Hi everyone,

I'm commissioning a dissolved oxygen system at a wastewater treatment works, and the PI controller is implemented by a Siemens S7 PLC. So I checked out S7 online and found a Siemens manual at the link below.

file:///C:/Users/35385/Downloads/s71500_pid_control_function_manual_enUS_en-US.pdf

The problem I'm having is with the PI description on page 32 to 33. It shows two graphs.



Then an equation which I believe is for the output y.



I'm assumming that 't = time interval since the step of the control deviation' means the time since the disturbance. Even if it doesn't, as time passes t will increase. This means that the integral term will decrease not increase as time passes and it will just be a proportional controller. Or, am I missing something fundamental.

The controller on site doesn't behave like the above equation. If the error doesn't decrease, the output starts to ramp up, as I would expect an integral to do.

Thanks for any help.

 

That equation looks wrong. Assuming Ti is the integrator time constant and t is the update interval, the units are not right because there are two times in the denominator. Also, I don't see where the output is being integrated.

 

Hi DaraghC,
Assuming Xw is the amplitude of the deviation step change (starting from zero deviation), the time t elapsed since that step change should be located in the numerator instead of the denominator in the equation : the integral term will actually generate a ramp on the controller output with a constant slope as long as the deviation remains constant. Proportional and integral terms are in series as in a standard PID structure, since the gain is a multiplication factor for both of them.

 

Don't ignore what I said above about there being two times in the denominator. How does a ( 1 + 1/(TI*t) ) make sense?
The first 1 is unit less. The second 1/(TI*t) has units of 1/seconds^2. How can these be added together?
What cancels out the 1/seconds^2 units?
There must be a misprint or you copied it wrong. It should be (1+t/TI). Now the times cancel out and the whole term is unitless.
This is very important! Do not pass Go do not collect $200 until you figure this out.
Doesn't the Siemens documentation define the variables?
There should be a SP ( set point ) and PV ( process value ). The difference is the error. I don't see and error term. If there is one it must be the Xw.

 

Hi DaraghC,
Assuming Xw is the amplitude of the deviation step change (starting from zero deviation), the time t elapsed since that step change should be located in the numerator instead of the denominator in the equation : the integral term will actually generate a ramp on the controller output with a constant slope as long as the deviation remains constant. Proportional and integral terms are in series as in a standard PID structure, since the gain is a multiplication factor for both of them.

So I'm not missing something. The equation is wrong. Thing is, this is from a Siemens document on PID implementation on a S7 PLC.
I think I'll write to Siemens. Feel a bit more confident now that both you and Pnachtwey agree it's wrong.

 

Don't ignore what I said above about there being two times in the denominator. How does a ( 1 + 1/(TI*t) ) make sense?
The first 1 is unit less. The second 1/(TI*t) has units of 1/seconds^2. How can these be added together?
What cancels out the 1/seconds^2 units?
There must be a misprint or you copied it wrong. It should be (1+t/TI). Now the times cancel out and the whole term is unitless.
This is very important! Do not pass Go do not collect $200 until you figure this out.
Doesn't the Siemens documentation define the variables?
There should be a SP ( set point ) and PV ( process value ). The difference is the error. I don't see and error term. If there is one it must be the Xw.

First of all my apologies the link I posted is from my PC. Correct link below.

https://support.industry.siemens.co...ic-s7-1200-s7-1500-pid-control?dti=0&lc=en-AR

And the graphs are on page 33 equation's on page 34.

It's not a misprint I copied that equation from a Siemens document using Snipping tool.
So, as both you and Fil agree it's wrong I'll contact Siemens and see what they say.