Catch Problems Before Process Shutdown

Plants are using diagnostics and prognostics to catch problems before they shut down the process.

Aw 1270 1011 Diagnostic
Until recently, South African Breweries Ltd. (SAB) was flying blind when it came to detecting the health of its multiple plants across South Africa. Plant operators had to inspect the line manually, and even then, they had difficulty detecting whether something on the line was about to fail. “Not too long ago, we had no tools available that could monitor communication between devices. There was no live view of what was going on within our industrial network,” says Cy Sotiralis, manufacturing technical specialist at SAB. “We had no way of identifying misbehaving or failing devices.”

The plant had numbers of devices communicating with each other, but the plant didn’t have diagnostics that would help operators troubleshoot and detect issues. In order gain that protective visibility, SAB deployed IntraVue from Software Toolbox Inc., a software supplier located in Matthews, N.C. “That changed the way we monitor and detect network disturbances throughout our manufacturing sites,” says Sotiralis. “On many occasions, it has helped us detect issues before they even happened. We now have a visual of what’s going on within our network.”

SAB uses a range of network devices from Cisco Systems Inc., the San Jose, Calif.-based network equipment provider, as well as distribution switches and industrial scalance switches. “IntraVue quickly and easily scanned our network after some minor configuration. Within an hour, SAB had a fully functional network monitoring system.”

Plants are deploying industrial networks that carry eyeballs across plant operations to make sure all the devices are operating correctly. As well as making sure that the devices are alert, the network is also carrying data back on the health of the plant, making sure that pumps are not clogged and that motors are not overheating. The result is the ability to fix a failing part before it shuts down the line, or at least keep the problem local and not system-wide.

Some plants are taking diagnostics data a step further into prognostics—rather than catching a pump just before it fails, prognostics can analyze the entire plant operation and fine-tune it for prolonged health and greater efficiency. “You’re getting more and more information along the network now. Plants used to use the data just to fix a problem. Now, plants are moving to prognostics to prevent the problem from happening. You get more information on the machine and use prognostics to check and replace things earlier,” says Jay Lee, Ohio Eminent Scholar and L.W. Scott Alter Chair Professor at the University of Cincinnati’s Department of Mechanical Engineering, and director of the National Science Foundation’s Center for Intelligent Maintenance Systems.

Checking system health

Some of the network health monitoring comes via a protocol so ancient that it’s odd to view it as a new addition to plant networks: simple network management protocol, or SNMP. Information technology (IT) has used it forever. It’s the pinging system that tells you whether your printer is on the network and whether it’s available. Yet, it’s new to industrial networks in plants.

SNMP tells operators whether all of the network devices are awake and operating correctly. SNMP has been somewhat of a breakthrough in plant operations lately. “A lot of hardware companies are adding SNMP to monitor the field of devices and gather the process data that’s coming from the devices,” says Boyce Baine, senior applications engineer at Software Toolbox. “With SNMP, you’re getting data from the devices such as uptime, number of cycles and internal temperatures for the device.”

As the monitoring system checks devices to make sure they’re operating and brings back data on the health of the system, it delivers efficiencies. One benefit is the efficiency of the automated eyeballs. With the network checking system health, fewer people have to walk around to see if everything sounds and looks okay. “Using intelligent devices plus fieldbus allows companies to manage their plants with fewer people,” says Adam Krug, product manager of industrial control and advanced motor protection at Cleveland-based supplier Eaton Corp.  “If you use these devices to protect the motor and provide information back as to whether the pump is starved, you don’t have to send people out to look. You’re looking at it remotely.”

Increased uptime is the strongest push behind network diagnostics. The goal is to reduce the risk of stopping production when part of the system fails. “It’s all about allowing the process to continue without shutting down,” says Marty Jansons, network consultant at automation vendor Siemens Industry Inc. in Norcross, Ga.. “It all goes to creating a fault-tolerant design, so you don’t have a complete plant shutdown when one area is having a problem.”

The difficulty and cost of diagnostics is not great if it is added to an existing network. “It’s a really hot item—not just for end-users, but for anyone troubleshooting a system or installation process,” says Bernd Schuessler, business development manager at Pepperl+Fuchs, an automation components supplier with U.S. offices in Twinsburg, Ohio. “It’s a low investment. You can prevent segment shutdown and save a lot of money by keeping your production up and running.”

Part of the efficiency of network diagnostics is the ability to get the right data to the right people. That means sending machine status data to maintenance and sending production data to the office. “By using intelligent devices and more processing at the input/output (I/O) level, we can push data to the right people,” says Tom Alford, product manager for integrated condition monitoring at supplier Rockwell Automation Inc., in Milwaukee. “If something is happening, you need to get it to the right person now. The maintenance guy should get vibration data a few times a day.”

Remote monitoring

The diagnostic and prognostic data is viewed remotely. In most cases, the monitoring is done internally. Even if the organization has multiple plants that send their data to a central location, it can be done through a protected and secure network. “There’s a big move to do remote monitoring within an organization where the plant’s expertise can be brought to bear from a distance,” says Harry Forbes, senior analyst at ARC Advisory Group Inc., in Dedham, Mass. “The internal network is already secure, so it’s not the same as bringing in someone from the outside.”

Yet, there are opportunities for vendors to provide remote monitoring of their own equipment. This task has become a value-added service offered by many machine builders. Rather than train plant operators to monitor the health of the machine—or send machine personnel to the plant—the machine builder can watch the machine from its own facility. “If you’re only concerned with a particular machine or system, it’s the vendor diagnostics that are meaningful,” says Forbes. “There is certainly an impetus for remote monitoring by vendors, but it is a security issue.”

Rockwell had partnered with Cisco Systems to help plants overcome the resistance to remote monitoring by machine builders. “The machine builder has to be able to look at the data at the manufacturing facility. But the IT group is concerned about security,” says Mike Hannah, manager for networks at Rockwell. “So we’re working with Cisco Systems on how to put best practices on these machines to make them IT-ready. We find it will be accepted by the IT group if you use these best practices for security.”

While much of the technology to run diagnostics has been around for years, it hasn’t all been gathered into system-wide network diagnostics until the last five or six years. Some say the recession was a spark to get plants to deploy network diagnostics. For once, they had time. Even so, adoption is still low when it comes to robust diagnostics. “About 10 percent of Ethernet users purchase more advanced Web Services,” says Dan DesRuisseaux, manager of the Ethernet Group at automation supplier Schneider Electric, in Palatine, Ill. “As far as diagnostic programs, the major suppliers are saying penetration isn’t great at this point, even though PLC diagnostics and Web Services have been around for a long time.”

Plant operators note that the diagnostics and prognostics that are available on industrial networks are quick to deploy and fairly easy to configure for health monitoring. Operators are setting up alerts along the system that indicate when things slip even slightly out of optimal. Thus, a slightly different flow rate in a pump—which would have gone undetected previously, can now be identified and corrected before a slight clog leads to complete failure.

Related Sidebar - Monitoring a Water Plant In Real-time
To read the article accompanying this story, go to www.automationworld.com/feature-7837.

Subscribe to Automation World's RSS Feeds for Feature Articles

More in Control