Thread Starter

Rajesh Mehta

I know, it has been a matter of debate for long but I would like to get fresh views on the subject in view of the fast changing technology.

There have been claims and counter claims from various manufacturers that their control system is a DCS or PLC. I would like to use a neutral term like control system to avoid confusion.

My question is what are the attributes that make a PLC to classify as a DCS or vice versa? In the modern day scenario, the benchmarks have to be different from the old notions like the DCS is used for continuous process control (PID/ CLCS) and PLC for discrete logic control (OLCS) as today's PLC's and DCS' are very much capable of performing both the functions. So where is the border line?

Or is using a neutral term like Control System the answer to this debate....

Appreciate your views.

Thanks in advance,
Rajesh Mehta

John Flanagan

A DCS can have any variety of PLCs or other control devices, as long as they can communicate over some sort of digital network. Usually, (but not necessarily) the operator interface will be on a computer screen on the same network.
The network connection can be made using LAN, serial or other proprietary communication system. We have connected stand-alone PLCs to our Honeywell DCS system, replacing pushbuttons and lights with CRT graphics, but the control logic did not change.
John Flanagan
Dear Rajesh,
I would base my differentaiation on two points:

1. Configuartion : Whetehr a single point configuartion is possible by using one software package supplied by the hardware vendor. In most PLCs, it is a two step procedure of PLC programming + HMI configuration.

2. Most PLC manufacturers , do not go into the minor details such as , for e.g., powering a two wire transmitters.The built in current limit functions are noot built into the analog input module.Panel designer has to use additional hardware to achieve currrent limit functions.

Kevin Totherow

There is not as much room for debate on the issue of DCS or PLC as some would have you believe. A DCS is a Distrubuted Control System. The system is made up of many controllers connected by a data highway system. A PLC is a programmable logic controller - which is a single processor.

The DCS was invented for the process industries to replace benchboards of pneumatic and panels of single loop analog controllers. All DCS's were capable of discete control but it was very expensive. All DCS's are easy to configure PID loops, usually have audit trail functions, have very sophisticated alarming capablitities.

The PLC was invented to replace relay systems. The PLC scan time is extremely fast to meet logic needs. The Operator interface is an after-thought as was PID control. PLC's generally scan so fast that they can handle multiple PID loops - but you have to engineer the solution.

The traditional DCS is now about dead for all but the largest process applications. Hybrid DCS control solutions are the way of the future for medium-scale process applications. Some of the "Hybrid" control systems are DCSs meant for the process industry and just make the I/O cheap enough, and allow function block or ladder programming, so that a PLC is not needed for the discete control.

Stringing several PLCs on a highway would constitute a "DCS" but the real DCS guys put years of work into building maintainable systems for the process industry.

A strong functional specification of needs will quickly weed out DCS from PLC and all those that try to confuse the two.

Hope this helps.

Take an example:
Bailey NET-90 could do lot of logic, and can close the loops. It's an " hybrid ". Take AB (some years ago) it was enough for logic. But insufficient for control strategies. So, what is the difference. In the Net-90 there is a lot of numerical maths helping and supporting control loop algorithms. Closing the loop in PLC means nothing. It's an ON/OFF animal. DCS may have ± maths available, but at least they opened the cracker jack box.
With the newer and fancier "Control Systems", a DCS can emulate the discrete functions of a PLC, usually at a higher cost; and a PLC can emulate the contituous control of a DCS. I agree. However, what separates the DCS from the PLC, and makes it more expensive, is

1. The ability to modify the configuration and programming of both the controllers and the HMI "on the fly" without having to take the system off line to recompile or reboot the system.
2. The ability to add or remove I/O cards from a controller without taking the process "Off Control"
3. The availability of redundancy throughout the process (including I/O) that is transparent to the operator.
4. The ability to remove and replace any single piece of hardware without taking the process "Off Control"

The client must make the decision as to just how much "bulletproofing" he is willing to pay for. If the process you are controlling is capable of frequent startups and shutdowns (such as batch processing), then the system based on low cost controllers and PC's running VB HMI's make sense. If however, the process is one that requires many hours or days to start up (such as Petrochemical or Refining processes), with costs in the range of $100,000 US per hour of down time, then the choice of the more expensive DCS is more easily justified. From experience, this will require acceptance and championing from local operating
management to prevent a cost concious project manager from "doing it cheap".

At least that is IMHO.

John Beck

Hall, Christopher W

It is perhaps worth remembering that with the current third generation DCSs that are IEC 61131 compliant, such as DeltaV and Freelance, the price
differential between PLC and DCS for anthing other than a very simple application has all but disappeared.

Hardware prices continue to fall. Manhour rates continue to rise. For larger systems the engineering manhours to configure, test, commission, troubleshoot, modify and document can be considerably less for the DCS route than for the PLC.

The great advantage of DCS is that there is only one database not two to maintain. However, this implies that configuration managment must be more
rigorous, for with greater power comes greater responsibility.

Chris Hall
Control Engineer - speaking for myself.

Johan Bengtsson

I think the border is slowly dissapering and basically like the idea of "Control System" (CS?) or perhaps "Programmeable Control System" PCS - well whatever - to the question

I think the border is located at the same place even if it is less significant now than it was before. Todays PLC can sure enough handle continous process control and todays DCS can of course handle discrete logic. BUT they are still more suitable (by system design, by programming tools, and so on) to do their "traditional" duties. I think it is somewhat easier to put up some 100+ PID controllers with some cascaded and so on in a DCS than in a PLC and make it do what you wanted with timing and such considered. At the same time it is more easy to make some 1000+ rungs of ladder in a PLC than in a DCS and make that work well considering timing and so on. This is changing and the border is dissaperaring but the difference is still there somwhere.

Ok, this is my opinion, let's hear others...
Especially if someone knows a system that is actually a CS (ie somewhere really *at* the border and not just on one side)

/Johan Bengtsson

P&L, Innovation in training
Box 252, S-281 23 H{ssleholm SWEDEN
Tel: +46 451 49 460, Fax: +46 451 89 833
E-mail: [email protected]

Rajesh Mehta

Well, here we have a no. of viewpoints and quite fresh ones too.

I would like to look at it from the point of view as to how the processor looks at a piece of code.

Fundamentally, the PLC goes about it in a sequential manner i.e. begin from the first line and get to the last (unless there is an interrupt, of course) and can not be really Multi-tasking. So essentially the code execution time has to be less than the cycle time (for complete code).

Whereas the DCS looks at the code in a time shared manner. Remember the cycle time specified to each task. The cycle time (time within which the processor MUST look at a piece of code) can be less than the execution time for the code. Any unfinished task is finished the cycle.

That is just about one of the differences.


PS:Any comments are welcome for discussion.
The Houston ISA section had a meeting June 6 titled "DCS/PLC/Hybrid shootout and panel discussion". Streaming audio (>1 hour) and speaker slides are available online at . Just following the link on that page.

Paul Gruhn, P.E., C.F.S.E.
Siemens, Houston, TX
Houston ISA Section Web Master

Heavner, Lou [FRS/AUS]

My take on this is if you are making things, you generally ought to consider a PLC. If you are making stuff, you ought to consider a DCS. Vendors know what their sweet spot in the market is and so they will tell you if they are selling a DCS or a PLC. That is what really defines what a control system is. :) Fortunately, most vendors have chosen to battle over where there
products are suitable rather than over what they are called.

If your requirements are such that only a DCS or a PLC will work for you and you are trying to figure out which, then you should probably ask somebody with more history in the business. If you are in the grey area where either can be competitive, then you better try to nail down the functionality that is really important and makes a difference to you. A lot of the grey area
is in batch or sequential processes where there is a blend of discrete and regulatory control. Despite all the harranging that salesmen get, they can usually tell if they have a decent chance giving the customer a competitive and functional solution. If they don't think they can, they are usually not going to make much of an effort to win the order.

I'm not sure it matters too much, anyway. If you believe some of the pessimists around here, there won't be many recognizable names left in the
automation business much longer and most of their product lines will suffer the fate of the dinosaurs. If you listen to CWW and his gang (no offense Curt) then everybody will be just rolling their own with open source software on Linux. But even none of that pans out, you can bet that control products including instrumentation are going to get smarter, smaller, faster, cheaper, easier, etc and there is a fair chance that fieldbus technology will become increasingly common if not dominant.


Lou Heavner
Emerson Process Management

Kevin Totherow

Many of the current DCS's have only one database, or one point to configure for both the controllers and the HMI, but that does not define a DCS. Many DCS's had, and still have, configuration of points in the controller and configuration in the HMI or two databases.

As an end-user I fought hard at user group meetings in the 1980s and early 1990's to get companies like Fisher to create a common database. Today, I will tell you that the common database is more of a marketing issue than a practical concern. Your plant probably has several databases if you count the points in PLC's and other foreign devices, the DCS, the historians, the online/offline models, etc., etc. Clearly, the number of databases will increase with open systems and the desire for process information that is now in the DCS.

I disagree that the difference between a PLC and a DCS is all but disappearing. Try getting a DCS to do a high speed sequencial manufacturing control or a relay fast safety system. Try getting a PLC to control a paper machine where there is a need for 99% uptime on controls, one week of outage per year and fifty small changes that need to be implemented online during the year.

It is the easy, common, applications where the PLC and DCS can overlap. You can drive one or two nails with a pipe wrench pretty well but you wouldn't want to build a house that way.

Several vendors are actively trying to fuzzy the definitions. Simultaneously, some engineering firms are clouding the issues by installing low cost, hard to maintain "systems".

Write a functional specification of what the system must do and how it will be used and the PLC/DCS issue will fall into place.


Kevin Totherow

Net-90 can do logic. So can every other DCS ever invented. The PLC is very inexpensive, very fast, programs in ladder logic, has easy ways to force I/O. I think the "hybrid" systems are those that can actually program like a PLC, and compete (a little bit) pricewise when an HMI and highways are needed.

Kevin Totherow
Hi Kevin!

That was really original.

Can you give some points to ensure that a specific control is done only by a PLC or a DCS.

I would really appreciate if you can give some such points.

If I am not mistaken, the ControlLogix PLC system can run subroutines on a schedule bases or a non-scheduled bases. Which means that subroutines that are on a schedule bases will run on a
periodic bases. So, if you have several subroutines as a scheduled task, the processor will execute them depending on the time interval and priority even if it is doing something else. Hence, it appears that the system "multi-tasks" in the old WinTel alliance definition of the term with the old PIC chips for interrupts and vector table at the first 256 bytes of memory ( the ControlLogix hardware infrastructure I am sure is a lot more sophisticated then that, but that is what it reminds me of. ) With a multi-processor configuration, a ControlLogix PLC system can easily be used as a parallel processing system with shared IO with ControlNet as the data highway for info transferring.

Mike Johnson

PS. I really like the idea of the user-defined data structure.

Kevin Totherow


Keep in mind that a PLC is a logic controller and a DCS is a control system. If the actual process to be controlled has unusual characteristics or extreme needs I would write a specification for that control. ie: sample frequency, control algorithm, adaptive control, pausing control execution, changing modes of operation, interfacing, redundancy, fail safe, etc. Look for examples or make vendors present options for solving the problem.

More typical are the user needs from a system of controller, highway, HMI, and configuration. When you start trying to build a documented, maintainable DCS from components you really get a respect for the old, expensive dinosaurs. Specify what is most important to making the company money. Redundancy, tuning, on-line control changes, configuration change management, HMI updates, HMI graphics needed - point detail displays in particular, interfaces with other systems such as historians, models, etc. Again, give the specification to vendors to see how they would accomplish.

For years we installed PLCs under the DCS to do motor and discrete control and pass the stop/start and status back to the DCS for a few process interlocks but mostly for HMI display and control. The discrete was in the PLCs because it gave the electricians a familiar device to modify, and troubleshoot electrical problems. Electricians did not have passwords to DCS configuration. It was also cheaper to do the logic control in the PLC. Either PLC or DCS could do the logic fine - but the system utilization made it better to use the PLC for tradition logic IN THAT PLANT. I would never assume all others have the same situation. Have a specification written to system needs and the PLC/DCS situation will get clearer.

Kevin Totherow
What is a PLC and What Is a DCS? is there a difference? There was once a concept called OCS in case anyone remembers it! It was basically a open DCS.

Though several textbooks try to tell us what is a PLC and what is a DCS, as time progresses, they seem to still leave some doubts. Once a vendor tried to push some standalone multiloop blind controllers and a good SCADA package as a DCS!!! Luckily, this novice engineer of those times, had a 30 year experience HOD to explain and guide.

The system uses multiple processors, has a central database and the functionality is distributed. That is the controller sub system performs the control functions, the history node connects the data, the IMS node gives reports, the operator station gives a good HMI, the engineering station alows engineering changes to be made.
The system has Processor and I/O's and some functional units like basic modules, communication modules and so on. Uses a SCADA for visualization. generally the SCADA does not use a central database and so if your clocks on two PC's are not synchronised, you will find that in one PC your plant tripped at 10:00 AM and at 10:02 AM in the other.

In the good old days life was simple we had Stand alone controllers or DCS for closed loop controls We had Relays and PLC's for Interlocking, shutdown and sequencing.

DCS had blocks, several stations spread on a lan and IEEE802.X was sacred and meant open system.
PLC had ladder and Statement list (from Germany) and SFC and the relative merits of these were discussed. Lesser beings had mnemonics and hand held programmers while the cherished ones did it through a PC. Most
DCS packages had their standard PLC's (like TDC-3000 and IP 620-35 combo).

Prices were like Macdonalds For veg burger (say 40 rupees) and French fries (say 7 rupees or even lesser) similar to DCS and PLC's respectively. Samaj gaye kya?(Gotcha!)

Speed of discrete operations were high. Micro seconds to milliseconds for individual operations, total scan time in a few milliseconds.
Speed of discrete operations was comparatively lower. I recollect that 250ms was considered good enough for most control applications.

DCS "Glory-days"
16 closed loops sometimes used 1:1 redundant DCS system costing millions of rupees (divided by 17 for dollars in those days). With Separate Colour Printers on the two Operating stations so that if one failed you could take history trend printout on the other!
PLC in those days
Contended with the Discrete logics.PLC knowledge was OK, Knowing DCS was
Gives lots of goodies for advanced controls. Foxboro, I hear has got Steam tables included as are several chemical properties.
Generally do not have a totalizer block and you end up with writing logic for totalizer. New PLC's have PID blocks (back in 90 I saw a PLC 2/05 where the PID was a mathematical subroutine). Even today I am not totally satisfied with the PID functionality of PLC's. Autotune is awaited in most PLC's.

The enginering is strong and you can write one database for a range of
controllers in a plant.
Making two PLC's to share data requires some special programming tools. The orientation of the engineering is still on PLC and its I/O basis rather than a central database.

Has a built in 3gl/4gl programming suite. Right from TCL in MOD-300 to
C++ and Basic clones in TDC-3000 or MicroXL.
programming limited generally to Ladder or STL or SFC. Though the
present standard provides for 6 languages, not all, if any, PLC's come
with a bundled 3GL/4GL support.

Moved faster to Enterprise connectivity.
Seems to be going in that direction.

End of PLC /DCS?
One vendor has moved rapidly to SCADA.

Will PLC-SCADA replace DCS, seems imminent in smaller plants but long away for bigger plants which require more functionality. DCS need not die, but can become open, available at affordable prices.

Or will PC based Systems replace PLC's and DCS. What do you do when the screen turns blue??

As time goes by the systems come closer and closer and in some time we may have the PDPCS or Programmable Distributed Process Control System as these DCS and PLC combines would probably be called along with the improvements that will come with PC based controls and Distributed Data Acquisition Systems.

Once again PLC or DCS or PC based or DDAS what does one go for?
Look for:
Functional blocks available.
future proof systems.
Open systems where you are able to connect to other systems in the
future and protocols are in the open arena.


P.S. I have no favourites between systems as their use is dictated by the

Leonard Hamilton

My primary background includes programming A-B and GE PLC's, and installation, upgrade, maintenance, and configuration of Honeywell DCS.
I will address four points:

1) Speed -- Most PLC's update their outputs in 10 to 100 msec.. In general you would be wasting the power of a DCS processor if you tried to cycle eveything at even 250 msec..

2) Connectivity -- PLC manufacturers have always been more interested in connecting to other products because they are so often used as SCADA systems. I know of few examples of a full DCS, such as Honeywell or Rosemount, making it easy to connect to stand-alone systems such as analyzers, flow computers,etc.. However, when it comes to security, most DCS are pretty much bulletproof.

3) Economics -- For many applications PLC's can provide tremenduous functionality with low-priced hardware. PLC hardware is generally much easier to configure for straightforward tasks. Rosemount Delta V appears to be much more flexible than Honeywell for small applications. The "curse" of both PLC's and DCS is that most good ones are nearly indestructible

4) Useage -- I think this the greatest area of difference. The PLC, so far, seems unable to match the power and integration of a dedicated DCS for large applications. PLC's can implement very complicated control schemes. However, viewing the algorithims and changing them usually requires much more time and prior knowledge of the PLC software than does the DCS. Even with a simple PID control. Screens for configuration and maintenance are generally built into a DCS system. Which also adds to their size and cost.

Most of the users I have met do not seem to mind that neither DCS nor PLC can do everything. EXCEPT those senior Control Engineers who don't know PLC's and are not comfortable with delegating tasks.

I hate to answer a question with a question, but, do yoy understand the term DCS? The term DCS stands for Distributed Control System. The key word being "Distributed". PLC's are simply the building blocks of a DCS. Every DCS has to start with some form of control device and most times that device is the PLC. When a process being controlled is very large, the PLC's are networked together forming the Disbributed Control system.
I hope this helps.
I would have to agree with rajesh. I have just completed a $500,000 wastewater treatment plant upgrade that now incoporates the "DCS" structure.
As Rajesh explains the operative word is "Distributed". The DCS is made up of numerous PLC's throughout the plant connected to workstations and each other via switched Ethernet. Process control of the plant is "distributed" to each PLC.

In short the system is distributed control because should the network 'crash' you can still control the plant at each control panel in that section of the plant.

I hope that makes sense!