Which DCS Systems use "Client and Server Architecture" Nowdays?

 

Hi Jorge

All DCS are client-server based systems.

The reason is simple. DCS means a distributed control system where the data (and hence risk of system failure) is distributed over a tightly integrated network.

All nodes fetch data from some other counterparts. Even the old legacy systems are client-server based (roughly speaking) since the data owners are distributed.

With onset of Windows platform in control systems we come across the concept of a server as the single data owner and clients, but even in this case there are systems designed such that the server alone is not the sole data owner.

I have come across arguments like failure of server would paralyze the system but it is not true. In case of a server failure (both redundant servers) you will not suffer any sort of data loss if it is designed well.

Goodluck

 

Hi,

Yes all dcs using server and client architecture. Server is the storage of all information which is getting runtime data from the connected CPU or PLC and sending to all connected clients depending on the purpose of clients again operator commands, setpoints etc all are transmitted to CPU from clients via servers.

In my experience, once both server (redundant servers)were down and there was no runtime data in all clients but the plant was running fine. At that time it was not possible to do or to see any thing from client computers. CPU can run all alone if there is no server or clients.

 

Hi SMS

What you said is correct for PLC-SCADA systems where clients are totally dependent on servers for their operation.

While in DCS, failure of a server does not impact the client operation (and hence plant operation) in any way.

If both servers are down, you will suffer:

1. On historized data retrieval if server stores history.

2. Inability of engineering modifications

3. Some third party interfaces like OPC/ODBC if configured.

Else plant operations remain unaffected.

 

While i agree with most of the points what were explained here, i would like to put down my own experience here (of Honeywell Experion PKS)

1.Honeywell EPKS is also Client-Server Architecture. It is truly distributed architecture in the manner that we can have more than one server having its own database in a single network. However, EMDB (enterprise model database) - where all the assets of the project are configured, will be a common one for the entire network.

To explain with an example - take a case where in a plant there are 3 major areas / sections to be controlled. Each area / section is having a controller, its own IOs & its own redundant database servers. So, in this case, you have 3 Experion Servers in a single network, have 1 EMDB & 3 different Real time databases. Realtime database refers to IO tag database, graphics, dynamic linking, logics related to IOs.

2. Now, in the Client - Server Architecture of Honeywell, there are two types of communication depending on the configuration of the client. There are 2 types of clients - Flex Station, Console station.

Flex station talks to server for data exchange. So, if both the servers fail, the flex stations do not receive anything & go blind.

The console stations directly talk to controller for IO status, set points, alarm setpoints. So, if the servers are down, still the plant can be visualized on these computers, dynamic data exchange is possible, you can start / stop the motors, you can open / close the valves through console stations. (motors can be started / stopped only if they are connected through hardwired IOs / connected to the controller serially).

In Honeywell Experion PKS, there is always a separate Engineering Workstation, which is configured as a console station. Even then, the engineering database is stored only in the server. So, you cant do any kind of engineering modifications.

All the units / equipments connected to the DCS through OPC, though might be running in the field, the status cant be viewed / updated in the systems as the OPC link will be down.

Systems connected to DCS through serial links can be viewed & operated.

Historical trending, event logger, reporting features will be down.

 

Which DCS Systems use "Client and Server Architecture" Nowdays?

Yokogawa’s Centum VP DCS is Distributed Control Architecture which is not based on Client/Server Architecture. In this architecture the database is distributed in each Field Control Station (FCS) which is entirely redundant with less than 1ms switchover time.

The Main Copy of this distributed database resides in the Master Engineering Station. However, this is only a copy with the main database distributed & residing in the Controller. This architecture offers unique advantages which client server architecture cannot offer.