Implementing Industry 4.0 in Three Key Steps
Smart Manufacturing Engineering Magazine May 2019
By Ed Sinkora
So far, Industry 4.0 is talked about much, but implemented little. Here, experts discuss lessons learned in introducing it to gain greater efficiency in manufacturing enterprises.
If you’re feeling a bit overwhelmed by the prospect of Industry 4.0, and perhaps a bit sheepish about your lack of progress, you’ve got good company. Global accounting network BDO USA, Chicago, recently polled 230 executives at middle market U.S. manufacturers—with annual revenues between $200 million and $3 billion—and just 5 percent are currently implementing an Industry 4.0 strategy.
Industry 4.0 describes a manufacturing environment in which all your equipment is networked both internally and with other devices in the supply chain outside your factory. Smart software analyzes data gathered from these sources and reports on production. Artificial intelligence then automatically allocates resources, buys supplies, or schedules maintenance.
That’s an attractive goal, so how do you get there?
Step 1: Connectivity
The first thing required for Industry 4.0 is digital connectivity between machine tools and the computers that analyze machine data. If you have newer equipment, this is a non-issue since it probably has connectivity built in. Older machines often lack this connectivity and are limited in the data they can produce. But according to Eskander Yavar, co-leader of BDO’s Industry 4.0 practice and national leader of its management and technology advisory services practice, “there’s a whole market retrofitting machines from the late 1990s and early 2000s with connectivity.”
Joe Gazzarato, director of engineering for FANUC America Corp., Rochester Hills, Mich., said his company has customers running robots made in the 1980s and he questioned the ROI of adding Industry 4.0 capability to robots that are 30 years old. “It may be easier to justify the expense if you’re talking about a custom machine tool, especially since there are MTConnect adapters to connect older machines. [These are] boards you wire in to extract data through I/O and other means and get them on an MTConnect network.”
Also, the R&D tax credit can be applied to Industry 4.0 investments, stated Yavar. He pointed out that with the cost of sensors coming down and the fact that most software vendors use a pay-as-you-go model, “barrier to entry has been reduced greatly.”
Wired networks seem to be the preferred approach. Furtado sees wireless data transfer playing a greater role in two possible cases: Smaller job shops and factories with lots of inexpensive machines, “where the cost to connect them via wires would be high.”
Wired or wireless, any network must offer both cybersecurity and “industrial robustness.” Sticking with the FANUC example, its Zero Down Time (ZDT) application connects over 20,000 robots in 16 countries
to send data to its ZDT Data Center in the cloud. To protect that data, FANUC partnered with Cisco Systems. Gazzarato credited that deal with giving customers “a lot of confidence in connecting with us.”
There is also the question of how best to structure the data for export and transmission. MTConnect is popular in North America, while Europeans (who are arguably farther along in Industry 4.0) favor OPC UA. Furtado said MTConnect is “limited to a few variables and thus limited in functionality and capability.” Bottom line: It’s not something that has been standardized, so you have to make a choice, get going, and continue to watch developments. Even given his critique of MTConnect, Furtado said it is important, “because you have to start somewhere.”
Step 2: Low Hanging Fruit
Once you have the necessary communications infrastructure in place—and it doesn’t have to cover everything at once—the next step is to implement a few Industry 4.0 functions that offer good payback for the effort.
That will help get management buy-in for additional Industry 4.0 investments and guide where they will be implemented. As Furtado put it, “the key applications revolve around productivity. Beyond that are key process indicators, like time to market.” He believes machine monitoring and preventive maintenance are the most widely used applications.
“People want to reduce machine downtime and they want to monitor not only what their machines are capable of, but also what their operators are doing and how effectively they’re using the machines,” said Furtado.
The reference to “applications” here is critical, because just pulling data out of the machines tells you nothing. You have to interpret the data to derive meaningful information, and that requires a good software program. There are already a number of such products, developed by subject matter experts and ready to use.
Launched in 2015, FANUC’s ZDT monitors robots remotely and reports information on a robot’s health, maintenance needs, and utilization rates. It does this automatically and while a company is running production, according to the company. Gazzarato reported that, to date, ZDT analytics have identified potential equipment issues and notified customers to take corrective actions to avoid unexpected downtime in over 600 cases, saving companies more than $75 million.
“But ZDT also does what we call condition-based maintenance,” Gazzarato said, “which goes one step beyond predictive maintenance. We look at the way a robot is being used and we can predict when maintenance should be performed.” This includes things like changing the grease on the drives or inspecting the cables that go on the arm. “For example, if the robot isn’t being used aggressively, you may be able to delay your maintenance routines. Conversely, if it’s being used aggressively, we may tell you to check the cables or the grease sooner,” he said.
Note that “aggressive use” doesn’t just refer to hours of operation. ZDT also tracks things like peak torque on the robot’s drives and how much energy the robot is using. By combining data from multiple sensors, applications like ZDT can make judgments about machine condition, according to the company. In one recent example described by Gazzarato, ZDT identified a potential problem with a customer’s robotic weld gun based on measured peak torque and disturbance torque as the gun closed on each operation. It turned out to be a missing bolt holding an adapter to the jaw, he said.
Siemens recently purchased the portfolio of OMAT Ltd., the Jerusalem-based makers of Omative adaptive control monitoring and vibration monitoring systems. Furtado said the software can be installed on the control or connected via a third party.
“The adaptive control and monitoring system (ACM) looks at the load during machining and adapts the feed rate automatically,” he explained. “This decreases the cycle time and increases tool life because it’s maintaining a constant tool load. It also improves the surface finish.”
The vibration control monitor (VCM) system does a frequency-based analysis to recognize chatter and can be used to optimize various machining conditions, he said. Both systems can perform these functions on the machine control without external connectivity, or Siemens can move this and other data to the cloud for further analysis via its Manage MyMachine function.
In the grinding world, abrasives manufacturer Norton|Saint-Gobain, Worcester, Mass., offers the Norton 4Sight process monitoring and diagnostic system. It collects spindle power load and other grinding parameters to provide insight into troubleshooting problems and optimizing grinding processes. That includes boosting wheel life, workpiece quality, and system productivity, according to the company. The system offers instant notifications, real-time dashboards, and historical analytics reporting. Saint-Gobain says Norton 4Sight easily integrates with any machine via MTConnect or OPC UA.
From there 4Sight can integrate fully to CNC controllers like FANUC FOCAS, Mitsubishi, or Siemens. This list can be expanded with some development time, and with older machines 4Sight is compatible with hundreds of PLCs, according to the company.
Step 3: Broaden the Scope
Once machines are connected and providing better insight into utilization rates, and perhaps tied to apps like ZDT or ACM, it’s time to “drill down to ask more detailed questions,” said Furtado. That can mean more apps, more sensors, connecting more machines, and expanding the scope of connections to more partners in the supply chain. Most of the data collected for the systems described so far comes from the digital feedback inherent in modern CNCs: things like the control knowing when the spindle is on, or the torque load on each drive. This digital feedback predates Industry 4.0, but connecting the data with outside resources like smart software increases its value.
Likewise, adding sensors to the machine can pay off, and DMG Mori, Hoffman Estates, Illinois, has addressed this. Partnering with the Schaeffler Group, Troy, Mich., DMG Mori added 60 sensors to its DMC 80 FD duoBLOCK machining center to gain real-time knowledge about the machine’s performance.
For example, the linear guideways in the X and Y axes have acceleration sensors that measure vibrations to collect reference data and define trends, set tolerance thresholds, and control lubrication consumption. The ball screw nuts have axial and radial vibration sensors. Bearings for the linear axes have axial force sensors. The rotary table has sensors for temperature, vibration, and grease.
All this data feeds into DMG Mori’s CELOS Condition Analyzer app, which visualizes, analyzes, and predicts machine status well enough to schedule maintenance. It performs this function for individual machines, across a factory, or across geographically dispersed plants via the cloud. A separate app called Performance Monitor captures and analyzes overall system efficiency while taking into account machine availability and component quality, according to the company.
Benefits in the Cloud
One big benefit to the cloud is the ability to combine data from large numbers of machines, even from machines dispersed around the globe. The more data you can amass, the more precise machine learning algorithms can be in predicting maintenance needs and other actions. There are three cloud-based apps running on MindSphere: Manage MyMachines, which shows historical and alarm data and to some extent allows for third-party connectivity; a “classic” OEE app that displays effectiveness based on a company’s own production criteria (KPIs); and Manage MyMachines Remote, which allows remote access to a machine.
“[This includes] its PLC/NC data and the possibility to transfer files to the machine using the secure connections. MindSphere can also connect to other enterprise-level systems like ERP systems,” said Furtado.
We’ve already touched on future capabilities, and it’s easy to envision some of the near-term improvements. FANUC’s Gazzarato pointed to widespread adoption of condition-based maintenance and gaining a clear handle on “real operational usage data and energy consumption data. We’re already doing it with ZDT, but we’re just touching a percentage of the robots out there. In the future, I think you’ll see it on all equipment. [With] Industry 4.0, you should never have a piece of equipment go down unexpectedly. You should know the health and needs of your equipment and you should be able to take proactive action.”
Even things like automated parts ordering based on machine data is doable and will occur. The question is when, because as BDO’s Yavar sees it “we’re at stage one of ten” on the journey to establish a universal infrastructure of data and nomenclature that can connect disparate types of equipment and business functions (like ERP).
Yavar and others also think Industry 4.0 will lead more manufacturers to offer performance-based rather than product-based contracts because they can monitor and maintain the performance of their machines remotely. “The hard part isn’t the technology. The hard part is that everybody is doing Industry 4.0 in their own way and we’re spending all this time building connectivity layers.” But don’t despair. The benefits are real and coming to those who dare.