In an effort to become one of the most responsible food companies on the planet, General Mills has embarked upon a massive global mission to make more nutritious products, reduce the company’s environmental footprint, and increase the safety and sustainability of key ingredients.
There are many moving parts in the overall plan, but one project underway has the IT and manufacturing team working to stitch together the supply chain from field to fork—that is, “the flow of information and attributes of the ingredient from the farm to manufacturing to its transformation into a finished product and onto the grocery shelves,” says Jim Wetzel, technical director at General Mills.
Given the proliferation of sensors and devices that populate the ever-growing collection of smart objects, known as the Internet of Things (IoT), one might think gathering information from outside of the enterprise is easy to do. But that’s not the case, especially when trying to keep data integrity intact while interfacing with myriad external systems. But Wetzel, who is also chair of the Smart Manufacturing Leadership Coalition (SMLC)—a non-profit organization comprised of manufacturers, technology companies, academia, and U.S. government agencies and laboratories—is leading another important mission: Solve the underlying interoperability issues across the manufacturing value chain.
SMLC is developing an open, ubiquitous manufacturing platform that, according to Wetzel, can be fully integrated—and knowledge-enabled—to meet cross-industry business-case objectives. “We are looking for an architecture that is not an Internet of Things, but an Internet of Manufacturing that allows solutions to be connected and applied,” he says.
SMLC is not alone in its efforts to create a smarter manufacturing ecosystem. There are many groups forming, from the Industrial Internet Consortium (IIC, with founders AT&T, Cisco, General Electric, IBM and Intel) to the AllSeen Alliance (a Linux Foundation collaborative project) to the German government-inspired Industry 4.0 initiative. Each group has a slightly different approach, but all have the goal of advancing manufacturing by taking advantage of the global IoT market, which, according to International Data Corp., will grow from $1.9 trillion in 2013 to $7.1 trillion by 2020.
Pilot projects and test beds are underway to prove the viability of each group’s respective reference architecture, and to create a connective framework that closes the loop between industrial automation and the enterprise. In the big picture, all have an eye on improving global operations, increasing competiveness, and even creating jobs by making it easier to set up local manufacturing sites.
In addition, the productivity benefits of smart manufacturing directly relate to decreased downtime and increased agility. “In today’s world, an almond farmer will fax us a document with product testing attributes, such as moisture, weight, taste, and that becomes our certificate of analysis, which we put in a file cabinet,” Wetzel says. “In tomorrow’s world of smart manufacturing, we run the fax through OCR [optical character recognition] technology, store the data in the Cloud and push it into the MES [manufacturing execution system] so that the system knows the [product] properties and adjusts the recipe of production accordingly.”
Of course, having this level of interoperability requires extracting data from control systems that traditionally operate in a closed loop inside the plant. So while SMLC and other industry groups work on open, flexible frameworks, the automation community is addressing the communication issues, which, hopefully, will result in industrial IoT standards.
When it comes to passing information from the control system to the Cloud, “we want to do it in a standard way,” says Mike Bryant, executive director of PI North America and administrative director for the OPC Foundation. There needs to be a common language, and “that piece is lacking right now.”
Developing the lingua franca of Industrial IoT
Enabling interoperability across such a large landscape requires a two-pronged approach that looks at the enterprise from the Cloud down and at manufacturing from the controllers up. “We are hitting it from both ends,” says Wesley Skeffington, principal engineer at GE Global Research’s Real-Time Embedded Systems Laboratory. GE has developed a software platform, called Predix, which can connect industrial assets with the Cloud. In addition, the company is working to embed knowledge into controllers so that they know what to do when a piece of information is received.
The key here is context.
“Besides basic interoperability, you have to get the boxes to talk the same language to make use of the data,” Skeffington says. And, for that, the industry needs new semantic standards. “Semantic interoperability happens by embedding context, so as the data aggregates up through the systems, you are not trying to find the Modbus register list to identify what the heck the data is.”
To that end, the OPC Foundation, the group responsible for developing and maintaining OPC UA, an open platform interoperability standard for industrial automation, is working to figure out what that IoT syntax model looks like. More importantly, the effort underway by the OPC Foundation and others is focused on leveraging existing automation standards so that manufacturers are not reinventing the wheel.
In addition to OPC UA, there are many network, fieldbus and machine communication protocols and standards that provide interoperability on the plant floor. And that, industry experts say, is providing a solid foundation for IoT.
“The goal of Industry 4.0 is not to [move everything into the] IT world, it is to make machines more efficient and easier to maintain,” says Peter Fischbach, automation industry sector manager at Bosch Rexroth. To that end, the company’s Open Core Engineering product connects the worlds of PLC and IT via open standards, function toolkits and an interface supporting numerous programming languages to enable the integration of smart devices and automation systems. “The basic architecture of a PLC doesn’t need to be changed, but the way the data is exchanged does have to change.”
In other words, the key to the next industrial revolution—in which big machines and Big Data merge across the Internet—is to combine information with action. More importantly, IT and automation systems need to speak the same language.
Industrial IoT action—mixed messages?
There are companies that already have technology solutions to tie together factory floor systems with the enterprise. In addition to Bosch Rexroth, Kepware Technologies and Opto 22 are working to ease the programming pain of connecting heterogeneous devices and machines to enterprise systems, either directly or through the Cloud.
Unfortunately, many manufacturing managers, including Benoit Lapensee, MES coordinator at Cascades Inc., a packaging and tissue paper company in Quebec, Canada, are still not sold on the idea of interoperability outside of the plant floor. Cascades, which has many locations throughout North America, has acquired a variety of equipment and plant floor software as a result of acquisitions. Inside the secure company wide area network (WAN), Lapensee links the different production systems to the company’s central MES, which is GE’s Proficy, using Kepware’s KEPServerEX, a communications platform that connects with a variety of device interfaces, including OPC and proprietary protocols.
It’s a one-way collection of data from the PLC to Kepware and then to the historian and MES for calculations on OEE and downtime. Data is then compiled into reports for supervisors and executives to see the status of operations around the world.
Kepware, through a recent alliance with operational intelligence provider Splunk Inc., has the ability to help Cascades extract even more meaningful information from the industrial data pulled out of sensors, devices and control systems. Through Kepware’s Industrial Data Forwarder (IDF) for Splunk, all of that information can be sent to the Cloud to apply Big Data analytics, providing real-time insights into operations.
But Lapensee is cautious when it comes to production data. “Every day my challenge is to build new reports based on data from different devices, but we are not yet to the point where we want to transfer data from one machine to another,” or outside of the organization, he says. “My concern is related to the control of the process. Doing something remotely could result in a safety issue.”
Indeed, many manufacturers are not willing to test new technology at the expense of their manufacturing processes. But Opto 22’s Benson Hougland says security shouldn’t be a concern. Companies like Google, Apple and Facebook have built their businesses on IT technology, and most people are comfortable enough to do their banking online. Of course, communication on the Internet is very different than the factory floor. It is done via Hypertext Transfer Protocol (HTTP) and information can be pulled into a “mashup” of applications via Representational State Transfer (REST) APIs.
This has led Opto 22 to take a slightly different stand when it comes to Industrial IoT. The company has turned to existing Internet architectures, like REST, for its new product called groov. Layered on top of Opto 22’s SNAP PAC System for industrial control, groov provides a way to create actionable views of different data sources. Dragging and dropping an icon from a web browser can present field device information on a mobile or desktop system.
“On the back end of groov, we are talking to OPC UA or Modbus TCP,” Hougland says. “In the future, when the Internet of Things takes off and the standardized way to communicate is in a RESTful way, groov will speak to those devices as well.” Imagine, he says, if you could quickly assimilate plant floor data with energy usage and even weather patterns mashed up to create a more valuable big-picture view.
Though Hougland is betting on REST and HTTP for the future, there are many methods emerging to connect machines on the factory floor and the Internet of Things.
“There are different groups creating different standards, and they all say their standard is the best,” says Erik Dellinger, product manager at Kepware. “But I don’t know if it will ever happen that one will rule them all.”
There will likely be a handful of standards that will play together in the IoT sandbox. But one thing seems clear: OPC UA will be in that sandbox, too.
What lies beneath IoT
SMLC’s Wetzel is working on a framework that sits on top of existing industrial automation investments. “We start with the data, and our mission is to act on the data,” he says. But how does that data get from sensors and other smart devices to a PLC or a historian to populate the smart manufacturing platform? “We’re not working on that.”
That falls under the OPC Foundation, which is tasked with creating a “dream team” of industry experts from the vendor and end user community, which will put all the pieces of the IoT interoperability puzzle together—from the physical network infrastructure to the fieldbus to the industrial network and out to the Cloud.
“The challenge is how to share data across all of the different levels and still have meaningful information,” says Carlos Pazos, product marketing manager at National Instruments.
“And, from a Profinet and OPC perspective, I’m trying to figure out what will be required of me in the next five years,” says PI North America’s Bryant. “What the control systems will need to provide, and what formats and speed. These are the things I’m constantly thinking about because I don’t want anyone to come back to me and say I have to change formats around, it’s not going to work.”
In addition, there are real concerns that an influx of information generated by a multitude of new smart devices populating a network could stress the infrastructure. “I’m not concerned about the speed of Ethernet; we have a lot of headroom regarding the performance of the network,” says Fluke Networks CTO David Coffin. “But we need to ensure the reliability of connecting up a sensor in a harsh environment.” For example, mechanical, ingress, chemical/climatic and electromagnetic conditions could impact wireless network performance.
Kepware’s Dellinger concurs that the network, including bandwidth, needs to factor into the overall equation. “There is an effort within Kepware to move closer to the edge,” including the millions of sensors in the field, Dellinger says. “But all of those devices can’t be communicating back to an app like Splunk at the same time, so there needs to be a middle layer that aggregates communication and manages network bandwidth.”
These are all things that need to be taken into account, which is why creating this next-generation manufacturing model could take years.
“Right now there is a lot of hype around IoT,” says Tom Burke, president and executive director of the OPC Foundation. “Everyone is talking about the fact that we will be able to get at data that we couldn’t get to before, but then what do we do with it?”
With some of the work the OPC Foundation has already done in oil and gas and other industries, the group is working on creating a set of industry-specific objects that describes data information—like a device library that sits in the Cloud. This is part of the semantic standardization effort underway to ensure all parties speak the same language.
Though this is just one small part of the overall IoT architecture, a setup such as this could make it very easy to hand off data to a smart manufacturing, Internet 4.0 or industrial Internet framework.
But all of this is still in the early stages of development, and how this will evolve is still a mystery.
“It’s like trying to predict what the Internet would be like in 1985,” says Skeffington. “No one [back then] could have predicted what it would become.”