but my intuition tells me this is stupid. I, my customers, have never been able to get two PIDs to control pressure at the same time. Normally we would control one axis in position mode and the other axis is controls the pressure. I would suggest controlling only one throttle using a PID. I would use feed forwards or model to control the other throttles. The throttle control by the PID would just correct for modeling errors of the other two throttles. Each of the throttles would need to be able to be a trim or PID throttle in case the original trim throttle fails.
 
Well, there are saturation and dead band blocks, which I didn't draw, because the sake of simplicity, I didn't think they are important. Limits of the saturation blocks is determined by the constants which are chosed depending on binary signal that validate analog signals. After delta A1 (or A2, A3) if formed (average - A1) signal goes to saturation block first (-1 to 1) and then to dead band block (-2 to 2) before adding C1*error. Maybe this is important and prevent this from windup, but I'll need more time to draw it. I didn't think they are important to get conclusion about how this working overall. I'm generally interestedin idea and principle of working.
I'll draw complete scheme later...
 
This is a complete scheme...
Other two control loops for A2 and A3 are exactly the same. Practicaly there are three PIDs for each flap (throttle). If two controllers are in auto mode, C1 has a different value, so C1*error is added to each signal before entering PI controller. The strange thing is the why constants 0.961, 0.99 and 1.053 are used to multiply "%" of openness of each throttle. Strange how three subloops are working. How would you call this control algorithm?
 
Main PV (air pressure) is air pressure in mbar. PVs for three loops are basically a measurements of how much in % throttles are open, while SP is same for all and it's average.
 

Similar Topics

Here is my project (using DeviceNet): I have a machine hall and a warehouse. I have to try to keep the air pressure differentials as follows: -...
Replies
0
Views
7,816
Hello Everyone , Hope you all are well and safe sound! I'm looking solution for the pressure transmitter which are installed in Blender or Lpg/...
Replies
8
Views
2,854
I need to ramp up the psi on a pnuematic cylinder until the force on the piston, measured by a load cell, is met. Then I need the air pressure to...
Replies
9
Views
4,516
On site we have Yokogawa EJA110E differential pressure transmitter connected to process air line. I'm not sure if this can be used to measure air...
Replies
9
Views
4,358
Can someone explain what are the consequences if the available instrument air minimum pressure in the plant is increased?. Will it affect...
Replies
15
Views
4,912
Back
Top Bottom