Hi Guys,
In most DCS, the scan time for the whole system is 1 sec or more depending on the logic. But whenever an field input change state the DCS detects it and time stamps it...
Events are time stamped on millisecond level, but DCS have time scan of 1sec or more... How such time stamping is made possible??
Regards,
Sameen
Time Stamp functionality in the DCS....
-
sameenkhan
- Posts: 32
- Joined: 07 Apr 2010, 17:54
- Area of interest: Electrical Engineering
- Contact:
-
ali.abbas
- Core Member
- Posts: 61
- Joined: 25 May 2010, 23:26
- Area of interest: Inst. & Control Engineering
Re: Time Stamp functionality in the DCS....
Sameen,
The scan time usually is much faster than 1 sec. Honestly, for PID controllers, 1 sec would be quite slow for maintaining a close process control.
Are you sure youre not confusing scan time with historization time?
The scan time usually is much faster than 1 sec. Honestly, for PID controllers, 1 sec would be quite slow for maintaining a close process control.
Are you sure youre not confusing scan time with historization time?
Regards,
Ali Abbas
Ali Abbas
-
sameenkhan
- Posts: 32
- Joined: 07 Apr 2010, 17:54
- Area of interest: Electrical Engineering
- Contact:
Re: Time Stamp functionality in the DCS....
Hi Aly,
A DCS scan includes following tasks:
1) Reading signals from Input modules
2) Input Signal Processing
3) Execution of Control Logic (i.e. PID, batch process, calculation etc.)
4) Output Signal Processing
5) Sending Signals to Output modules
Task 1 & 5 take most of the time in completion but it remains fixed.. it only depends on types of signals & electronics of modules used. Whereas, time taken by task 2 & 3 is mostly constant. Time taken by execution of control logic depends on complexity, size and data used and etc. so it may vary from logic to logic.. Vendor define time taken by each control block (i.e. PID, PVI, PID-STC, CALC etc.) in their product documentation, we can use them to calculate the time taken by control logic.. but it would be theoretical..
You are right.. for PID controller 1sec is very slow.. Well we set scan time as per size of application (i.e. control logic) & also it depends on process reaction time (such as for antisurge control you need faster control, etc)... in DCS a PID controller is given scan time of 1sec.. it does not means that it actually takes whole 1sec.. the PID algorithm executes much faster.. it is mainly dependent on how much DCS controller is loaded.. the object of setting 1sec scan time is to give a target to DCS controller to complete the task within predefined time...
I have researched the time stamp of signal variations... whenever, field signal changes it state.. it is detected by module electronics & then change is stamped.. and then data is forwarded to PID controller for control purposes.. input modules detects all changes which occurs within 1ms (subject to modules design)... You are right that is the same time stamping is also used for historization... the same concept also applies to output..
Time Stamp is done at both levels... at module level & also at controller level... Controller time stamp is more easy to understand if we are manipulating soft switch or soft/internal registers or words... but in industries we are mostly interested in time stamps of input & output signals... and time stamping done by controller is mostly ignored...
Looking forward to your comments....
Regards,
Sameen
A DCS scan includes following tasks:
1) Reading signals from Input modules
2) Input Signal Processing
3) Execution of Control Logic (i.e. PID, batch process, calculation etc.)
4) Output Signal Processing
5) Sending Signals to Output modules
Task 1 & 5 take most of the time in completion but it remains fixed.. it only depends on types of signals & electronics of modules used. Whereas, time taken by task 2 & 3 is mostly constant. Time taken by execution of control logic depends on complexity, size and data used and etc. so it may vary from logic to logic.. Vendor define time taken by each control block (i.e. PID, PVI, PID-STC, CALC etc.) in their product documentation, we can use them to calculate the time taken by control logic.. but it would be theoretical..
You are right.. for PID controller 1sec is very slow.. Well we set scan time as per size of application (i.e. control logic) & also it depends on process reaction time (such as for antisurge control you need faster control, etc)... in DCS a PID controller is given scan time of 1sec.. it does not means that it actually takes whole 1sec.. the PID algorithm executes much faster.. it is mainly dependent on how much DCS controller is loaded.. the object of setting 1sec scan time is to give a target to DCS controller to complete the task within predefined time...
I have researched the time stamp of signal variations... whenever, field signal changes it state.. it is detected by module electronics & then change is stamped.. and then data is forwarded to PID controller for control purposes.. input modules detects all changes which occurs within 1ms (subject to modules design)... You are right that is the same time stamping is also used for historization... the same concept also applies to output..
Time Stamp is done at both levels... at module level & also at controller level... Controller time stamp is more easy to understand if we are manipulating soft switch or soft/internal registers or words... but in industries we are mostly interested in time stamps of input & output signals... and time stamping done by controller is mostly ignored...
Looking forward to your comments....
Regards,
Sameen
-
ali.abbas
- Core Member
- Posts: 61
- Joined: 25 May 2010, 23:26
- Area of interest: Inst. & Control Engineering
Re: Time Stamp functionality in the DCS....
Sameen,
That was quite a comprehensive analysis by you, and certainly most informative. However, we seem to be deviating from your core query and therefore i'd bring ourselves back.
If I understood correctly, your initial confusion was that if scan times are in seconds, how can time stamping be done down to the millisecond level ... right?? Well, I was just going thru the manuals of the DCS at our site based on what I read, I could come up with the following explanation. In case you have something to add or correct please do so.
All control systems have a scan cycle in which they would scan all their parameters at least once. The scan cycle might be sub-divided into smaller cycles to allow for scanning of points that are supposed to be scanned at a faster rate. These scan cycles are determined by clocks/crystals internal to the control systems. One can obviously presume that if the control system is to maintain a scan time of the order of 1 sec or less (minimum scan time in our DCS is 0.25 sec), the internal clock should follow time to an order much below a second. In other words, the control system should be precise enough to establish at what time in that specific second was the event recorded and then assign that time stamp to the respective event.
I hope ive been able to explain my ideas clearly. Still looking forward to your comments.
Regards,
Ali
That was quite a comprehensive analysis by you, and certainly most informative. However, we seem to be deviating from your core query and therefore i'd bring ourselves back.
If I understood correctly, your initial confusion was that if scan times are in seconds, how can time stamping be done down to the millisecond level ... right?? Well, I was just going thru the manuals of the DCS at our site based on what I read, I could come up with the following explanation. In case you have something to add or correct please do so.
All control systems have a scan cycle in which they would scan all their parameters at least once. The scan cycle might be sub-divided into smaller cycles to allow for scanning of points that are supposed to be scanned at a faster rate. These scan cycles are determined by clocks/crystals internal to the control systems. One can obviously presume that if the control system is to maintain a scan time of the order of 1 sec or less (minimum scan time in our DCS is 0.25 sec), the internal clock should follow time to an order much below a second. In other words, the control system should be precise enough to establish at what time in that specific second was the event recorded and then assign that time stamp to the respective event.
I hope ive been able to explain my ideas clearly. Still looking forward to your comments.
Regards,
Ali
Regards,
Ali Abbas
Ali Abbas