Using an Ethernet for plant control as well as information offers manufacturers a host of benefits, including easier integration of shop-floor to top-floor systems and the chance to use a single network infrastructure for several different functions. Clive Barwise of Rockwell Automation believes it is only a matter of time before Ethernet becomes as widely accepted on the factory floor as it is in the office.
The development of plant control networks such as DeviceNet and ControlNet over the past decade has brought many benefits to manufacturers and OEMs. No more complex, time consuming and fault-prone hardwiring of individual devices back to centralised I/O racks through a myriad of multiwire cables and junction boxes. Instead, a network allows intelligent devices and sensors to be directly integrated into the control system and simple device's discrete I/O interfaces to be located near the machine.
Installation time and risk is cut dramatically, configuration - and reconfiguration later when required - is simplified and reliability improved. And, with this, devices and I/O have become more capable, providing higher quality information and allowing for more informed decision-making.
The attraction of moving to an Ethernet for plant control is based on the fact that Ethernet is already well established on the factory floor for information and visualisation - it is often used to link programmable controllers to HMIs, scada and manufacturing execution systems. Where these networks use standard TCP/IP, this information can be easily and cost effectively accessed remotely via the Internet or a company intranet.
Debunking a redundant, perceived shortcoming
"Too often it is argued that off-the-shelf commercial Ethernet and IP-based solutions lack the degree of determinism and predictability required for industrial control, and if your experience of Ethernet is a 20th century view, that may be true," says Barwise. "But one of the growth drivers for 21st century commercial Ethernet is Voice over IP (VoIP) - demanding the same type of predictability and reliability (or quality of service) as the most demanding industrial control application. Obviously a device must react in a predictable way to information not being delivered on time but that is no different to today's industrial networks which rely on single access bus technology."
To overcome this perceived shortcoming, several Industrial Ethernet systems have been developed with the assumption that standard Ethernet cannot deliver the benefits offered by existing plant control networks without re-engineering, resulting in special silicon or non-standard media access methods. However, only one - EtherNet/IP - offers the industrial optimisation of a specialised control network with the openness and flexibility of standard Ethernet, taking advantage of today's Ethernet technology without resorting to modified switches, silicon or stacks.
This does not mean that there is no need to industrialise commercial components; the widespread adoption of PCs on the factory floor has resulted in a need for industrialised PCs. Industrial Ethernet must be as close to commercial Ethernet as a PC on the factory floor is to the one in the clean control room.
In the commercial world, Ethernet is being securely used for all communications from voice and video to data and Internet - not just because it can be, but because it financially makes sense. The decision to put Ethernet on the factory floor similarly is not taken just because it works, but because there is a business case to add automation to the already long list of services that are being carried over the current infrastructure. Increasingly, the question being asked is: "What is the business case for keeping the automation services separate?"
In building this business case, we need to look at some of the places where costs are currently duplicated with parallel networks or where capabilities are not available because of the absence of an Ethernet network.
Ethernet's broad bandwidth allows it to carry far more data at higher speeds than any control network - this bandwidth should be utilised for non-automation services. A 100 MB Ethernet could stream video images from a vision system or webcam live to a HMI or PC screen, enabling remote monitoring of a machine or production line. This can allow safer operation using fewer operatives or enable an OEM to monitor a machine remotely to help diagnose problems.
With Ethernet running around the plant floor, it is possible to hook up local printers and print maintenance work orders, production run instructions, wiring diagrams or other useful documentation from a PC in an office or another part of the factory connected to the network, taking information to the operator, not the operator to the information.
Automated storage and retrieval of production data in PC-based databases would be simplified with the use of a common network and communication, and extending Ethernet down to every device on the factory floor would allow transparent access for remote configuration and troubleshooting.
Taking the machine information to the maintainers, even in their own homes, is more efficient than taking the maintainer to the machine.
Combining the information and control functions into a single, multi-purpose network offers the possibility of reduced cost and installation time, and once installed an Ethernet network could carry any type of compatible data or voice communication.
Start up time for plant control systems using Ethernet is greatly reduced because faults are reduced when pre-made cables are used, and any faults that do occur are isolated and so easier to diagnose. This is partly because the PC and Ethernet have become a mission-critical part of commercial enterprise and so fault-finding and maintenance tools are not only available, but widely deployed. The star or tree topology of an Ethernet network also makes adding extra devices at a later date very easy with no impact on the existing installation.
Industrial Ethernet considerations
With all these benefits on offer, it would appear to be only a matter of time before Ethernet becomes as widely accepted on the factory floor as it is in the office. But there are several important factors manufacturers need to take into account when considering implementing an Industrial Ethernet.
The first is that, while Ethernet is very versatile and can perform some functions that dedicated device level networks cannot, it does not offer all the features of a purpose-built control network. Just as a single, general purpose golf club could never work as well as a specialised putter, driver or sand wedge, so a single multipurpose Ethernet does not do everything as well as a more specific single-function network.
ControlNet for example is designed as a high performance I/O network, offering guaranteed determinism, the ability to handle large volumes of I/O traffic, and native redundant media for critical applications.
DeviceNet on the other hand is ideal for integrating simple digital devices such as switches and actuators in a highly scalable and cost effective network.
So, for specific control applications requiring the robustness of ControlNet or the scalability of DeviceNet, an Industrial Ethernet is unlikely to fit the bill and, as a result, Rockwell Automation remains committed to the development and support of products using these networks.
But these are considerations when putting together a business case for not using Ethernet; for a significant number of installations these features are not required or are less important, and a single Ethernet able to fulfil a number of roles offers the end user significant benefits above and beyond those possible on any dedicated industrial network.
This is why it is important to remember that not all Industrial Ethernets are compatible with the off-the-shelf variety used in the office. If the Industrial Ethernet is to be used for extra functions such as streaming video, connecting printers or carrying voice over IP it is vital that the required devices can co-exist on the network without limitations.
Just because a network is called "...net" or "Ether ..." it does not necessarily mean it has anything in common with Ethernet, and many implementations of Industrial 'Ethernet' are in fact two or three separate networks each dedicated to I/O, information or motion control. Any network that requires special infrastructure, chips or topologies is not a true Ethernet and therefore restricts or eliminates the financial benefits that can be gained from a truly converged network.
This is why Rockwell Automation is committed to EtherNet/IP, which uses the standard TCP/UDP/IP protocol (just the same as the Internet) for all automation services including configuration, information, I/O control, time synchronisation, safety and motion control. EtherNet/IP allows unmodified IEEE802 Ethernet to carry both time critical control data and all the information normally handled by a standard Ethernet. Further, EtherNet/IP uses the same Common Industrial Protocol (CIP) as DeviceNet and ControlNet, bringing best of breed industrial optimisation to best of breed networking technology.
Developed under the auspices of the ODVA, EtherNet/IP is a completely open network and many fully tested plug and play products from many vendors are freely available.
While standard Ethernet cannot offer all of the more specialised features of a dedicated control network, the power and flexibility of EtherNet/IP means it offers some exciting new possibilities for very many control applications and as commercial Ethernet moves forward the benefits offered by EtherNet/IP will continue to grow.
For more information contact Jeff Sandison, Rockwell Automation, 011 654 9700, [email protected]
| Tel: | +27 11 998 1000 |
| Email: | [email protected] |
| www: | www.rockwellautomation.co.za |
| Articles: | More information and articles about Rockwell Automation |
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version