Hi guys!

I am a control engineer of a future thermal plant for research purposes, containing boiler and flue gases analysis. I have to decide the whole DCS system.

What configuration do you think fits better in the process: having the controllers distributed next to the I/O in the cabinets on the platform, or only having the I/Os next to the process and the CPUs next to the global control room of the plant? I mean distributed CPUs or not. I will use Profibus for field, and ind.ethernet interconnecting the CPUs.

There will be around 3000 signals and maybe 3 controllers. If speed required, I could add PLCs, right?

Thanks for you help, cause I am new in this field.

 

You should also look at the advantages of Foundation Fieldbus. That gives you the option of doing the control in a central location or right out in the field Transmitter or control valve.

Roy

 

No way. FF can only be useful in open loop. In close loops the device response is very sluggish around 31.5 kbps. In distributed fashion go for Optical port processors and s/w for ethernet like foxboro at upper level of control hierarchy and at lower end profibus is advisable.

 

You are totally wrong. There are thousands of plants using foundation fieldbus for closed loop control. Control response period can be 250 ms or even lower, even for relatively complex loops.

You should look at 31.25 kbit/s from the perspective of each PV being only 5 bytes long and that each FF bus has only a few loops. It closes the loop with time to spare.

31.25 kbit/s was selected to enable long wires from the control room to the field where the transmitter sits. FF permits 1.9 km. PROFIBUS, Modbus, or any RS-485 bus can do 12 Mbit/s, but only for 100 m. Once you want to go long distance you must slow communications down due to capacitance of copper cable.

In my personal opinion, using fiber optics for transmitters and valves is impractical.

Going back the original question, mounting conventional I/O modules or controller in the field (remote-I/O and distributed-I/O respectively) is also inconvenient. The I/O cards are mounted in field panels scattered around the plant. The I/O cards, backplane, power supply, and communication processor are, in spite of protective enclosures, exposed to environmental more stress such as heat and vibration. Maintenance such as inspection and card replacement such as a power supply becomes more difficult for technicians because the I/O panels on the platform are not as accessible and should not be opened due to sea spray, dust and moisture, water spray from washing, rain, and hazardous area etc.

Cheers,
Jonas

 

I believe that Profibus-PA (the version meant for process control) is also 31.25 kbps. Profibus-PA and Profibus-DP are two completely different things.

The original poster did not give enough information for anyone to give a useful answer.

 

To be honest, it may not make a lot of difference. As ever, the answer depends on your individual circumstances. Some things to consider ......

The advantage of fieldbus systems is that the I/O can be very remote from the controller.

What is the environment like at the I/O locations and in the proposed Control room? What are the environmental specifications for the equipment you want to put in those locations? I.e. Is the CPU rating as "good" as the I/O modules or does it need a "cleaner" environment.

Do you need easy access to the CPU's or are you happy to walk around the plant to get to them?

Will physical separation of the CPU's give you better fault tolerance if the control room burns down?

Yes, in general, if you want to make the code execute faster you put less code in more CPU's. Note however you will need to design your I/O and fieldbus layout carefully if you intend to add more CPU's later in the project. For example, you need to make sure the PV and CO of a PID loop will always be connected to the same CPU.

Personally as a programmer, unless there's a good reason, I prefer not to have to walk so far to the CPU.

Rob

www[.]lymac.co.nz

 

Thanks guys for you help !
The Profibus I mentioned was Profibus PA. So the strategy I thought is:

HMI, Servers using Ethernet, Controllers' network using Industrial Ethernet, link between controllers and remote I/Os could be Profibus DP in case I use then Profibus PA to bus devices.

Another question is: the plant is for experimental purposes, I mean research not only in the process but also using advanced control strategies.

I have the following offers:
ABB 800xA, Honeywell ExperionPKS, Siemens PCS7, Emerson Ovation, Yokogawa Centum 3000.

Anyone knows which could be the best in engineering tools for control strategies research and get knowledge in this field?

Thank you for your help ! You are really experts, guys!

 

Any of those solutions will probably work for you.

You mention that you are looking at 800xA. Why not ABB's Compact Products? The controllers and engineering tools are the same but the size of the system is limited to 5 workplaces. This could be better for a small or research type plant as it could save you quite a lot of money. Contact us if you are interested

Rob

www[.]lymac.co.nz

 

I have just done a project using 800xA Profibus DP and PA. The company I work for has extensive experience with Emerson, Siemens and other DCS systems. Our 800xA project was a first of type development plant, we are about to start their next 2 commercial plants.

Basically choose what you feel best with, as all of these systems work. Its going to be your plant and you will have to live with it. Good idea is to look for the installed base, i.e. how many systems does this vendour have in this part of the world. Its a very good indication of the amount of spares and support available.

You have said this is a pilot/development plant. Who is doing the design, setup, programming and maintenance on this? Who is selecting your instrumentation?

Our recent experiences shows us that the last people you want programing is the hardware vendor. Yes, ABB, Emerson etc know their gear very well, and are fairly good when they have a defined specification. However they are useless when they do not have a defined specification or it is getting changed. They are quite poor at lateral thinking. Its just not what they do. You have a plant that has a flexible spec, I don't need to know what it does, the fact it is a development plant tells me this is true. You will need people who can help you solve the problems as they come up and DEVELOPMENT PLANTS GO FROM ONE ISSUE TO THE NEXT. Hardware vendors are notoriously bad problem solvers. DO NOT confuse problem solving and technical support. Technical support regards how to make the hardware do things, problem solving is what its doing with or needs to do with your plant. Find someone who can solve problems, use them to design your system and pick your instruments. Find a hardware vendor who will technically support their system.

Beware all the hardware vendors will try and keep the whole project in house (design, programming, construction, everything). Do not fall for this as one of our clients did.

If you don't want to talk to me ok. But don't listen to idots like the guy saying FF can't do closed loop control. Find someone who has worked on development projects before. It doesn't matter if they have never worked on your type of plant, its the EXPERIENCE with DEVELOPMENT and PROBLEM SOLVING you need.

You BIGGER ISSUE issue will be the instrumentation, its just not as simple a "I need a pressure sensor", go have a look at the emerson catalog and see if you can do the full code on a pressure transmitter.

 

Thanks for your explanation Tony...

Well I also agree that I have to look for good experienced support.
The idea is that the system will be installed with a typical and conventional control, and then I will be the one to deal with the whole system, changing the code to test new control strategies.

The system should be the most flexible, so it has to be capable to deal with Profibus and Fieldbus Foundation, because the instrumentation is shared between Profi and Fieldbus.... I mean the FDT of the DCS should not be limited to one bus.

I don't care about the difficulty in learning programming and complexity of the application environment. I want the one who can do more in engineering related to furnace control and flue gas analysis. There is no electrical generation system involved.

Thanks.

 

You can consider Siemens SPPA3000

 

Hello,

May I suggest that you should also look at how open your prospective system is. You may realize tomorrow that a certain algorithm of a third party vendor performs better that those included in your DCS, or you yourself can write up one in a third party environment. (N)MPC and real-time optimization may be such cases. If it is a research plant, vendor lock-in is very undesirable.

Best Regards,
Andrey Romanenko
Ciengis, SA

 

Thanks Andrey...
That's right. The system should be open due to the fact that this is a reseach platform and part of the research will be in the control field, testing new control strategies.

I need to be able to communicate and the asset management should be optimized for Profibus and Foundation Fieldbus devices. Also it's not only a typical boiler system (like in a power plant) but also flue gases treatment system, so the best, I think is to chose something not specific for power generation.

As I will be control system engineer in the platform, which system gives you the most capabilities for upgrading, adding new devices, well, in other words "playing" with the system. It won't be a continuos 24x7 system... There will be tests, where the platform will be running for at most 10 days.

Thanks.