I have a client with a legacy control system that holds in to among other
functionalities about 100 PID control loops with some 150 AI and 60 AO.
They want us to check if it would be possible to work in a PLC
How many PIDs I can place in a PLC? Siemens, Modicon...
How many are recommended?
As a retired controls person I may be a little out
of touch with the latest innovations in PLC technology. So don't flame
me too bad.
I have written several very large PLC programs that had
over 100 PID loops including high-level math capability.
The systems were HVAC with space temp and humidity, reheat temp,
chiller valve control, and basin water level.
All the loops had a relatively slow execution rates, anywhere from 3sec.
to 20 sec. AI count about 400, AO count 150.
I know that the TI/Siemens 505 series with a 555 processor
will do the job, and I'm sure that some of the AB PLC5 processors will
do equally as well.
The TI/Siemens PLC has the advantage over the PLC5 because it has
dual processors, one for ladder logic and the other for loops
and high-level math functions.
Best of luck
With the right PLC you can run dozens of PID loops, possible even your 100.
It depends on the PLC, and of course the required sample rates.
The overal PLC response time may decline as more PID loops are executed,
however that should not be a problem. Each PID must be executed on a fixed
rate periodic basis and provided the scan rates are such that the total load
can be executed faster that the required scan rate there is not a problem.
And there is often a problem that many PLC programmers do not understand PID
A good example that I have seen is a Siemens PLC controlling an extruder
that had a dozen or so PID loops, some fast (pressure control) and others
slow (zone temperatures)
There are a few catches is what you write:
1. You talk about dozens. that is OK. it is not a 100.
2. Cycle time of a PID should be fixed for proper action. It works on a
different time scheduling mechanism than a DCS. Can a PLC guaranty this?
3. What will happan to the current applications? they are useually motoros
and valves and other devices that may require speedy action.
4. The example you give is great. But this is a small application.
The plant has some 400 D-IO that are being controlled int hat PLC. I
strongly advise against puting the PID in there. I wanted to know what are
others experiance in this.
Basically, no. And FWIW most PLCs can't "auto-tune" either, so be prepared
to work the loops out the hard way. Keep in mind that even modern PLCs are
still primarily designed as relay-replacements and anything else they can do
(which is a lot in some cases) is a bonus.
Good question. It depends upon the brand of PLC you select and how you
As mentioned by Trajen, the good old Siemens TI range had a different
processor setup and could handle a bit of analog, and as Francis mentioned,
the Siemens S7 range is not bad.
Both ABB and Yokogawa (amongst others) make small DCS hybrids with the kind
of I/O counts you are looking for - but, to answer your original question,
they are not PLCs. The closest to a PLC would be the AB ProcessLogix -
which wasn't actually developed by AB..
" 2. Cycle time of a PID should be fixed for proper action. It works on a
different time scheduling mechanism than a DCS. Can a PLC guaranty this?"
Yes, if properly programmed. With Siemens you would call the PID from one of
the Time based OB's . These interupt the normal scan. With Rockwell use the
Selectable Time interrupt, which is similar.
You also need some logic to spread the PID's out, so that you only call a
few on each of the time based scans.
And Yes, you can have hundreds, provided they do not have to scan too
frequently. Not knowing what your loops do I do not know if that might be
Of course, it may be better to use a 'DCS' product for the better loop
handling. But DCS is now an almost obsolete description. For example is
Rockwell RSLogix a DCS? (what about in it's Honeywell form) - and is DeltaV
more like a PLC? But I do not want to start the PLC DCS argument again!
From what I read I can figure out that it may be possible to do.
1. I'll have to manage the PID scheduling.
2. No tools for PID (Autotune, smart algorithms etc.).
3. Other SW may suffer.
4. Choose the PLC wisely and than maybe I'll be able to handle it.
I can hear from the tone of everyone that this is not advisable. I better
choose a system that is build for these kind of applications.
I thank everyone for their thoughts.
Best of all and Shana Tova (Happy new year)
After reading these posts, I thought I'd check out how long the PID
instructions take on my current Allen Bradley CompactLogix project. 4 PIDs
scan 10 times per second, each scan of the routine (containing only the PIDs
and a few OTL/OTU instructions) takes a maximum of 430usec. The routine
interupts others, to make sure it is on time.
Crudely calculated, I should be able to do over 900 PIDs without skipping
any updates. Of course performance of other routines/comms etc. will
degrade, so I wouldn't recommend it, but I would think 100 should be fine.
Latest Logix5000 PIDE instruction does have some Autotune support, but I
haven't tried it. I have used ABs tuning software (written for them by
someone else, can't remember who) - and that works just fine.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.