Manufacturing is witnessing a transformation through the implementation of the smart-factory concept and factory-automation technologies, according to Markets and Markets, which has just published a report on the topic. The firm expects the IIoT market to grow from $113.71 billion in 2015 to $195.47 billion by 2022, at a CAGR of 7.89% between 2016 and 2022.
The Markets and Markets report1 covers devices, technologies, and software—including sensors, RFID devices and systems, industrial robotics, distributed control systems, autonomous haulage systems, condition-monitoring systems, smart meters, camera systems, wired and wireless networking technologies, and PLM, MES, and SCADA software. In addition to manufacturing, the report covers industry verticals including utilities, oil and gas, metals and mining, retail, healthcare, transportation, and agriculture. Among the verticals, the firm found that the manufacturing vertical accounted for the largest share of the IIoT market in 2015.
In addition, government initiatives such as Industrie 4.0 in Germany and Plan Industriel in France are expected to promote the implementation of the IIoT solutions in Europe, the firm notes. Moreover, leading countries in the manufacturing vertical such as the United States, China, and India are expected to further expand their manufacturing industries and deploy smart manufacturing technologies to increase the contribution of this vertical to their national GDPs.
Matt Newton, director of technical marketing, Opto 22, commented, “During 2015 and 2016, two organizations dominated the IIoT headlines: Plattform Industrie 4.0, rooted in the concepts of efficient manufacturing and the smart factory, and the Industrial Internet Consortium (IIC), which takes a more cross-domain approach to the IIoT.”
He added, “Both groups developed reference architectures to help streamline standardization and adoption of IIoT technology. While similar in some respects, they also differ on many points.”
An Opto 22 white paper2 elaborates on the differences. IIC published the Industrial Internet Reference Architecture (IIRA) in 2015; it’s a standards-based architectural template that IIoT system architects can use to design their own systems. “The IIRA is designed to address the intelligence and connectivity now being built into the sensors, actuators, and other low-level devices deployed in a variety of applications, including smart manufacturing, the smart grid, the connected hospital, smart transportation, and many others,” the paper notes.
In contrast, the paper notes, the goal of Industrie 4.0 is to optimize production through the development of the smart factory built around four pillars: interoperability (with machines, devices, sensors, and people communicating with one another), information transparency (with systems creating virtual copies of the physical world through sensor data), technical assistance (with systems supporting humans by solving problems and performing difficult and unsafe chores), and decentralized decision-making (with cyberphysical systems making simple decisions anonymously). The two organizations have been conversing since 2015, but differences remain, making companies unsure of where to invest.
The arrival of the IIoT also can spark tensions between operations technology and information technology teams, which must work together to realize IIoT benefits. The Opto 22 paper describes IT as living “… in a world of constant change and never-ending upgrade cycles, seeking the newest, fastest computing hardware and software to gain some competitive advantage the enterprise can use to its benefit.”
In contrast, the OT team “… handles the installation, maintenance, and occasional upgrade of equipment….” This team sees upgrade cycles of 10, 20, or even 30 years or more, with the equipment itself being expensive and often proprietary.
With OT and IT teams evaluating competing visions for the IIoT, it’s not surprising there still can be confusion over definitions. Sewing further confusion, Industrie 4.0 has been adopted by companies worldwide who can put their own spin on it. The translation “Industry 4.0” was ubiquitous on the exhibit floor of the recent IPC Apex Expo show in San Diego, bandied about by American, Asian, and European companies alike.3 (In this article I’m using the English “Industry 4.0” unless specifically referring to the German Plattform Industrie 4.0.)
According to Prof. Birgit Vogel-Heuser of the Institute of Automation and Information Systems at the Technical University of Munich (TUM), Industry 4.0 has “… a number of definitions, but no single one which is generally accepted. Or definitions are kept so general that they really don’t mean anything. Industry 4.0 is actually a concept with many different facets. It’s just not possible to summarize them all in a single sentence.”
In a Q&A posted at the TUM website,4 she explains what Industry 4.0 is not: something you can buy off the shelf. She notes, “Trade-fair booths will often advertise with a banner saying something like ‘We have an Industry 4.0 PC.’ Utter nonsense. There’s no way a PC as such can be Industry 4.0—it’s just a device with software.”
Vogel-Heuser continues, “At training events I often hear things like, ‘Tell me how Industry 4.0 will work at my company.’ There’s no such answer in general. Every company has to figure out for itself what parts of this plethora of Industry 4.0 components is interesting for the company, for its business, and its customers.”
Others concur with possible difficulties extracting value from Industry 4.0 and the IIoT. Philipp H. F. Wallner, industry manager, industrial automation and machinery, MathWorks, said, “Industry 4.0 and Industrial IoT are huge topics and come up in every conversation or conference. However, we see that many companies have a hard time figuring out which aspects help them create value for their business. For instance, they collect lots of measured data from their machines but then struggle with getting valuable insights from this data.”
Collaboration and demonstration
Demonstrations can help. Vogel-Heuser collaborated with other professors on an Industry 4.0 implementation called the MyJoghurt demonstrator (Figure 1). She explains, “… in a manner of speaking, we have a yogurt jar that knows how it wants to be filled. Let’s say you want mango and strawberries. First you have to find out if that can be produced—that is, are all the ingredients available and can these fruits be processed at the plant in the first place. Then the jars start to move and get what they need: In a sense they’re in contact with the machinery. The idea itself is rather old—but now it can be actually put into practice.” A German-language paper provides background on the demonstrator.5
Courtesy of Eckert/TUM
Collaboration is a key to effective IIoT implementations. “Implementing the IIoT and achieving its benefits can’t be done alone, which is why we work closely with industry leaders and partners to develop IIoT standards and prove out concepts,” said Kyle Voosen, National Instruments section manager for DAQ and embedded product marketing. “In this spirit, NI and 12 other industry leaders recently opened the NI Industrial IoT Lab, which showcases the latest IIoT technologies and provides a collaborative space for partners with different expertise to work on solutions that will change the way businesses operate.”
He said the Industrial Internet Consortium helps coordinate such activities by bringing together industry leaders from around the globe to accelerate the development and adoption of the IIoT. “The consortium plays a strong role in shepherding the development of testbeds that ascertain the usefulness and viability of IIoT technologies, applications, and processes,” he said. “One prominent testbed is the Time Sensitive Networks for Flexible Manufacturing Testbed. Nineteen companies, including NI, Cisco, Intel, and Bosch, are working together on this testbed to ensure network performance and interoperability among vendors.”
Wallner at MathWorks cited two major aspects where customers seeking to implement IIoT solutions can benefit from MathWorks’ MATLAB and Simulink. “The first one is predictive maintenance based on our solutions for data analytics. More and more companies in the industry use MATLAB to design their algorithms for predictive maintenance and integrate them into their IT systems. This allows them to make use of the measured data from their machines and components in the field and create a completely new—service based—business area.”
The second major aspect is virtual commissioning based on model-based design. “Modern production machines as well as components like sensors or servo drives have become sophisticated mechatronic systems with a significant amount of software onboard,” Wallner said. “The Simulink platform allows engineers to model their (software) functionality and their environmental models (including mechanics, electrical components, external influence, etc.) in one single environment and simulate the behavior before the physical system is available. This workflow enables early verification of the machine or component and commissioning of the functionality virtually (in the model) before it runs on an embedded controller or a PLC.”
Customers have been using MATLAB and Simulink for analyzing data and for model-based development and testing of embedded software for many years, according to Wallner.
“Customers in the automotive and aerospace industries and innovative industrial customers like wind turbine manufacturers have developed standards for using simulation models and automatic code generation for building their systems,” he said, adding that the companies choose MATLAB for its capability to work with all kinds of engineering and business data and for its integrated environment for applying advanced analytics capabilities such as machine learning and optimization.
MathWorks offers support for both edge and cloud IIoT implementations (Figure 2). “Algorithms developed in MATLAB can be integrated into the machine infrastructure in two ways,” Wallner said. “The first option is to generate real-time code (for example, C/C++, or IEC 61131-3) that runs on a PLC or on an embedded controller. The second option is to create standalone applications that integrate into the existing enterprise IT systems (for example, a process-control system or cloud).”
Copyright 1984-2017, The MathWorks Inc., Used with Permission
He continued, “In many projects, we see that both options are used in parallel. Typically, the time-critical part of the algorithm (for example, for filtering or cleaning the raw measured data) is integrated into the real-time controller, whereas the more time-consuming operations (for example, machine learning) run on the cloud. However, we are seeing increased interest in running machine-learning algorithms directly on the real-time
controller.”
From cloud to edge
Much publicity around the IIoT centers on the cloud. As Christopher Alessi puts it in The Wall Street Journal, “Germany’s Siemens AG and larger U.S. rival General Electric are duking it out to develop the definitive ‘Internet of Things’ cloud platform for industry.”6 GE is fielding its Predix platform while Siemens offers its MindSphere platform.
But the cloud is only part of the picture. Voosen at NI cited IDC estimates that by 2019 at least 40% of all IoT-created data will be stored, processed, analyzed, and acted upon at the edge. “For nearly 15 years, NI customers have boosted their operational efficiency and lowered costs by deploying NI CompactRIO, and the rest of our software-centric platform, at the edge to obtain and act on insights from their assets,” he said. “When our customers place more intelligence at the edge with CompactRIO, they drastically simplify the link and reduce the data passed between IT and OT infrastructure.” Using real-world measurement data, he added, CompactRIO can locally perform feature extraction, run machine-learning models, and perform nanosecond analytics and control—all at the edge.
Voosen noted that in the IIoT, decisions happen both at the edge and the cloud, with the flow of data normally dictating how customers choose to architect their IIoT solutions. “Maximizing performance and reducing unnecessary data transfer are two primary reasons for pushing decision-making to the edge,” he said. “Closing control loops at the edge is the most obvious way to maximize control-loop speed and reliability in the IIoT. Just as you wouldn’t place the cloud between your foot and the brakes of your car, you wouldn’t expect the cloud to sit within the obstacle-avoidance or safety systems of your autonomous earthmover. These kinds of high-speed decisions based upon sensor fusion need to happen as quickly and reliably as possible.”
He continued, “Until each asset has a dedicated, multigigabit link to the cloud, the edge will continue to serve a role in data reduction, feature extraction, and decision making. For instance, by pushing machine-learning models for predictive maintenance to the edge, an edge node can locally detect an anomaly and determine what the potential impact of the anomaly is on the lifetime performance of the asset without burdening the IT infrastructure. From there it can work, through the cloud, to schedule the appropriate repair/replacement service.” He added, “Flowserve, one of the world’s largest suppliers of industrial and environmental machinery, is working with PTC, HPE, and NI to implement such a solution for their customers today.” Figure 3 shows a Flowserve demo in the NI Industrial IoT Lab.
Courtesy of National Instruments
Opto 22’s IIoT product offerings also begin at the edge, with the SNAP Ethernet I/O system, which bridges the real world with the digital world through a collection of input and output modules designed to connect with virtually any electrical, electronic, mechanical, or environmental device. The company also offers programmable automation controllers such as its SNAP PACs, complete with a built-in HTTP/HTTPS server and RESTful API. According to Newton, “We have what is effectively an edge-processing system built into our programmable automation controllers.”
For visualizing information, the company’s groov platform offers a simple way to build operator interfaces that can be viewed on any screen, from a smartphone to a big-screen HDTV. The platform logs events and notifies you when events occur in your plant, in your remote assets, or within your building.
Fourth industrial revolution
Michael Guckes, HBM Germany product manager, describes Industry 4.0 as the fourth industrial revolution supporting an entire production chain that’s intelligent, self-adjusting, and connected to the Internet. He said HBM, headquartered in Darmstadt, follows the German initiative very closely and is involved in a university project to implement Industry 4.0 into local businesses. “We can already conclude one thing with confidence,” he said. “A flexible, self-adjusting, and super-efficient production chain can only be achieved using high-performance measurement systems.”
The smart factory places high demands on both measurement technology and sensors, Guckes said. Implementations must make exact measurements and provide feedback to control systems, and reliability is critical because measured values may trigger other parts of the production chain. “Measurement technology plays a key role in the fourth industrial revolution and cannot be overlooked,” he emphasized. “The selection of amplifiers and systems for test and measurement becomes a strategic factor for introducing new production concepts because they have to work seamlessly with the rest of the smart factory. Only in this way will it be possible to achieve the promised benefits of Industry 4.0, such as efficient and cost-saving production and good quality control.”
Smart production, he said, requires data transfer and processing in real time, high memory capacities, the capability to visualize measurement data; integrated diagnostic and monitoring options, and simple parameterization. “The challenge is to meet the increasing market requirements with existing methods and tools,” he noted, “because the technology to realize the dream of the smart production chain is already here.”
In addition, Industry 4.0 encompasses more than just production. “Ever faster product cycles mean that product development and production planning also must be part of the initiative around Industry 4.0,” he said, adding that despite the availability of IT tools for simulation, “… the products must still be tested in real life, using prototypes.”
He explained the tests are partly in order to make sure how the products function in a production setting. “But it is also about simulating the product’s entire life cycle,” he said. “A test simulating several years of usage can be carried out in a few hours, exposing the products to mechanical stress, physical stress, and temperature changes as similar to real life conditions as possible. To conduct and evaluate such tests require highly developed measurement technology.”
For Industry 4.0 applications, HBM offers products such as the PMX industrial amplifier—the basis of an intelligent data-acquisition system that is able to monitor and control the entire measurement chain. In addition, browser-based PMX software (Figure 4) eliminates complex and error-prone software installation difficulties while ensuring access to all device parameters for configuration, control, and analysis.
Courtesy of HBM
Customer examples
Both Wallner at MathWorks and Voosen at NI cited some specific examples of successful customer IIoT implementations. “Companies like Baker Hughes7 and Mondi8 are using MATLAB to design their predictive-maintenance applications,” Wallner said. “Using MATLAB for machine learning as well as connecting to different data sources … gave them the ability to convert previously unreadable data into a usable format and to predict the ideal time to perform maintenance.” Those data sources can include databases, cloud platforms, or data conforming to the OPC Foundation’s OPC Unified Architecture machine-to-machine communication protocol for industrial automation.
Voosen cited three customer examples. “London Underground improved reliability for 200 million annual passengers with remote condition monitoring based upon CompactRIO,” he said, by implementing a large-scale distributed system on its Victoria Line to simultaneously monitor assets in real time from a central location. “This system allows London Underground to proactively respond to potential failures and gain better insights into asset lifecycle, which has resulted in improved uptime for the Victoria Line and an estimated 39,000/year reduction in lost customer hours and a £350,000/year savings in passenger disbenefit.”
In addition, he said, “FireFly Equipment designed a smart turf harvesting machine that allows customers to double productivity compared to manual harvesting, harvest 20% faster, and use half as much fuel as other turf harvesters. CompactRIO controls everything in this machine, from data acquisition and the user interface to motion control. As an IIoT solution, FireFly can perform remote diagnostics, push firmware updates, and provide on-the-spot training and instruction to end users.”
And finally, he said, “Jaguar Land Rover implemented a data-management solution to manage and analyze up to 500 GB of distributed test data per day.” A fully automated analysis routine allows engineers to search, inspect, analyze, and report on the data 20 times faster than any previous manual method and increase the amount of data analyzed from about 10% to over 95%. “These improvements allowed Jaguar Land Rover to address more issues before passing final products to the customer, resulting in increased customer satisfaction ratings because the products became more robust,” he said.
Conclusion
The IIoT will continue its rapid pace of adoption, sometimes intersecting with other emerging technologies. For example, at the Hannover Fair in April, Fraunhofer detailed how 5G would unleash its potential in Industrie 4.0 applications and what it calls the tactile Internet. “We’ve already shown in the lab that it’s possible to transfer data at 10 gigabits a second with a latency of 1 millisecond and with absolute reliability,” said Prof. Slawomir Stanczak, co-head of the Wireless Communications and Networks department at the Fraunhofer HHI, in a press release issued in advance of the show. “And we are ready to present solutions that can easily be applied to products in some interesting application areas”—such as reliably controlling robots from afar.
Fraunhofer also sees an important role for edge computing, using cognitive mechanisms at the network-node level, with intelligent algorithms and dynamic learning software detecting faults in advance and respond accordingly. “Let’s say a fork-lift truck in a factory drives past a machine and disrupts the transmission path,” explained Stanczak. “The system knows this in advance and either reroutes the transmission of data or initiates a timely response.”
Yet despite the emergence of fast, low-latency communications and network-node cognition, accurate measurement will remain the key requirement of IIoT implementations. “To realize a vision like Industry 4.0 is not easy,” concluded Guckes at HBM. “However, at HBM we are convinced that test and measurement technology, used in the right way, is one of the most important keys to make industry smarter—with or without a national initiative like Germany’s.”
References
- Industrial IoT Market by Device & Technology, Software, Vertical, and Geography—Global Forecast to 2022, Markets and Markets, February 2017.
- 2017: State of the IIoT, Opto 22, White Paper, 2017.
- Nelson, R., “Test, inspection drive Industry 4.0 at IPC Apex Expo,” EE-Evaluation Engineering, May 2017, p. 20.
- “Many people think Industry 4.0 is something you can buy off the shelf,” Interview with Prof. Birgit Vogel-Heuser, Technical University of Munich, Jan. 24, 2017.
- Mayer, F., et al., Deutschlandweiter I4.0-Demonstrator, Technical University of Munich, 2013.
- Alessi, C., “GE, Siemens vie to reinvent manufacturing by harnessing the cloud,” The Wall Street Journal, March 6, 2017.
- “Baker Hughes develops predictive maintenance software for gas and oil extraction equipment using data analytics and machine learning,” MathWorks, User Stories.
- “Mondi implements statistics-based health monitoring and predictive maintenance for manufacturing processes with machine learning,” MathWorks, User Stories.
For more information
