Behaviour of Controller on loss of Input Signal 2

[divine7]

Hello All,

I have this challenge with a controller I am trying to program. I need to understand what is actually happening and that is the reason why I have created this forum. I plan to contact the manhufacturers of the controller for help in addition but I do not want to be lazy. I want to do my research first.

Basically my controller reads a PV (pressure) and varies the output (actuator). When I break the input or remote setpoint signal, I get a controller eror, which is normal. On restoration of the input or PV signal, the controller error clears, which is also fine. It is what happens to the output which is of concern for me.

The actuator output does not 'freeze'when I break the input signal but goes down to zero. On restoration of the signal, the actuator output falls to zero, stays there for some time (maybe "integrating")and then rises to the value it was before the signal break.

I would like a bumpless transition from loss of signal to restoration, where the controller just continues acting from where it was without having to ramp the output first to zero, 'soak' for som time and then rise again to the value pre-signal break.

Has anyone successfully accomplished this bumpless transfer on any controller?

Regards.

 

[IRstuff]

Why would you want that? How will you tell later on that your sensor messed up? Seems to me that you want the controller to lie to you.

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[divine7]

Thanks for your reply.

What I am aiming to achieve is to prevent the valves from closing (being driven fully shut) when there is a sensor break. It is an undesirable state for the process and so I need to program this anomaly out, if you see what I mean.

I have an alarm that will warn me that there has been a controller fault and then I can send an engineer to the local panel to investigate. So, yes, I have a means of knowing that the controller 'messed up'

 

[DonUK]

I've not done it but I'd continually write the current input value to a "last scan" register (or a timed back up register if more suitable) and then when I got the device error I'd switch to using this saved value.

When the device returns from its error condition switch back over after a suitable time delay to allow for settling. Or, neater, compare the difference between actual and saved input values and have a ramp routine to slowly adjust the value used by controller from the saved value back to the actual value. That might not be necessary if you have a sluggish controller response anyway.

 

[divine7]

Thanks for the reply DonUK. I do not have data logging facility as part of this simple control system but what I will try is to use the Sensor Break Mode facility.


If configured as ‘Hold’ the output of the loop will stay at its last good value.
On exit from sensor break the transfer is should be bumpless – the power output will step from its held value to the control value.

That's the first thing I have come up with.

 

[DonUK]

No you won't need data logging - everything I mentioned can be done with basic PLC functionality and either a bit that tells you the device has been disconnected or your own code to detect it.

Personally I would manage the controller input as in my post above since that is the source of the issue but, as is usual in PLC control, there's more than one way to skin a cat.

 

[divine7]

DonUK,

Thanks for your recommendation. I will get work on it and will update you on any progress over the next couple of days.

Regards.

 

[IRstuff]

One thing to consider is false alarm rate. You may need to "filter" the disconnects if they happen often and intermittently, so that the tech doesn't wind up going there to find nothing wrong.

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[divine7]

Thanks IRstuff. Actually, loss of the remote set point signal is going to be a rare occurrence but I am considering it in my design so I can improve the reliability and availability of my system.

I think I have come up with a modified program and I hope to test this tomorrow and feed you guys back with what I did and the outcome

Regards.

 

[djs]

From your description you appear to be using a PID to control the output. A solution to your issue is that when a sensor failure is detected, freeze or place the PID into manual mode which will freeze the output. When the sensor condition clears, place the PID back into auto. What you are experiencing is call PID "windup". This occurs when the PID looses control of the loop for some reason. Another example is when the output actuator is placed in hand or manual mode. For safe stable control you must monitor the status of all devices in the loop, and if something happens to any of them, place the PID into manual mode.

 

[divine7]

Thanks for your tip djs,

The system is expected to be in auto during normal operation. Operation in manual mode will initiate an indication in the control room. We do not want to distract the operator too much.

I get your point though and that sounds like a robust design.

 

[djs]

Loss of the sensor signal should trigger an alarm.

 

[itsmoked]

It should certainly be known about in the control room!!

Keith Cress
kcress -

http://www.flaminsystems.com

 

[divine7]

itsmoked. It is. I agree. There is an alarm in the control room that is initiated when there is a controller fault.

 

[itsmoked]

Whew.

Keith Cress
kcress -

http://www.flaminsystems.com