provided by: Design News Ask any automation expert what the future holds, and the word "integration" comes up every time. Usually it's preceded by "increased," "greater," "tighter" or "closer." But what exactly does integration mean to the engineers who design machines? In automation circles, the word has become a catch-all that can describe everything from the physical combination of the simplest components to the unification of complex control and MES systems.
For engineers trying to make automation decisions with an eye to the future, what may matter most is the tightening integration among very different worlds of corporate information technology (IT) and factory floor automation.
Engineers and IT professionals have traditionally spoken different languages and still do to a large extent. But if anything will bring them closer over the next few years, it's a length of CAT 5 cable. "The most significant trend in factory automation is the connection of the factory floor with enterprise systems using Ethernet as the common network," says Sujeet Chand, senior vice president and chief technical officer for Rockwell Automation.
And that connection only stands to grow. A report from the ARC Advisory Group, a firm that tracks industrial automation, has forecast that shipments of industrial Ethernet devices would grow from 840,000 units in 2004 to 6.7 million units in 2009 for a compound annual growth rate of 51.4 percent. (To purchase the report, go to http://rbi.ims.ca/4946-562.)
One reason these connections matter so much, according to Chand, is that factory floor data has simply become too valuable to leave on the factory floor. End-users are demanding the data be passed up to MES and ERP systems, where it can be used for supply chain optimization, improving asset utilization, capacity planning, product-tracking and other make-or-break business decisions. "End-use customers expect more data from the machines they buy. They're increasingly demanding 'dashboards' that let them drill down to the machine level," he says.
Engineering Implications
Engineers who build and run machines will get something out of the integration with IT too. Enterprise systems can make machine performance and maintenance data more visible to engineers, according to Filomena Wardzel, manager of the automation systems business unit of Siemens Energy & Automation Inc.
An IT-friendly approach paves the way for technologies that could allow engineers to collect the data they want. Wardzel gives wireless (see sidebar below) and RFID as two examples. "RFID is not only for tracking big parts once they leave the factory, it also tracks the little pieces on the factory floor," she says. That's good news for engineers responsible for troubleshooting manufacturing problems.
And while RFID is nothing new in more advanced manufacturing operations, the technology will become even more popular as companies try to integrate factory floor operations with supply chain and logistics systems.
To get an idea of just how far IT integration will go, consider that Web servers are increasingly being built into motion and pneumatics components. Numatics Inc., to take one example, has been shipping a valve manifold with a built-in Web server and wireless communications. Why does a manifold need its own Web server? Enrico De Carolis, director of technology development at Numatics, says it provides access to diagnostic information without disturbing control algorithms or real-time operating systems. "Everybody loves the idea of diagnostics, but they don't like the idea of programming it in, especially into real-time operating systems," he says.
With valve manifolds, data on the number of valve shifts might be made available over the Internet. De Carolis believes a similar approach could be taken with a variety of motion components, opening in the way for a new preventive and predictive maintenance model. Rather than the OEM trying to program maintenance functionality, it could be handled by the component supplier or a third party.
Remote maintenance of high-level components and machines has taken place for years, but one debate in the coming years will be how far down to extend remote monitoring now that it's so easy to implement. "Does every component need its own Web server?" asks Chand. "I think that would be overkill." He instead sees machine controls serving as "aggregation points" for component data.
And even De Carolis believes the jury is still out on whether component-based Web servers will be the way to go. "Sure, we have a valve manifold that will send you an e-mail. The question is does anybody want e-mail from their valve manifold," he says. Still, he does add that the Web-enabled manifold has proven to have one of the fastest adoption rates of any Numatics product.
Software and Security Barriers Remain
For all the benefits Ethernet and IT integration could bring to the factory floor, there are some opposing forces at work here. Control engineers and IT guys may increasingly plug in the same kinds of wires, but they still have their differences to overcome. "You're already starting to see control engineers coexisting with IT personnel at larger companies. It's not always pleasant," says Carolis.
In Chand's view, software is one thing that needs to change. "We can't assume everyone on the factory floor will be comfortable with Java and Web servers," he says, and that lack of comfort makes it more difficult to move data off the factory floor. Chand says he would like to see easier ways to integrate data movement into the programming environment engineers use, and additions to the PLC programming methods spelled out in IEC 61131-3 would help. "I'd like to see IEC 61131 send data and information as a Java message. That's not there today," he says, though he predicts it will be in the future.
Another thing that would help, according to Chand, is a move toward Service Oriented Architecture (SOA) - or software composed of loosely coupled, interoperable, preprogrammed services. A service could, for example, track material consumption. Or it could collect uptime data. "SOA exposes information from the factory floor domain in IT-speak," Chand says. "We're looking at SOA very carefully and believe it has promise." He predicts in three to five years SOA will be a much more common way of building software that ties the factory floor and front office together.
Security issues, or at least the perception of them, represent an even bigger source of friction between the IT and factory floor mind sets. "With the old proprietary networks, security on the factory floor wasn't really an issue," says Wardzel. Open Ethernet networks make it an issue. "Simple firewalls can add a lot of security," she says. But she acknowledges that engineers don't trust that firewalls won't interfere with their real-time systems.
Simply throwing up a firewall won't satisfy most engineers for another reason. "IT and the factory floor have fundamentally different approaches to security," Chand says. He says IT generally works within a homogeneous environment - in which they work with one operating system. "The typical factory floor is IT's worst nightmare," he says. "It's a dog's breakfast of multiple networks, fieldbuses and now Ethernet control." He doesn't expect this heterogeneous environment to change anytime soon. As a result, IT's approach to security - firewalls and pushing patches - won't cut it on the shop floor. "IT doesn't usually have to worry about the effects of their security efforts on real-time operating systems."
While plenty of automation vendors already offer factory-floor security products, Chand acknowledges that security still has to evolve before it addresses the needs of factory floor engineers. "The challenge will be integrating all these heterogeneous technologies and creating firewalls for real-time networks without impairing performance," he says. "These are things we're all still working on."
Going Wireless
For all the current buzz about wireless communications on the factory floor, what will engineers use it for? Siemens, for example, has already delivered wireless safety systems for use in robotic cranes and other machines that sprawl over a large area, and this kind of application requires deterministic control over wireless. "The technology for control is there now," says Filomena Wardzel, manager of the automation systems business unit of Siemens Energy & Automation Inc. "But there has been little interest in using wireless for control."
She doesn't expect that lack of interest to change any time soon, nor do other key suppliers. "The biggest need is not for real-time control over wireless, it's for monitoring," says Sujeet Chand, senior vice president and chief technical officer for Rockwell Automation.
Chand notes much of the current discussion about wireless' suitability on the factory floor usually revolves around real and perceived quality-of-service issues, but he sees a far more basic hurdle to wireless adoption: battery life. Wireless tends to get the nod in those very applications where it's difficult to run cables, including power cables. "Think of thousands of wireless sensors, all battery powered. Now image the nightmare of changing those batteries," he says.
Suppliers have already started to deliver self- and remote-powered wireless devices. In one job, Rockwell put wireless vibration sensors on rotating machines found on an oil tanker, using the motion of the ship to power the units. So the problem of powering wireless devices in large installations isn't insurmountable, but it is something that might have to be taken into account from a design and cost standpoint.
Industrial Ethernet Resources
Several different variants of industrial Ethernet vie for dominance on the shop floor. These systems differ primarily in how they implement real-time control. Scott Hibbard, vice president of technology for Bosch Rexroth drives and controls, says some protocols differ very little from commercial TCP/IP. "Others take a 'scorched earth' approach and pretty much use just Ethernet cables and connections," he says. The following links give an overview of each protocol:
For even more technical information, visit the online Industrial Ethernet Book at http://ethernet.industrial-networking.com/ . The site contains scores of articles on various Ethernet-related engineering issues.
Simulating the Factory
Engineers do like their computer simulations. And one thing they can already simulate is how a proposed production line will function. In the coming years, though, these simulations not only become more common but also try to reach further back in the product design cycle. "Engineers will increasingly start to think about automation when they design the product," says Ralf-Michael Franke, president of Siemens Automation and Drives' Industrial Automation Systems Div. "They want to optimize the product and the factory together."
Franke lays out a vision of the future in which engineers will use mechanical CAD data describing their products and production machines to run virtual or "digital" factories from a control-and-automation standpoint. That simulation would then yield the PLC, CNC, or PC-based control software that ultimately runs the real production line, cutting both engineering and commissioning time.
These "digital factories" do presuppose links among the various CAD systems and factory simulation software. But Franke says with today's open systems there are really very few technical barriers to "opening an interface to the world of 3D design."
Automotive OEMs are already taking this approach, using factory simulation systems such as Siemen's year-old SIMATIC Automation Designer. Over the next 10 years, though, Franke believes the link between design and control engineering will strengthen in a much wider range of discrete manufacturing operations.
Safety and Robotics Move to the Machine
As engineers try to find the lowest cost way to build machines, expect them to pack more and more functionality into the machines while keeping the controls as simple as possible. "Looking down the road a year or two, I think you'll see much more integration in the machine than you do today," predicts Helmut Kirnstötter, vice president at B&R Industrial Automation. "In the past, limits in processing power and network speed imposed limits on how much functionality could be put on the machine. Now open networks are improving and much more is possible," he says.
As an example, he points to the evolution of Ethernet Powerlink, the industrial Ethernet protocol B&R developed and later made public. At the SPS Drives Show in Nuremberg, Germany, the Powerlink group rolled out 1 Gbit/sec Ethernet, a tenfold improvement. And it offers cycle times in real-time control applications down to 200 µs. Other industrial Ethernet protocols are likewise getting better (see table below).
These network improvements, along with recent changes to North American safety standards, will soon reshape the way North American engineers implement machine safety, allowing them to adopt integrated safety systems that share space with an Ethernet-based primary control network. "You need a certain level of network performance for integrated safety to make sense," says Chand. He puts the minimum cycle time at 10 ms for safety applications.
But once you have that level of performance, integrated safety makes a lot of sense and is in use in Europe. Machine builders can pull a lot less wire once they eliminate the need for a separate safety network. Kirnstötter notes that integrated safety can have other benefits for machine builders. As safety networks react faster, you can move light curtains, guards and other safety components physically closer to the action. "You can end up saving a lot of real-estate on the factory floor," he says. And integrated safety makes it easier to implement more efficient safety routines - like safe slow downs - that don't require the complete shut down of components or machines.
That notion of integrated safety still frightens some engineers. "They don't really trust silicon yet," says Scott Hibbard, vice president of technology for Bosch Rexroth drives and controls. But that attitude is changing, in part because revisions to NFPA standards will soon make integrated safety viable in the U.S. "Customers here are starting to ask for it," says Kirnstötter.
While there's little doubt integrated safety will become more popular over the next few years, there's plenty of disagreement about the best way to implement it. For example, Hibbard believes integrated safety and distributed control architectures are a natural fit. "You can put safety places where the reaction time is faster," says Hibbard. He argues that drive-based controllers offer some of the fastest reaction times, detecting and reacting to unsafe conditions in a little less than 2 ms versus 30 ms or so for a PLC. "There's a big difference between 2 ms and 30 ms," he says. "It could be the difference between losing four fingers or no fingers."
Improving Ethernet performance also fosters robotics integration. Rather than having a separate robotics controller, the robot controls can be integrated into general motion controllers. Complex robotic kinematics will arguably need their own controller for some time. "But the majority of robotic applications have simple kinematics with three or four degrees of freedom," Chand says.
The benefits here are similar to those driving integrated safety: Eliminate the need to wire in the robotics and synchronize separate controllers and you can save space.
"One of the advantages is the size of the controller," says Hibbard. At the recent Pack Expo show in Chicago, Bosch Rexroth demonstrated a Delta robot controlled by one of the company's 3 × 3 × 3-inch IndraMotion MLC controllers.
| Comparison 1 Safeguarding a Transfer Line Hardwired (ANSI TR6) | Comparison 2 Safeguarding a Roofing Line Hardwired (ANSI TR6) |
| Material and Services | % of Total Project Cost/$ Amount | Material and Services | % of Total Project Cost/$ Amount |
| Electrical Devices | 5% or $20,000 | Electrical Devices | 8% or $136,000 |
| Electrical Interfaces | 16% or $40,000 | Electrical Interfaces | 18% or $272,000 |
| Electrical Installation | 24% or $80,000 | Electrical Installation | 24% or $408,000 |
| Electrical Material | 3% or $7,500 | Electrical Material | 3% or $51,000 |
| Total Cost: | $250,000 | Total Cost: | $1,700,000 |
| Safeguarding a Roofing Line Safety Network (ANSI TR4) | Safeguarding a Transfer Line Safety Network (ANSI TR4) |
| Material and Services | % of Total Project Cost/$ Amount | Material and Services | % of Total Project Cost/$ Amount |
| Electrical Devices | 8.8% or $22,000 | Electrical Devices | 5.8% or $149,000 |
| Electrical Interfaces | 18% or $45,000 | Electrical Interfaces | 18% or $308,000 |
| Electrical Installation | 6% or $115,000 | Electrical Installation | 6% or $102,000 |
| Electrical Material | 1% or $2,500 | Electrical Material | 1% or $17,000 |
| Total Cost: | $207,000 | Total Cost: | $1,411,000 |
| Total Savings of 17% or $43,000 | Total Savings of 17% or $289,000 |
References:Useful Links
//Check out the links below for more info//
author: By Joseph Ogando, Senior Editor, Motion Control
Design News. Copyright © 2007 Reed Business Information, a division of Reed Elsevier Inc. All rights reserved.