Case History 161: Differences between PLCs and DCSs – a discussion from a practical control point of view
July 2018, Motion Control & Drives
Eliakim Musikiti read one of my articles in SA Instrumentation and Control and emailed me to ask what the differences are between PLCs and DCSs. I started replying to him and then thought that it would make an interesting article, as I am often asked why I always prefer working in plants which run their analog feedback control systems on a DCS rather than on a PLC. To understand this, one must go back into the history of the development of these devices.
A friend of mine recently commented that he was born BBC. Translation: ‘He was born before computers’. Feedback control, as developed by mankind and used in plants for almost a century before the digital computer came on the scene, is also BBC. It was developed into an engineering science by several brilliant mathematicians in the early part of the 20th century, and really came into general plant usage with the development of pneumatic technology. The technology then remained pretty much static until the mid-1960s when electronic instrumentation and control systems came into commercial use, and took over from the old and relatively crude pneumatic equipment.
Enter the electronic era
The electronic era was a marked leap forward for control equipment, as electronics is far more accurate and much easier to configure.
At this stage there were two departments dealing with controls in most plants, namely C&I (control and instrumentation), and electrical. The former dealt with the feedback control systems, and the latter with on/off controls like motor starters, etc. and with sequencing and interlocking of various machines and drives, which was accomplished by the use of devices like electro-mechanical relays, and special rotary switches when it came to sequencing. Electro-mechanical timers were also employed.
For simplicity, and before going further, I would like to define the two different systems as ‘analog control systems’ for feedback and feedforward type controls, and ‘digital control systems’ for on/off and sequencing type controls. Obviously today, everything is done digitally in computers, and this makes things much easier.
Digital computers came into commercial use in the 1960s, but it was really only in the 1970s that they started moving into the two control fields. It should be noted at this stage that the computers being used for control systems were heavily ruggedised and had extremely high reliability, as plant operation was completely reliant on their operation. This is still the same today with the exception of scada systems, which will be discussed later.
Old control panels at the St Helena Gold Mine.
Analog control field
The first major development in digital computing control systems occurred in 1975 when Honeywell and Yokogawa introduced their DCS (distributed control system) platforms to the market. Most of the major C&I manufacturers followed suit fairly quickly.
The important point of these DCSs is that they were, and still are, manufactured by companies with many years, if not decades, of experience in producing feedback control systems. Bearing in mind that the average human being is resistant to change, and at that stage many of the C&I practitioners and plant operators were not very conversant with computers and were quite scared of them, the DCS manufacturers tried to make the transition from the old analog to the new digital equipment as painless as possible.
Operators were used to controllers and recorders sitting in large panels and being able to read the information on them quite easily on meters or pens on the front of the instruments. The DCS manufacturers accordingly built specialised screens on computer monitors that more or less showed the same type of displays and operating controls that the operators were used to. As far as configuring control loops was concerned, the manufacturers tried to make loops configurable in a way that was similar to the old loop configuration. In addition, they built many safeguards into the software to prevent errors in programming.
Another extremely important point is that they very carefully ensured that the actual implementation of the PID controllers in the software was correct, so that controllers would operate dynamically in exactly the same way as the old analog controllers. The result of all this is that plants changing over from the old electronic and pneumatic systems did so with a minimum of problems.
Digital control field
The first computerised digital (on/off) control system was introduced by Richard (Dick) Morley, an employee of Bedford Associates, a company involved in the automotive industry. He proposed something called a modular digital controller (Modicon) to a major US car manufacturer. This was the birth of the first programmable logic controller, now commonly referred to as a PLC. There were huge advantages to this device as it eliminated the cost of the expensive electro-mechanical components, as well as the need for complex physical wiring. In particular, it was extremely easy to configure the programs, and to make changes, which previously took a lot of time and effort. Aside: in the early 1970s, and after PLCs really took off in the market, Bedford Associates changed its name to Modicon PLC.
Once again the manufacturers of PLCs tried to make the system easy for practitioners of digital control systems. The first units were programmed by ladder logic, which is almost exactly the same as the wiring drawings used for the old electro-mechanical devices.
Initial differences between PLCs and DCSs
There were no displays for the early PLCs, and the use of push-buttons, rotary switches and indicating lamps mounted on control panels continued for quite a few years.
Thus, the two control fields carried on separately for several more years, but obviously people started querying why two expensive computer systems were needed in a single plant. Surely a single computer could combine both functions, especially as computing power had grown exponentially?
The DCS manufacturers were quite quick to start incorporating ‘digital’ capabilities into their machines, but there was a problem for the PLC people. There was no ‘easy’ analog operator interface on their devices, whereas the DCSs had operator screens as an integral part of the system. Some of the PLC manufacturers offered special interfaces for PID controllers, for example a panel-mounted module with a front display, which had the capabilities for an operator to switch between as many as eight PID loops. However, these displays were expensive and not terribly user friendly.
A solution to this problem appeared in the early 1980s, when an engineer, Steve Rubin, who owned a company called Intellution in Boston, US, started wondering why people were using expensive ruggedised computers for control. When I met him he said that as far as he was concerned, the ordinary desk PC had all the power and reliability to run plant controls, and they were a fraction of the cost of the control computers being used. He then developed an extremely clever system for use on desktop PCs to operate both analog and digital controls. The best part of his system was that it was extremely easy to configure ‘screens’ to display any type of control device, for example from PID controller faceplates to alarm screens, to on/off control buttons and indicators, and he also included powerful trend and real-time recording and logging.
He introduced this system, which he called ‘The Fix’ (fully integrated control system) at an ISA show in Houston in 1984. This was the world’s first configurable PC-based HMI/scada software programme.
From the actual plant control point of view, very few people ever agreed with Rubin that desktop PCs were reliable enough to run a plant, in fact I have only ever come across two small plants where everything ran off a PC. However, the scada part took off like a rocket. Here at last was the easily programmable and configurable HMI interface the PLC makers needed to be able to introduce a full variety of analog control systems into their products. Fairly quickly the PLC manufacturers also started offering a full menu of both analog and digital controls in their machines.
A modern digital control room.
So, from the analog control point of view, what are the differences between the two systems?
The thing to note is that the DCS manufacturers had a background of decades in making PID controllers and ensured that they could be configured and operate correctly. The PLC people on the other hand had very little if any experience and knowledge in this complex field and they put controllers into their PLCs that on the whole caused all sorts of problems. My own experience, over many years of working in all types of plants, is that 85% of the time when I go into a plant using PLCs for control, their controllers are set up incorrectly. Therefore, whenever I am called into a plant using PLCs for control, the first thing I do is to test the controllers to see if they are working correctly – before even looking at loops.
The cost to plants with control systems working incorrectly can be unbelievably high. My most classic example is of a pharmaceutical plant in the UK that was losing an estimated £14 000 000 per year all because they had not set up a PID controller correctly. I have come across three plants in South Africa running for many years on PLCs with the PID controllers not really working at all, so that actual operation was done in manual control. In most PLC plants the controls are causing many problems because they are not set up correctly. Many plant people do not really understand their control systems properly and often think their problems are caused by poor tuning. They don’t realise the controllers are not doing what they are supposed to do. People find these things hard to believe, but I have seen all these problems first-hand.
The lack of knowledge of the analog control side, and the fact that PLC/scada systems are unbelievably wide open so people can program pretty much anything they like, has caused many serious problems. PID control programming is in fact very complex and few people have the knowledge of how to set up the normally basic PID blocks provided in PLCs so they operate correctly.
I have also found that programming personnel find it much harder changing things around in a PLC than they do in plants with DCSs. However, on a positive note, things have started to improve in recent years. Several of the big PLC manufacturers have started incorporating DCS software into their products so the controllers work correctly.
From my point of view though, I am still far happier working in plants using pure DCSs rather than PLCs.
Michael Brown is a specialist in control loop optimisation with many years of experience in process control instrumentation. His main activities are consulting, and teaching practical control loop analysis and optimisation. He gives training courses which can be held in clients’ plants, where students can have the added benefit of practising on live loops. His work takes him to plants all over South Africa and also to other countries. He can be contacted at Michael Brown Control Engineering cc, +27 (0)82 440 7790, firstname.lastname@example.org, www.controlloop.co.za
I did a loop tuning course with you when I was still working for AECI (in the 1990’s), 22 years later I’m still with Staro Process Control distributing Rockwell Automation products (Allen Bradley). My career started with DCS systems in AECI and PLCs were smaller standalone systems mostly found in MCC rooms. I am glad to see that you are still in the business.
I am very surprised with your article choosing superiority of DCS PID control systems. I believe PLC algorithms, speed of data and processing power together with the continuously developing scada systems outperform DCS systems, and wonder if it is even comparable nowadays. Many of the current ‘DCS’ systems are actually nothing other than a PLC hardware platform, and we have replaced DCS systems by keeping the Allen Bradley I/O cards (branded something else) and replaced the DCS CPU with latest Rockwell PLC CPUs. The controls in all cases outperform the previous DCS systems by far, mostly because the latest CPU technology is superior.
Speculation does not appropriate in our world as far as better is concerned, but, I wonder for how long the term DCS or PLC will still last. Nowadays, many young engineers work just with PLCs as a control system, sized according to the application. Of course, there may be small not so high-performing logic controllers on the market, but one should not generalise these with the major brand offerings.
It was good to see that you are still around and respect for your practical loop tuning mind.
Buks Pretorius, managing director, Staro Process Control.
Great to hear from you after all these years, and yes, I do remember you.
I am writing this from Cuba, where I am doing some consulting at a mine and having a difficult time convincing metallurgists that tuning cannot overcome process problems caused by faulty valves and pumps.
I agree with many of your comments about the coming together of the machines. However, you possibly haven’t worked on some of the latest DCS platforms, which have also developed out of all recognition. I still find most big continuous processing companies will always use a DCS, which is generally far more geared to feedback control systems, as opposed to ‘digital’ control such as sequencing etc. The main point I tried to make in this article is that unless PLCs have specific DCS type software in them, such as the Siemens PCS7, users still configure the PID controls incorrectly in nearly 80% of all plants. Implementing a PID control is very specialised and one needs a lot of knowledge and experience to get it right. This is why most DCS systems protect the user against bad configuration. One of the most horrific jobs I ever tackled was sorting out the controls in a paper plant that decided to go with a PLC that is still in common usage. I had to rewrite all the control blocks on a running plant!
I am sure that you are correct in the fact that in a few years the distinction will disappear and one will just buy a control system.