Delivering Enhanced Insights and Remote Access, IIOT Adoption Skyrockets

Editor’s Note: For more information on the IIoT, predictive maintenance, and how they can help organizations thrive in this turbulent time, register for our free webinar, Predictive Maintenance in the Time of COVID-19.

The novel coronavirus pandemic has changed our world in many ways. In the case of industrial automation, the pandemic has dramatically increased the adoption of digitalization in the form of the industrial Internet of Things (IIoT). The technology is not new. The pharmaceutical, semiconductor, and medical device sectors, as well as many process industries, have developed digital data gathering and analysis to an art. In other sectors, particularly in discrete manufacturing, manual readings taken on paper still dominate. That state of affairs is poised for change, however.

The IIoT is a network of sensors and devices that feed data to a central platform for analysis and visualization. It is combination of intelligent (and not-so-intelligent) devices connected together to create an interoperable ecosystem. The IIoT provides real-time information on how the equipment is operating: whether it needs repairs, whether the machines are running at full efficiency, whether consumables need to be replenished, and more. Users can view dashboards presenting key performance indicators (KPIs) or drill down to data as granular as gearbox temperature, motor vibration, or drive current. Once the data is in the cloud or the enterprise network, it can be further distributed, interfaced with manufacturing execution systems (MES) or provide the raw material for big data analytics. The result is improved operational equipment effectiveness (OEE), better supply chain management, and more effective operations and maintenance.

One of the key benefits of the IIoT is its ability to serve up information from any networked device to any permissioned user located anywhere on the globe. As essential businesses during the pandemic, manufacturers are trying to operate while dealing with work from home orders, social distancing, process changeovers to produce pandemic-centric products, and reduced staffing due to sick leave or layoffs. Hopping on an airplane to service faulty equipment is no longer an easy option. In this environment, interest in IIoT deployments is skyrocketing.

Figure 1: the IIoT can present data and analysis from a network of sensors and devices to any web enabled device anywhere in the world. (Courtesy of Inductive Automation)

“Manufacturers were thinking about IIoT long before [COVID-19],” says Craig Resnick, VP at ARC Advisory Group (Dedham, Massachusetts). “A lot of that was driven by workforce transition with baby boomers retiring and being replaced by engineers that hadn't yet accumulated the skills and expertise to maintain the machines in the plant.” End users were interested in digitalization, but only to the extent that it enabled OEMs to offer remote maintenance expertise. The advent of the pandemic changed that calculus. “Now, it's not just being driven by people retiring, it’s being driven by people who either can’t go on the factory floor or are resistant, or even just by social spacing requirements,” he says. “Engineering and management have to be able to work remotely from anywhere. They need real-time access and visibility, and supply-chain flexibility so that they can make rapid decisions.”

And those needs can be served by the IIoT.

Figure 2: The IIoT brings comprehensive information and analysis about the equipment to the engineer. (Courtesy of Inductive Automation)

IIoT 101

In many ways, the IIoT is the logical evolution of networking. Communications began with connecting physical locations across hardwired networks, first as telephone to telephone and later as computer to computer. The next level of progression was the connection of person to person over wireless handsets. The IIoT represents the latest evolution in communications, connecting device to device over a hybrid network consisting of wireless and wireline connections.

The devices that make up the IIoT are smart components with onboard processing power and memory. Beyond that, the architectures very widely.

Where are the readings generated?

• Systems with external sensors for monitoring components, e.g. temperature sensors, vibration monitors, and current sensors.
• Components with integrated sensors such as drives with onboard temperature or current monitors

Where is the data aggregated?

• Dedicated data loggers that collect input from sensors and components.
• Components with integrated data loggers such as HMI capable of storing data in a rolling buffer so that when the system defaults out, technicians can review the conditions immediately prior to the event

Getting the Data From the Nodes to the Cloud
The IIoT is only useful to the degree that the data gathered can be converted into actionable information. “The point of digitalization is to get the right information to the right person at the right time, and that includes machines,” says Bernard Cubizolles, senior marketing manager, Digital Plant team, GE Digital (Foxboro, Massachusetts). Where the data is stored and analyzed directly affects its effectiveness. A wide variety of architectures, technologies, and products exist to accomplish this task, as do the capabilities of different products. As with most things in engineering, the optimal choice depends upon the specific situation.

In the most basic architecture, sensors/devices (nodes) can send their data to a server in the enterprise network or the cloud. Alternatively, the data could first go to a server located on the factory floor for backup and storage in the event of a network outage. There, it may be housed in a so-called data lake, which is a repository of aggregated raw data. By including all of the data available and leaving it untouched, data lakes maximize the options available to the data scientists. They are undifferentiated, meaning they can even combine sensor data with customer data, a major departure from conventional database formats. On the downside, all of that data needs to be stored somewhere and there’s a cost of management. Also, the value of data degrades over time because the actionable time might pass before the data is evaluated.

Probably the biggest issue with the simple format is the sheer volume of data involved in IIoT systems, particularly raw data from the sensors. Sending that data costs money, a particular disadvantage when so many of the readings are steady-state. These factors have led to the emergence of edge computing for IIoT applications.

In edge computing, the data undergoes a certain amount of preprocessing on premises at the network edge. The closer to the source of the data this processing occurs, the thinking goes, the more efficient the use of resources. The data volume is significantly lower, reducing latency, bandwidth demand, and storage requirements.

Edge processing occurs in a device known as an edge device, or sometimes an edge gateway. This can be a server or simply a microprocessor. A more effective version of an edge network inserts an aggregation layer between the sensor/device nodes and the edge devices. IIoT build outs can incorporate hundreds, or even thousands of nodes. Adding a layer of gateways aggregates data more efficiently.

With noting that implementations today can be quite different. The edge device can be a separate box or can be a microprocessor embedded in another device such as a device's PLC. In other cases, the edge devices combined with the router. Alternatively, existing automation layer can be used to perform the edge work, so no extra investment may be needed from an end-user perspective. When it comes time to start a pilot project (see below), be sure to consider your options for this very important aspect of the topology.

IIoT Use Cases

Remote Monitoring and Predictive Maintenance
Sensors are part of the fundamental fabric of the IIoT. The data they generate can provide valuable insights into machine condition. An overheating gearbox could indicate lubrication breakdown. A sudden increase in motor vibration could be caused by a bearing-cage defect. The IIoT builds up a history of asset behavior over time, both to establish a baseline and, eventually, to characterize the lifecycle of the asset.

Figure 3: Smart Drive with built in friction and vibration sensors detects problem with ballscrew actuator. Note the blue circle in the upper left corner, indicating an alert. (Courtesy of Mitsubishi Electric Automation)

When real-time sensor data departs from baseline, it could indicate a developing problem. The system can be configured to send an alert if the changes exceed some threshold. At that point, a decision needs to be made. “Predictive maintenance is being able to look at those changes and predict when this will become a problem,” says Daniel Zachacki, Senior product marketing engineer, Mitsubishi Electric Automation (Vernon Hills, Illinois). “It’s a question of how well you understand that particular portion of the machine.” The history built up by the IIoT makes it possible to understand when a component is approaching end of life and requires quick action.

The beauty of predictive maintenance is that it enables maintenance and operations to respond in the way most convenient for their schedule and production goals. Rather than being hit with unscheduled downtime, they can delay repairs until the end of the shift or the end of the week. They may even be able to use the continuous monitoring of the IIoT to run the component to the next scheduled maintenance shutdown. Early warning gives them time to order replacements and lineup any skilled technicians or specialty equipment necessary. Downtime drops, OEE rises, and cost of ownership improves.

Product Quality
The fully integrated picture of the operation enabled by the IIoT makes it possible to identify the source of product issues and ensure consistent production values across the enterprise. Cubizolles points to an automotive manufacturer that struggled with cracks developing in the middle of the body parts. The company gathered data from the equipment and integrating it with inspection results and supply-chain data to provide a full picture of the process.

“Our engineers went to work on it,” says Cubizolles. “At some point, it has nothing to do with the products. It's all about physics. It's all about analytics. You have to solve it first but then you want to make sure it doesn’t happen again, and this is where the software comes in. You need a real-time view of this machine. It’s not as glossy as AI, but just digitizing procedures has tremendous value for the customer because it allows for repeatability, making sure that quality issues will be fixed because the job will be done in a similar manner all the time.

Performance Consistency
One of the challenges in manufacturing is maintaining product consistency and optimal machine performance from location to location, machine to machine, shift to shift, and even operator to operator. IIoT analytics can compare the data from the high-performing operator, shift, etc. against their counterparts to determine which factors are responsible for the differences. “The IIoT solution would identify those variables and make them highly visible,” says Chris Baldwin, VP of product management at PTC (Boston, Massachusetts). “It’s easy because it has captured and studied what the top-performing operator does and what the top-performing shift does. Then it’s very easy to standardize that in other environments.”

He points to a customer that invested around $100 million in machine lines across multiple facilities. When operations started, there were clearly problems across the board. An average of one unit in three was so far out of spec that it had to be scrapped. Perpetually recalibrating the machines didn’t help. The company used an IIoT solution that incorporated machine learning to troubleshoot a test machine. It diagnosed and corrected the issues – for that piece of equipment, in that environment. “The model retrains and learns as it goes in the local operating environment, but certain KPIs and other data that were pertinent or heavily weighted in one environment may or may not be in the others,” he says. Instead, the software was able to update at each location using data from the local IIoT. The increased throughput amounts to multiple millions of dollars of production per machine. “The net result now is about 75 additional units produced per machine per day which is very, very substantial for their business,” he adds.

The Digital Twin
A digital twin is a detailed, evolving virtual model of a real physical machine, production line, or even factory. By applying machine learning and other sophisticated algorithms to both historical data and real-time readings from the IIoT, the digital twin becomes a highly accurate virtual replica that wears in the same way as the physical system. The digital twin can be used to explore designs, speed commissioning, predict component failure, and test out new control algorithms, to name a few. The IIoT provides constant updates in an easily consumable format.

Augmented Reality
Virtual reality might get all the headlines right now but augmented reality (AR) is quietly making a difference in the industrial landscape. Augmented reality involves overlaying information on the field of view. A maintenance technician looking at a motor through a head-up display, for example, might see temperature readings or drive current and voltage superimposed over the device. The data could be as simple as part name and serial number or a detailed rendering drawing of the contents inside a black box.

Supply-Chain Management
Although it does not involve motion control and is only peripherally related to automation, we would be remiss not to consider supply-chain management as an IIoT use case. The technology is growing increasingly important, as manufacturers lobby for more granular visibility into the location of raw materials and subassemblies. Real-time updates improve inventory and support more effective production scheduling.

Figure 4: Augmented reality overlays data from IIoT system onto images seen. (Courtesy of PTC)

Getting Started

Despite the many benefits of the IIoT, the digital transformation can seem overwhelming. Here we list a few tips for getting started:

Have a Plan
Success starts with a strategy. Don’t launch an IIoT initiative because it’s the buzzy thing to do. Review your operation. Identify your biggest pain points. Consider ways that the IIoT could help you address them. What data will you gather and how will you use it? What resources do you have to support the initiative?

Start Small
Don’t try to instrument the entire factory or line all at once. This is the surest way to wind up drowning in data without results. Start with a pilot project. Get experience with the technology, with from installation to data capture and analysis. A modest project will simplify approvals. Any problems will take place on a small stage, enabling you to learn what works and what does not.

Look for Low-Hanging Fruit
Choose an asset likely to quickly demonstrate the value of the technology, for example through predictive maintenance. Likely candidates include as assets with a history of failure or those that fault out frequently. A quick win will demonstrate rapid ROI. It will also give you an opportunity to learn more about the system.

Know What Success Looks Like
Establish quantitative objectives – cut downtime by 20%, increase throughput by 10%. Meeting goals will provide a demonstration of benefit and make follow-on projects easier to justify.

Don’t Forget the Operators
Be sure to consult the people on the production floor, who work with the equipment day after day. They have a tremendous amount of institutional knowledge in terms of how the equipment works, how to maximize performance, frequent issues and trouble spots, etc. Make sure that your maintenance team buys in, as well. Carrying off the digital transformation, especially when applying predictive maintenance techniques or comparing performance among operators, machines, and shifts requires communications from leadership and buy-in from the participants.

Avoid “Pilot Project Purgatory”
You may be starting small but the ultimate goal is to implement the technology across the organization. Choose a scalable solution. Ensure that you have sufficient bandwidth to expand from a local level to an enterprise-wide application. Bring in IT from the beginning. “If you do your homework in terms of architectural capabilities and technology offerings, you won’t wind up in pilot project purgatory or Gartner’s trough of disillusionment,” says Don Pearson, chief strategy officer at Inductive Automation (Folsom, California). “Think through your architecture even before you start your pilot project. Choose solid building blocks that can scale.”

The most essential point, according to Resnick, is to just get started. “One way or another, companies are going to have to go through the digital transformation process,” he says. “Even if they survived the first wave of the pandemic, the odds of them surviving the next one or any other global economic disaster are very, very low if they have not begun the digital transformation process. So, digital transformation is no longer optional, it’s really mandatory. Don't feel as though you have to transform into a completely digitalized plant immediately, but if you have not begun the proess yet, the time to start is now.”

Mitsubishi’s Zachacki puts it most simply. “The IIoT is here today,” he says. The need for it is increasing. Educate yourself and familiarize yourself with the concepts and technology because it could be of immediate benefit to your company.”