Hi All,

I m a DCS engineer, newly Introduced to the Gas Turbine based Power Generation Industry. I was trying to figure out the scope of DCS controls on the plant operations.

What ponders me is that, of the many plants that I've visited, all of them have separate PLC controls for systems like Fuel Gas Compressors, Instrument Air Compressors, Water Treatment Plants etc. which are later connected to the DCS control panel through Remote I/Os.

Wouldn't it be easier and economical just to integrate the automation of all these systems directly into the DCS rather than using a PLC - DCS approach?

Please advice.

 

Hi George,

Technically, especially with the capability of hardware today, I think it is totally possible.

But it wasn't so back then. PLCs or dedicated control system has significantly smaller scope and specific control function compared to DCS, so they are equipped and programmed with much faster IO scanning and control sequencing program execution cycles. For the DCS, as it might be connected to a myriad of control systems or RTUs, that might have different IO scanning cycles, it needs to synchronise the signals from all sides, and so it would be much slower compared to a dedicated PLC or control system.

But, that is history. And due to this history many vendors of dedicated control systems or PLCs try to protect their bread and butter...you know what I am trying to say.

Technically, we know the truth is different now.

Totally my opinion.

Cheers,
Kevin

 

George,

I agree with Kevin that it's totally possible these days, and was more difficult even just 10 years ago.

However, I would definitely NOT want to be the person in charge of getting all of the manufacturers of the various pieces of equipment to submit their control requirements (hardware and software) to a control system integrator or DCS vendor for incorporation into a single control system.

Most vendors have invested a significant amount of time and effort (which translates to money) in getting their equipment to work using a particular control system (PLC; PAC) in the manner they want it to work. And, most vendors are NOT going to want to provide their control schemes (some considering them to be proprietary) to a control system integrator or DCS vendor and have to deal with questions which will certainly arise ("Do we have to do it this way?" or "Can we substitute this or that?" or "Can you provide more details on how this works or should work?").

And, I would definitely not want to be the commissioning person for equipment that was being controlled by a DCS when it's always been controlled by a local PLC/PAC. It would involve a lot of time and effort if there were any issues (and there would likely be some issues!).

And what about warranty concerns? YIKES!!!

While it's certainly possible, and has been done on some smaller plants (with varying degrees of success), and on some larger plants (with a lot of schedule delays during engineering and commissioning/start-up), it is really difficult to implement and oversee. And, the warranty issues which might arise if the equipment is damaged or broken and there is any possibliity that it could be attributed to programming/configuration of the DCS would mean the problems could continue for a LONG time.

Plant owners and operators salivate and complain loudly that this is what they want--a single control system AND, more importantly, a single operator interface for all the equipment in the plant. In my personal opinion a powerful, flexible and programmable HMI application which can communicate natively with many different controllers/control systems is about the best compromise for this problem. And there are a lot of HMI applications that claim to have native drivers, but don't really have very good ones.

Fortunately, OPC and OPC DA and various versions of this are helping to allow control systems of all kinds to speak to each other or to an HMI application, or even a DCS, to make things easier for operators.

But, it doesn't solve the problem of multiple control systems--and the reality is, that's going to remain for a very long time. It's just very expensive (in terms of engineering and start-up/commissioning and warranty considerations) to try to control all the equipment in a power plant with a single DCS.

And, purpose-built control systems (for turbine control and protection, for example) are much better for some applications (primarily because of the various unique types of I/O which many DCS systems haven't been engineered or designed to work with).

 

Thanks Kevin and CSA for the reply,

Why this question arises in the first place is that many of these PLC based sub-systems are directly constructed by the same company in-charge of the DCS. And any how, all analogue and digital signals including control commands are interfaced wit the DCS through Remote I/Os.

As of right now for a single block in a combined cycle / co-generation power plant we require to interface approximately 1000-2500 Inputs and Outputs. We keep about a 5-10% spare slots on each remote I/O module for any small future additions. And further I/O modules can be added for additional blocks. All these sub-systems corresponds to just 50-100 I/O slots. Connected to the PLC or not many a value that are read from these sub-systems are calibrated.
So why not just add it to a particular Remote I/O module and make a sub routine for that sub-plant. Anyhow the DCS controls the system starting and stopping.

I do understand CSAs' concerns of a vendor not agreeing to give Guarantee on such systems. But wont it be possible if no vendors are involved.

I've also heard the fuss of the DCS programmers / Controllers having to call the vendor now and then to make small tweaks on the program, and it causing great delays (Mechanical system tweaks are OK though - It will require the vendor as it may be in the Vendors' scope).

Even in DCS programs we work one system at a time in sub-routines rite.

Food for thought...

 

George,

I don't understand what you're trying to describe at all; sorry. One company might be in charge of procuring and supplying all the equipment required for a power plant, but "...directly constructing..." all of the equipment? That would seem to be very expensive, especially if they are designing all of the equipment, and procuring all the necessary parts for the equipment, and then manufacturing, testing and shipping all of the equipment.

It would certainly seem they would then incorporate all of the equipment into a single control system (DCS). And you say commands and analog signals are routed through the DCS.

This might be happening because different divisions of the DCS company have developed their own control schemes using different control systems and the parent company hasn't exerted enough persuasion to coerce all of the divisions to use the DCS equipment for control. Does the DCS vendor not make smaller control systems/modules which use the same programming/configuration software?

There may be a lot of reasons why the "parent" company isn't controlling all of the power plant equipment and auxiliaries. I submit, again, that it's just not practical for the same company to engineer, design, produce, test and ship all--or even most--of the equipment used in a power plant. That company may be acting as the EPC (Engineer/Procurement Company) for the plant, but based on the Vendors you checked in the post it would seem that the EPC isn't manufacturing the turbines, or the switchgear, or a lot of the equipment/auxiliaries.

Various signals (operating data/parameters) and commands may be done via the DCS, but if the EPC is also "directly constructing" all or most of the power plant equipment then it would certainly seem they would also want to control that equipment with the central DCS.

What you describe (using a single control system, the DCS, to control all of the equipment and auxiliaries in a power plant) is, indeed, the ideal. We can't know all the particulars of why things were done the way they were done at the plant where you are currently working; only you can discover these for yourself. But, ideals are something that are rarely achieved, and everyone's ideal isn't always the same as everyone else's.

Great idea, though. This is how lots of companies get started--with an ideal.

The implementation, however, is, often, quite another thing. One of the most important lessons I have learned in my engineering career is: Engineering is a series of compromises. And, the longer you work in this business the more you are going to realize just how true this is. Decisions made today, without consideration for months or even years or even decades later, can have huge impacts on what can be done tomorrow, next year or next decade--requiring compromises to be made.

 

One more reason to keep the subsystem control with the subsystem supplier: This also lets the subsystem be factory tested with its own control. Any problems with either the control or the subsystem are much more economically corrected at the factory than on site.

 

Thank you CSA for your view,

I'm sure that it may take some time as well as expertise to achieve such a goal.

Mainly was a bit frustrated about sub-plants build with less consideration to the overall control of plants.

Thank you for the discussion.

 

> One more reason to keep the subsystem control with the subsystem supplier:
> This also lets the subsystem be factory tested with its own control. Any
> problems with either the control or the subsystem are much more economically
> corrected at the factory than on site.

I think this was the best answer,

DCS stands for
Distributed Control System
How come they always like to centralize it?

Roy

 

Roy,

> How come they always like to centralize it?

This one's easy--DCS stands for:

Distributed
Centralized
System

I've never personally understood the whole 'distributed' thing. Newer control systems, and some technologically advanced older ones, do have the capability to have I/O modules, and even controller(s), remotely located from the "central" controller(s) so I've always thought that's what was meant by 'distributed.'

My first exposure to a large power plant (having worked initially at small gas turbine peaking plants) was a 400 MW coal-fired steam plant. And there was a LOT of PLCs (as they were known at the time) controlling a lot of auxiliary equipment, but there was a large control system used for the boiler and coal feeder and -pulverizer and basic steam turbine commands (the steam turbine control/governor was still a purpose-built control system provided by the manufacturer). I, too, was amazed at the whole idea of having a lot of different types/manufacturers of smaller PLCs that were loosely connected to the large plant control system, usually just by a small number of analog and discrete signals (some used MODBUS).

I also feel otised's answer was excellent, but the original poster indicates all of the equipment is provided by the DCS vendor, which just seems unusual unless there is some kind of owner's requirement that all equipment be warranted/guaranteed by one manufacturer (which is something else all owners and operators desire). Some Asian EPCs/contractors have tried this in the past, again, with mixed results. Some of the pieces of equipment have just been minimal and barely functioned which seemed to stem from a basic lack of understanding of required functionality or operation.

Again, the ideal expressed by the original poster is extremely difficult to obtain--laudable, but fraught with potential pitfalls and difficulties. In today's business world where companies want to stick to their core competencies and outsource or source auxiliary equipment from other companies (who are sticking to their core competencies) it's just very hard to get everyone to "share" their control schemes and requirements with any other company, and when it's been tried (in my experience) it hasn't worked very well. Not well at all.

A major multi-national combustion turbine manufacturer has even developed their purpose-built control system to be a DCS/BOP control system, primarily because of their advanced gas/steam turbine designs requiring steam cooling. They don't want to have to deal with several different boiler manufacturers who use various controllers to provide critical steam flows to the turbine (more for cooling that for power generation). They just want to be able to provide a single control system for the plant that is intended to be a "one-button" start (which is yet another thing plant owners and operators strongly desire--and which is much easier with a Distributed Centralized control System (DCS)... But requires a LOT of plant experience, and auxiliary experience, to integrate into a single control system (centralized or not).

 

The original definition was "Distributed Control System!

Phil

 

Roy,

Yes its kind of funny that it calls itself distributed.
Well, as you may know already, this niche requires a Centralized control and operation strategy to run the Plant. The operator has to make decisions comparing data on screen and are critical for the smooth Operation of the Plant.

CSA,

You said it (right). I am part of such an Asian company trying to centralize controls and Guarantee. Being part of the contracting company we've been providing sub contracts to companies in their respective niche for the corresponding parts of the Plant.

According to the owners requirement(and common sense to me), the operator must be able to completely control the plant from the control room. Here all data and actions are logged and are closely monitored (more like a space program control room). Like I mentioned before, many a time you get only partial control of the sub- plants through limited I/O's. and changing these sub-plant control programs according to updated plant requirements is a hectic task ( Need to call the vendor, need to get their head office's approval, need to wait for the PLC engineer to come and analyze and then calibration). This takes a long time, which could be rather solved within hours and less cost and more reliability. ( Any contractors/ Owners dream).

You are right again on the talent part. Its hard to get expertise in all areas of a plants working. Its a similar case for vendors as well. Most of them specialized in the Mechanical and Instrumentation systems and very few in the PLC or controls part. Whereas the plant already will have an experienced DCS engineer with great to do the task. That is one of the major concerns I had when starting this post.

I understand the importance of checking with the local control systems before each vendor presents their product at site. To my astonishment this activity never happens. I affirm never. The mechanical systems would be first installed by vendors, then they design the controls hardware according to P&ID and then a PLC programmer designs the controls and commissions. Lab testing is done by specialized tools.. always.

Its only for the critical systems like GTs / Generators that a control hardware is designed , tested and brought from the main factory along with the main equipment.

IT should be only the concern of guarantee by vendors that prevents a centralized control in such cases.

Happy to know it was food for though for you all.

Thanks again for the replies...

 

> The original definition was "Distributed Control System<b>!</b>

Chill, Phil.

It was a play on an abbreviation, since they are distributed AND centralized control systems these days. And they were when the original DCSs were released, too. It's almost a misnomer--really just a marketing ploy to try to distinguish one system from another.

 

Sorry about the typos (was sleepy)..

 

It's always been a bit of wishful thinking on the part of the marketing guys ...

When I started out (mumbledy-mumble years ago) control was done in some places by a man and a panel out on the plant in the noise and dirt. He communicated by telephone with a centralized control room who made the strategic decisions.

Then plants were built with pneumatic controls, with many many runs of small-bore pneumatic tubing connecting back to a central control room. This strategy was copied when the first analogue electronic control systems became available. However, electronics were 'easy' to extend compared to pneumatics and the number of connections started to skyrocket. The initial digital offerings also had a single central computer (anyone remember the PDP8-based systems of the early '70s?) in an air-conditioned room with well-filtered and secured power supplies.

Mid-70's and the microprocessor started to make its presence felt. Some big firms wanted to get into the market and decided that offering a control system built up from a large number of small number of microprocessor-based modules could be sold on the grounds of (a) reliability (a number of small modules made redundancy an attractive option, compared with the difficulty and expense of making a single CPU-based system redundant)
and
(b) the horrific number of connections between the field and the control room could be reduced by sticking small racks of equipment close to the operating areas of the plant, rather than centralising them.

The first advantage was real and very attractive to end-users: the second one never really got any traction (possibly because the end-users expected relatively poor reliability and weren't comfortable with having to do maintenance in the field, or because the techs working on the computer based systems had got used to a certain degree of air-conditioned comfort and clean workspaces, and didn't fancy having to go out into the real world again).

So you can interpret the "distributed" part of DCS as relating to the "distribution of control functions over a large number of semi-intelligent modules, instead of being centralized in a single item of equipment, or referring to the concept that control equipment would be "distributed" physically around the plant rather than in a single location. Take your pick.

Bruce.

 

George,

I think many of us would take exception at some of your statements about testing before shipment. For auxiliary equipments which are "mass-"produced in some quantity (say, four or ten per month) which have very little variability (say cooling water modules, or filter skids) there's little reason to test a known working design before every module or skid ships. (The exception to this is when a new manufacturer has been chosen to make a piece of equipment that is usually made somewhere else in the world--then, YES! It's incumbent on the company which is outsourcing the production to check for proper operation. Even on the simplest of control schemes (a couple of differential pressure switches and a couple of vibration switches) mistakes leading to blown fuses and worse can occur if the wiring isn't checked on the first few skids--at a minimum--from a new producer, even using designs supplied by the original equipment manufacturer.)

Many companies are trying to reduce costs by eliminating factory testing for smaller auxiliary pieces of equipment, preferring instead to have the commissioning personnel (rarely engineers these days) do the initial check-out and commissioning since they have to send people to site. Sometimes, the commissioning personnel are experienced and knowledgeable but these days many are just contract personnel and are not really familiar with a particular piece of equipment even though they might understand the concept involved.

A major multi-national manufacturer of heavy duty gas turbines puts out a specification for various auxiliary pieces of equipment, and companies "bid" on the production of these pieces of equipment, and the lowest bidder wins. About the only quality checks that are done by the manufacturer is that they receive a written document from the producer that the equipment meets the specification. Rarely does the manufacturer send a quality control person to the factory to observe testing and confirm suitability and quality before approving shipment of the first modules/skids. Rather, the problems become apparent at site when the modules/skids arrive and are being installed/commissioned and then there's a lot of finger-pointing about who's to blame. Which only make the commissioning process longer and leaves a bad taste in everyone's mouth. Yes, the problems eventually get "solved" but at a cost in man-hours, aggravation, frustration and schedule over-runs.

I would argue that it's very possible for an operator sitting on his/her arse in an air conditioned control room to operate an entire plant with multiple control systems for various pieces of equipment in the plant. It all depends on how well the various pieces of equipment were integrated into the "central" control system, or more correctly, the HMI the operators use to control the plant.

Would it be easier for the maintenance technicians to only have to know and understand a single control system to keep the plant running? Yes, indeed. But, as you've noted--it's very difficult to have and maintain an engineering design and procurement organization that's capable of specifying, designing, integrating them into a common control system, and manufacturing and testing pieces of auxiliary equipment--there's just too many auxiliary pieces of equipment. And, there's usually some bean-counter who will checking in-house costs versus purchased costs and can show that it can be done cheaper (though not necessarily better!) using outside suppliers--though they rarely mention that the outside suppliers will be using their own control systems!

Ideals, again, have been the impetus for many an entrepreneur and new companies. But, the devil is in the details. Even many managers and vice-presidents of large corporations have the same ideals--but leave the details to underlings and don't take the time to try to understand all of the intricacies and details. They sell their vision and then can become quite upset when they didn't explain their vision to their reports, or when their vision can't be implemented in the way they envisioned--because they were using rose-colored glasses which prevented them from seeing all of the potential pitfalls and problems.

But, this is the world. Ideals are great, but bean-counters usually win. And they have little accountability.

 

ha ha...Well Said CSA,

What i expressed was an engineers headache (you might be too familiar with it). I always appreciate sub-systems purchased with a proper control system from the original company (If they have one). It normally can transfer all functionality controls to the DCS interfaced with It. And is much hassle free.

I'm more worried about the latter majority (as you may have understood).

Well these are the challenges we have to face in the industry. Smaller and faster centralized controls have taken over the plant. Operators lose direct interaction with these systems over time..which is also a threat for the future.

Intelligent systems replace operators at many a plant sub-system. Growing more towards a One man control. Having its own advantages and disadvantages.

And we are faced with the great task of re-designing and incorporating these properties to the industry. Keeping us up to date with proven technology.Some challenge.......