DCS: an open road
In spite of the advantages offered by truly “open” distributed control system platforms, the industry standard has been based largely on proprietary systems. This is set to change, however, bringing both users and vendors lowered costs and greater flexibility.
Roger A. Leimbach
MAX Control Systems
The conservative requirements that govern the design and application of distributed control systems have previously forced vendors to utilise architectures based upon largely proprietary hardware and software platforms. However, there is a widely accepted trend toward more “open” systems based upon industry standard architecture and components.
The advantages of open systems to both the user and vendor are well documented:
– A reduced cost for the overall system through local sourcing of primary components which drives costs down
– An increased pace of innovation – next generation versions can be integrated by the user without help from the original equipment supplier
– The ability to mix and match different vendors` equipment due to standard communication protocols
– Reduced dependence on vendors for service. Universal hardware and software is more widely understood by maintenance personnel
– Fast acceptance by plant operators – graphical interfaces with high familiarity speed training.
Note that there are some aspects of open systems that can give rise to detrimental effects if not considered carefully. Security, for example, is critical, and access must be restricted to only those with authority to take action.
Further, some characteristics of open systems can be disadvantageous for the vendor. For example, while long-term support is now made easier by standardized equipment, there may be a subsequent loss of profitability in terms of after market sales and servicing. Also, vendors are driven toward becoming application software suppliers and integrators. On the other hand, development costs are reduced and the time to market is shortened.
MAX Control Systems, based in Lansdale, USA, recently introduced the MAX1000+Plus system based on “open computing standards”. It makes use of commercially available industry standard software and hardware platforms such as Microsoft Windows NT, Intel based workstations, fast ethernet, and Windows CE.
One of the first MAX 1000+Plus systems to be put into operation is for a large diesel engine power plant in Chennai, India. The 200 MW plant, being built by Hyundai Heavy Industries, has four diesel engines each driving a 50 MW generator. Each engine is 70 000 hp, and is as high as a five storey building.
All distributed control systems have multiple levels, each with specific functions, and each with its own level of security and “openness”. At each level the system designer must confront the desire to utilise either open or proprietary technology.
The highest level of the system is typically reserved for plant information management. It has connections to any number of different workstations and computers from different suppliers. Security must be considered; while the free flow of data must be supported, it should not be possible to make changes to the system parameters or implement control of the process.
The hardware at this level has evolved from mainframes to RISC workstations, to high end PCs running Windows NT. Operating systems have moved from IBM to VMS to UNIX and now NT. Data interchange standards are now emerging which utilise the ethernet and TCP/IP standard. The TCP/IP standard allows data to transmitted over a wide range of media and to most computers. The MAX1000+Plus system provides for remote Windows so that all the displays available to the operators are also available to management.
At the operations level distributed control systems first incorporated proprietary operator workstations. Over time vendors introduced RISC processors running in the UNIX/X-windows environment. In 1992 MAX Control Systems introduced the MAX1000 system with high-end industrial grade PCs running Windows operating system. This experience has led to the introduction of secure Windows NT with the MAX1000+Plus system.
The communications networks provided by all vendors have been, for the most part, proprietary. There have been de facto performance standards adopted by the industry, but no universal standard that permits intersystem data communications has emerged. Field bus does show promise as systems become more fragmented and move out to the field equipment. However, the need for security and deterministic highway loading has led most vendors to maintain proprietary systems.
Max Control Systems developed a new communication network, MAXNet, based upon switched fast ethernet. This network meets the requirements for distributed control and allows for intersystem data communications using open ethernet standards.
– High availability – there is no single point of failure
– High speed – it has fast response, and no delays or latency is permitted
– Unlimited access – stations connect at any time. Simultaneous traffic is supported
– Self-diagnostics – message checks are made on all transmissions
– Expandability – adding more stations is simple, even if they are the next generation
– Operation under harsh conditions.
Control processing hardware and software remains largely proprietary. Reliability and performance considerations result in “home-built” operating systems. I/O bus designs are a mixture of open (HART, Fieldbus) and closed solutions. The need for system-wide sequence of events recording is an example of a function that drives vendors to custom I/O solutions.
The MAX1000+Plus system incorporates a new Distributed Processing Unit (DPU) that utilises a Pentium processor running Windows CE that promises to open the lower end of the DCS, thus enabling users to import their application software directly into the controllers.
The DPU provides modulating and sequence control, utilising the world standard IEC1131.3 configuration tool set. System-wide sequence of events recording is provided to one millisecond resolution.
A global database
The greatest single benefit of open systems is the potential to easily incorporate information from other dissimilar systems into a single plant global database. Typically, information can be classified for data exchange as follows:
– Real-time process data – current process variables reflecting plant operating conditions. At the very least, a common name and tagging system must be assigned for all systems in the plant
– Historical data – process and event summaries created from queries defined by specifying points, dates, time, and duration
– Operator actions – manual operations, tuning changes and changes in alarm values
– Configuration – the most difficult, and only with common configuration tools.
While there is no common standard protocol that can be used to access data in this format, MAX Control Systems has developed a mechanism to support a wide variety of standard data exchange systems. MAX1000+Plus employs a “software backplane”, that when installed in a foreign system with an ethernet interface, can permit the foreign system to coexist on the MAXNet data highway with other MAX1000+Plus stations. A software equivalent of a “firewall” isolates the systems.
The software backplane has only four basic functions:
– Read – get information now
– Write – put information now
– Subscribe – receive information that has changed
– Register – make information available.
And it has only four data types:
– Process variables
– Historical information
The introduction of secure Windows NT, advances in Developer Studio, and the implementation of Object, Linking, and Embedding (OLE) has made it very advantageous to use the Windows approach. Like it or not, Microsoft Windows is the most recognisable operating system in the world, and its use assures fast acceptance by users.
MAX Control Systems adopted the Windows system before it was popular for control systems, and utilises a native 32-bit graphical user interface called MAXVue. All stations in the system run MAXVue and interface with Distributed Processing Units through the software backplane.
MAXNet is based on the fast switched ethernet standard. Fast ethernet is based upon conventional ethernet, but is ten times faster. It runs at 100 Mbits/s and is based upon the familiar CSMA/CD standard. It can coexist with the slower 10Based T networks. No protocol translation is necessary to interconnect the networks. It uses the same media, coaxial cable, copper, or fibre.
The performance of fast ethernet can be enhanced, and the disadvantages of CSMA/CD systems can be eliminated with the addition of intelligent switches. Switches also permit simultaneous access to the network by all stations. This variant is called fast switched ethernet.
Intelligent switches are employed at various locations in the network to form a backbone for the system. It can be as simple as a star, but it is more likely to include multiple redundant sets at various locations. Each station is direct connected to a switch. Stations do not share a line as in a multi-drop system. A switch may have as many as ten different stations direct connected. The backbone connects the switches together over a wide area. The network is always redundant.
Major advantages of switched fast ethernet are as follows:
– Full duplex operation. This provides for simultaneous two-way communication, thus doubling the bandwidth to 200 Mbits/s. Throughput is effectively doubled.
– Switches support multiple simultaneous communication, enabling more than two sets of stations to address each other. The performance is dictated by the switch`s bus arbitration scheme, which in turn is based on the activity at any given moment. Switches sustain traffic in both directions and between stations simultaneously.
– No collisions occur. No collision detection is necessary, and traffic is never stopped. Stations that monopolise a node are throttled back by the switch. Switches produce “wire speed” transmission rates through the nodes.
– Connection of 10 Mbits/s links to 100 Mbits/s networks
– Increased physical size of the network
– Access is permitted at any time. Switched fast ethernet is a democratic system. There is no fixed scheme that determines when access is permitted.
There are many manufacturers of switches since they are used throughout the communication industry. They come in many combinations and sizes. They can have fibre optic ports and combinations of these in the same switch.
A major advantage
A typical distributed control system for a 300 MW coal fired power station has subsystems associated with various functions in the plant. These may have multiple distributed processing units (DPU), each connected to a switch with a redundant 10 Mbits/s link. They do not share a link with another DPU. The same is true for the workstations. All links are redundant. The 100 Mbits/s backbone is full duplex and will most likely be fiber optic.
The system uses standard off-the-shelf components (switches) and software (TCP/IP). The workstations are Intel based and run Windows NT. The DPUs are also Intel based. Each type of station uses the same ethernet board.
A major advantage of the switched fast ethernet system is the addition of an alternate path for each transmission. The workstation automatically switches to the alternate path when required. In fact, the path will be changing with every new transmission. This permits the system to survive multiple-fault scenarios.
The MAX 1000+Plus System installed at the Chennai diesel power plant has four operator stations located in a control room overlooking the engine hall. In addition, a workstation is provided in the engineer`s office. The workstations are connected to the DPUs via the new MAXNET 100 Mbits/s communication network.
In addition a remote workstation is located over 300 m away in the administration building. This workstation can access all plant data and see all the graphics available to the operators.
The system has 20 DPUs (ten redundant pairs), which process over 3000 I/O. The system controls everything from balance of plant operations to generation.
The plant will provide a welcome break for the electricity-starved local economy. From the time of ground breaking to first power generation has been approximately one year – a major feat for India. Hyundai Heavy Industries plans to add two more engines to the plant and to build another similar plant in India.
“Open” systems can therefore provide many potential advantages to both the user and the vendor. The use of Microsoft technology is also an advantage; and Windows NT and CE will become more prevalent as operating systems for distributed control.
As standard networks are utilised, bandwidth will increase significantly to the extent that it will no longer be a consideration. Switched fast ethernet provides a cost-effective alternative to token passing systems.
Open systems will continue to evolve and be accepted by distributed control system users because of the overwhelming advantages. However, a good data exchange standard will be required, such as MAX Control Systems` software backplane, that when implemented in foreign systems will provide for a common plant database that is transparent to the user.
Figure 2. Typical distributed control and data acquisition system with multiple levels
Figure 3. MAX1000+Plus system. Note that all MAXNet components are redundant
Figure 4. Typical MAX1000+Plus system for coal fired power plant. Note that the system is functionally distributed. A remote processing unit is provided for each subsystem