Tuning of cascade control

A

Thread Starter

Anonymous

For a cascade control system, would it have made any difference if we tune the primary controller first and then the secondary controller?? If yes, then why???(the usual order is to tune the secondary controller first then the first)
 
C

ControlNovice

Yes - the secondary has to be tuned first!

The primary controller will see the response of the secondary controller and it uses this as part of its tuning. So, if the primary controller is tuned first, it will 'see' the reaction of the secondary controller. This affects the tuning parameters of the primary controller.

If the secondary controller is tuned after the primary, the secondary controller's reaction has changed, and now the feedback response that the primary controller 'sees' has changed....and the previous tuning constants may not work.

So, the secondary must always be tuned first.

One more note: another requirement of cascade control is that the secondary loop should be faster than the primary loop. A good rule of thumb is at least 3-10 times faster.

Also, the secondary controller must respond faster to adverse conditions involving the valve, such as a sticking valve, mis-sized valve or wrong valve characteristics. This will confine the problem to the secondary controller and will have a much less detrimental effect on the primary loop.
 
F

Friedrich Haase

Moin all,

It makes a big difference!

Tuning a controller will change the transfer function of the closed loop. Simply put, that's why you tune it.

For a cascade the outer controller delivers a control action which in turn will be used as set-point of the inner controller. Or in other words the outer controller sees the closed loop with the inner controller and the plant.

If you tuned the outer control loop first then modifying the inner controller will change the behaviour (transfer function) of the inner control loop which in turn will change the behaviour of the outer control loop.

So, always the innermost loop comes first.

regards
Friedrich Haase

Ing.-Büro Dr. Friedrich Haase
Consulting - Automatisierungstechnik
email [email protected]
WEB http://www.61131.com
 
G

G.Navaneethan

It is not important whether master first or slave first, after all both should have been completed at the end. It is individual's convenience. It is very important how to start.

Mostly boiler control cascade loops, i am doing this and is very satisifactory.
Master controller = Gain 0.75 and Reset Middle of the scale.
Slave controller = Gain 0.25 and Reset the in
seconds. (fast)

Allow both controllers in auto and cascade control. System will start hunting and the size of the hunting will be in the increase. At this, Increase only the master's gain and you will see the whole system will come under total control.

Navaneethan
 
B
I just want to make sure that there is no doubt in this chain of emails. I have worked on chocolate factories and large power generation units and the rules are the rules the inner loop (slave) has to be tuned first,then the outer (master) loop. Even Boilers have to follow this rule.

The reason is very simple: Once the slave loop is closed (automatic) the process and the slave controller result in one single process dynamics. These dynamics are the one that you will use to calculate your controller tuning for the master controller. Every time you re-tune the slave controller you need to verify the tuning of the master controller.

Slave then Master no matter what - that's it.

Ben Janvier
Senior process control consultant
[email protected]
 
M

M. Moosavinia

In a cascade control loop the rule of thumb is that the inner loop is and must be a faster loop (>10 times) than the outer loop. The Optimum process performance, therfore is to 1st tune the inner loop before the slower outer loop is tuned.
 
Yes - the secondary has to be tuned first!

The primary controller will see the response of the secondary controller and it uses this as part of its tuning. So, if the primary controller is tuned first, it will 'see' the reaction of the secondary controller. This affects the tuning parameters of the primary controller.

If the secondary controller is tuned after the primary, the secondary controller's reaction has changed, and now the feedback response that the primary controller 'sees' has changed....and the previous tuning constants may not work.

So, the secondary must always be tuned first.

One more note: another requirement of cascade control is that the secondary loop should be faster than the primary loop. A good rule of thumb is at least 3-10 times faster.

Also, the secondary controller must respond faster to adverse conditions involving the valve, such as a sticking valve, mis-sized valve or wrong valve characteristics. This will confine the problem to the secondary controller and will have a much less detrimental effect on the primary loop.
No... Inner loop need not be faster.. Study more...
 
There is too much misinformation here.
For the best performance the inner loop should have a faster response time than the outer loop. However, it isn't necessary if you are willing to put up with sub optimal control.

If the open loop transfer function of what is being controlled is known then the solution is easy. In another thread on control.com I stated that being able to do an accurate system identification is necessary. The should also be a separate feed back for the inner loop and the outer loop. If there is only one feed back there is no point in having an inner and outer loop.
In the example I post I assume there is a velocity feedback for the inner loop and a position feedback for the outer loop. However, if the inner loop feedback is derived from the outer loop feed back then both loops should be combined into one transfer function.

I have an example from the CTM PID.
https://deltamotion.com/peter/Mathcad/Mathcad - DC Motor Velocity.pdf
The ratio of the inner loop bandwidth to the outer loop bandwidth is 3.32 not 10. I can do this calculation for any system.
There are too many myths about control theory. I don't know where they start but I know that who ever started them did not do the math and whom ever repeated them didn't ask for the proof.

I admit that my example is just one but I can use the same technique on any system. The ratio of the inner to outer loop band widths will not always be 3.32 but I can calculate what it will be given the open loop transfer function and where you want the closed loop poles to be. So shame on the false engineers that didn't ask for or do the math.

BTW, I think that Matlab is great for getting answers but it is a poor tool for teaching students. Getting answers without understand the techniques necessary to get the answer is not good.
 
Yes - the secondary has to be tuned first!

The primary controller will see the response of the secondary controller and it uses this as part of its tuning. So, if the primary controller is tuned first, it will 'see' the reaction of the secondary controller. This affects the tuning parameters of the primary controller.

If the secondary controller is tuned after the primary, the secondary controller's reaction has changed, and now the feedback response that the primary controller 'sees' has changed....and the previous tuning constants may not work.

So, the secondary must always be tuned first.

One more note: another requirement of cascade control is that the secondary loop should be faster than the primary loop. A good rule of thumb is at least 3-10 times faster.

Also, the secondary controller must respond faster to adverse conditions involving the valve, such as a sticking valve, mis-sized valve or wrong valve characteristics. This will confine the problem to the secondary controller and will have a much less detrimental effect on the primary loop.
Hello, ! I have a question. If the inner loop control time is less than three times that of the out loop, can I regard the cascade control as unnecessary? thank you
 
Top