Sign in to follow this  
Followers 0
Greg

DCS vs PLC

9 posts in this topic

Hi to all, Could any one explain me the exact difference between the DCS Controllers and PLC Controllers. at every point of view, their performance, the advantages, their disadvantages, when should i use PLC and when should i use DCS controllers. i have been working with them without knowing the difference, and i am quite to be ashamed of this situation. i have asked some people at work, seems to be they not that specialists as i thought. all answers will be excepted with pleasure and appreciation. Thanks in advance, Best Regards, Greg.

Share this post


Link to post
Share on other sites
Hi Greg This is more of a general question than an "Other PLC" topic so I'm going to move it to the "General Topics" category and see if we can get you some answers

Share this post


Link to post
Share on other sites
Literally: A ProrammableLogicController is a single piece of hardware with I/O attached. A DistributedControlSystem is a network of several controllers. However these lines have grayed in recent years.

Share this post


Link to post
Share on other sites
In this day and age - absolutely nothing. DCS is a hangover from the old days (10 years + ago) when PLCs were rather simple limited systems. Limited in the amount of I/O they could attach and the networking capability. Today most mainstream PLCs can do everything a DCS could do. About the only real difference is that a DCS was really a PLC and SCADA supplied as an integrated system. DCS engineers would have you believe otherwise - if you asked them to explain and they were truthful - it would sound just like a modern PLC. DCS = Distributed Control System. PLC = Programmable Logic Controller. A DCS contains PLCs - A PLC can be a DCS (B&R being a good example). PLCs were considered a single controller with local I/O only. A DCS was generally multiple PLCs connected via a network and most often included SCADA. When I refer to PLCs I am talking about modern systems that use the full IEC language set, not those stuck on Ladder Diagram.

Share this post


Link to post
Share on other sites
Hi to all, Thanks for all, it helped me to understand the difference between that 2 systems. Best Regards, Greg.

Share this post


Link to post
Share on other sites
this one is about 7 years old - and as it says upfront, "there are exceptions to the rules" ... still, it's helped quite a few people over the years with the basic questions that you're asking ... http://www.plctalk.net/qanda/showthread.php?p=12395&postcount=10

Share this post


Link to post
Share on other sites
At one time, the controls world was split up into 4 distinct "regions": 1. VLSI 2. FPGA 3. Embedded controls. 4. Motion controls. 5. DCS (distributed control system) 6. PLC's. Today, the bottom two have effectively collapsed into one another and there are efforts afoot to gobble up #4. #1 has almost faded away except with very high volume production, and often #2 has taken over a lot of #3. Using the traditional model, this is why you'd use them: 1. VLSI. Whenever cost (at high volumes) or speed is the overriding factor above all else. 2. FPGA. When you need the speed of VLSI but not the volumes. Basically, this is "design your own chips" territory. 3. (embedded controls). When you have very specialized hardware or need high speed data processing for things like vision or DSP work. Also often used when you have an utterly simple controller integrated into a small control screen to run something like a kiosk or a washing machine. Response times are usually microseconds. Also very helpful when doing PC integration since often the target IS a PC. 4. (motion controls). When you need to run servo's. Response times are microseconds. 5. (PLC's). Running batch or assembly line type stuff (lots of start/stop controls). Response times are tens of microseconds. 6. (DCS). When you need hundreds or thousands of IO points and especially for doing analog control loops. Response times are hundreds of milliseconds. At one time you'd implement your system in this order...often a DCS was the "master controller" for the whole plant and would either interface directly to IO or use PLC's as "slaves" to it, just as you'd put in a motion controller to run a servo or a "robot" type device, with a PLC running the overall process. With a DCS system, there'd be one overall plant-wide DCS with individual PLC's in each plant unit operation or building. However, things have changed. The "PLC" has become the "PAC", which takes over most if not all of the functionality of the DCS. There's also a distinct way in which you pay for and do development. With a PLC, you pay once for the software and pay for individually controllers as needed. Anyone who has a software license can do development and system integrators are encouraged. Frequently the system is split into layers with the hardware, HMI, historian, and controls all taking distinct layers. Each layer can be implemented with different hardware/software, and you can even omit layers without any severe repercussions. Recently there have been motion controllers implemented inside PAC controllers as well which means that the remaining hardware in the motion controller is just the amplifier...the separate controller is gone. Systems are now operating at the millisecond/microsecond border. With a DCS, they take great strides to keep the system integrators out. You license the development licenses per "platform" and each site is a "platform", so having a license at one site is useless at another. The systems have gotten faster, down to tens of milliseconds. There are usually annual license fees and you pay "per point" (basically, per IO device) rather than pay once for the whole system. The HMI, historian, and control system are integrated together. "Small" systems are mid to high 6 digit price tags. Large ones are 7 digits. Supposedly as you scale up the costs for DCS systems are less overall than a comparable PLC (PAC) system, but so far in every model I've run, I haven't found where this point is yet. DCS systems do not play well with others or even with themselves. Expect your system to be an "empire" in it's own. Whatever capabilities it comes with are all that it will ever have. You can't add on extra modules later very easily (say for a bigger historian or on newer hardware), and getting it to interface to anything that the manufacturer doesn't specify (aside from IO devices themselves) is an exercise in frustration equivalent to getting Windows to play nicely with anything else. There are major differences in terms of reliability. A DCS is based these days on commodity PC hardware (there are exceptions). So it tends to be about as reliable as your typical server PC (not very). Thus, redundancy with a DCS system especially for the servers is an absolute must. With PLC/PAC hardware, even the original designs were intended for an operating life of about 20-30 years. With the newer safety-PLC designs, this has been pushed out to centuries...basically the limitation on the PAC/PLC is now external factors like support from the manufacturer, and redundancy is usually not used at least at the PLC/PAC level because the controller is no longer the weakest link. This drastically simplifies the overall architectural design, compared to the "herd of servers" issues that you have with DCS systems. Another advantage is that since the PLC itself is relatively low cost (on the order of a single DCS server or less), they tend to be much more numerous and the overall system is a little more robust because it is cost effective to have one PLC per unit operation whereas with a DCS installation, it is extremely rare to install more than one. Perhaps the one holdout for DCS systems over PLC's that is still very nice is that a DCS is designed such that once you start it up, you never shut it down. You can reprogram parts or the whole system in online programming at any time. PLC's are getting CLOSER to that ideal but there are still lots of little gotchas and limitations...we're not quite there yet. These days, if you have a large centralized plant-wide HMI, it tends to do all the things that the DCS formerly did. You can use it to monitor and control any part of the operation from a centralized control room. The PLC's do the "control" functions while the operators "live" inside the HMI. Of course this is one model of how to do things. With separate pieces (HMI, historian, PLC, hardware), you can pretty much mix and match to achieve any particular level of centralization/decentralization you want. There's a 7th category that takes somewhat from DCS and somewhat from PLC's that I'll just call a "true DCS". Traditional DCS systems still have a centralized control system but it tends to be a network of redundant PC servers. The "true DCS" that I'm speaking of comes in other flavors, but the most popular one is Foundation Fieldbus. In this system, EVERY device (IO or otherwise) has a set of programmable blocks. You load the programming into these blocks. There is no central "controller" like with a PLC or DCS system. For instance, you can configure the programming in a valve to run a PID loop where the valve is adjusting to a temperature or flow meter reading. The valve directly accesses the flow meter output, without any separate controller. Speed in this system is hard to describe. The underlying protocol is a relatively slow serial communication protocol. But since you don't need every single IO message to go to and from a relatively "centralized" controller, speed isn't usually all that important. So I'd still say it responds in 10's of milliseconds, but it can definitely be slower at times, especially when you make configuration changes.

Share this post


Link to post
Share on other sites
+1 A few years ago I had to visit a plant of ours that was in the process of deciding how to implement a control system upgrade. The purpose of my visit was to give the plant an experienced assessment of PLC based control. One camp (ours) was in favor of a multi-processor PLC based system on a fast ethernet backbone and the other camp was in favor of a DCS system (only because they "knew" that "big plants" had DCS systems in them). There was a significant cost difference between the two (the DCS was much more expensive even though the IO count was the same). When my part of the presentation started I went to the white board and started drawing a networked control system. When I was through I asked them to tell me what I had drawn. Every person in the room stated that it was obviously a DCS. I had drawn the network diagram of a control system that we had implemented a year earlier. It was PLC based. 10 processors in all (my "standard", if you want to call it that, says that if it rotates (compressors), it gets it's own processor, if it has a fire in it (fired heater or boiler), it gets it's own processor, and if it controls any form of energy, it gets it's own processor (PID loops, dehy sequence, and motor control are typically assigned one processor though sometimes motor control gets it's own) with 5 independent Wonderware nodes. Inter-processor communication has to be considered as there are usually numerous instances where a bit or word of data is in one processor and it's also needed in another. If it's critical information (ESD) we hard wire the points between each rack. Otherwise we do simple data transfers using the ethernet (or ControlNet) backbone. The point is that as has already been said- there is no appreciable difference in systems under the size of 1,500 - 2,000 I/O except cost. Edited by Michael Lloyd

Share this post


Link to post
Share on other sites
PLC's scale just as much as DCS systems do. If you are doing large systems, the initial fixed cost gets amortized out but the cost per IO point (including that nasty annual license fee PER POINT) still exists, so this is not a scalability issue at all. The major differences are: 1. Typically PLC's have a separation of concerns. The control, operator interface, programming, data collection, and reporting systems are all separate "layers" in a large system. In a DCS, they are all essentially the same layer. The separation of concerns idea would seem to be a disadvantage but in fact it has been shown to be better in practice. In fact in the programming world I would point to "Visual BASIC" as the pinnacle of "integrated systems". Modern programming environments and systems are now actually returning the the idea of breaking out everything into distinct sections again. The only equivalent PLC system that I'm aware of is the GE Fanuc "Cimplicity ME" HMI/PLC environment. Even then, it's not quite the same. 2. Pricing. With a PLC, you pay once for the hardware and once for the software. You might pay a "support" contract which is either based on the software or the number of PLC's but this is optional. With a DCS, you pay up front as well as a an annual charge per point (not per device or per seat) which is nonoptional. And the pricing...well, even the most expensive PLC systems with all components included are still half the price or more of a stripped down, very limited DCS system. 3. Support. With a PLC, you typically use a local "system integrator" if you don't develop in house, and the business is highly competitive. With a DCS, the vendor takes steps to shut out the system integrators. You pay a much higher charge to the vendor and all outside development comes only from the vendor. In fact they are so nasty about this that even a lot of the initial setup can only be done by the vendor. This is one of the reasons that DCS systems are outrageously expensive. 4. Reliability. DCS systems pride themselves on heavy use of redundancy to avoid system failures because they use cheap PC hardware. Hardware is designed with the PC life span in mind (3-5 years). PLC systems are designed for rugged reliability over a period of decades or more, so although redundancy (other than at the IO level) is sometimes available, typically it is not necessary or used. 5. Online programming. PLC's are getting a lot closer but DCS systems are designed from the ground up to be configurable online without ever shutting the system down. On the other hand since you would typically shut down a process unit for a major change, and you'd use one PLC per process unit, this point might be a bit mute. With a PLC system, you don't need the level of online programming that you get with a DCS, but it sure would be nice sometimes. 6. Interoperability. A DCS is an island. PLC's work with each other, with different IO platforms, etc.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0