The New Industrial Revolution promises big things for big manufacturers, but what about the tens of thousands of midsize companies making everything from components to original equipment? Are there enough smarts in “smart manufacturing” to go around for all manufacturers?
The answer is an unequivocal “yes,” according to members of the Smart Manufacturing Leadership Coalition (SMLC), an organization at the forefront of the infusion of artificial intelligence into machines to transform the manufacturing enterprise. The coalition is composed of experts on the subject from academia and major companies like Owens Corning, Rockwell and General Motors. “Even some rudimentary information technology and modeling capabilities can go a long way for small and medium-size manufacturers,” says Jim Davis, UCLA’s vice provost, information technology and chief academic technology officer and co-founder of the SMLC. “The time is right for them to start getting involved.”
Smart manufacturing is described as the convergence of enterprise IT with production IT. In this environment, disconnected and dumb machines no longer bang away at making things, completely segregated from the rest of the enterprise, supply chain partners and the demands of customers. Rather, the plant and the machines in it are integrated with data analytics software in the cloud to create agile, informative and demand-driven supply chains.
In this environment, the entire manufacturing process moves in concert with the rest of the enterprise. “Smart manufacturers are those that have evolved from production processes involving intensive labor to highly automated processes,” says Keith Nosbusch, chairman and CEO of Rockwell Automation. “They’ve gone from isolated plant operations to integrated, responsive supply chains.”
By migrating away from traditional, vertical silos like manufacturing here and supply chain there and customer orders somewhere else, the entire process is realigned through automation. Benefits include faster time to market, greater agility and speed in responding to customer trends, improved asset utilization and predictive machine maintenance, among other gains.
“Any company in the market to buy factory equipment today would be making a mistake not to pay a bit extra and buy smarter machines with microprocessors and an Ethernet port on the back side,” says John Bernaden, an SMLC member and director of corporate affairs at Rockwell Automation. “Even if you have no plan to plug in the equipment right now, some day you will want to do so because the intelligence deriving from that machine will be invaluable.”
This intelligence is gleaned from various sensors, and network-based data and modeling software embedded inside plant equipment to manage throughput, product quality, maintenance, inventory and the supply chain. This information is simultaneously integrated with other enterprise systems, infusing manufacturing intelligence throughout the lifecycle of design, engineering, planning and production.
The supply chain benefits alone are enticing. By digitizing the production process and having this information flow to and from the plant and the extended enterprise, smart manufacturers can provide real-time visibility to suppliers so that the exact quantities of a particular part or component are delivered by a certain date, allowing for minimal excess inventory. “Systems simply know how to produce whatever is ordered, whenever it’s ordered,” explains Michael Yost, an SMLC member and president of MESA International, a manufacturing membership organization focused on driving greater productivity. “The workflow integrates all responsible personnel in real time, regardless of where they exist across the globe.”
By hinging the supply chain to real-time and forecast requirements right at the manufacturing location, companies have greater flexibility to match production with demand, as well as more agility in delivering orders of varying order sizes. Any defects or drifts in a part’s tolerance also can be corrected immediately, reducing disruptive incidents caused by order changes.
Grabbing Manufacturing by the Antlers
This expanded visibility into—and control over—production processes is evident at Moosehead Breweries, an early, midsize company that converted to smart manufacturing. The Saint John, New Brunswick, Canada-based maker of Moosehead beer has invested $35 million to date in a capital-investment program to improve manufacturing efficiency, beginning with its bottling plant. At the facility, bottles are pasteurized, filled, labeled, capped and packaged along a continuous loop of conveyor belts.
The various machines along this loop are embedded with sensors coupled with data analytics software to improve quality, increase throughput, discern bottlenecks and reduce costs. Previously, human beings determined when to increase or decrease throughput. The shift to intelligent machines has enabled a 50 percent reduction in the brewery’s labor expenditures.
The plant’s smart machines are integrated with Moosehead’s enterprise data systems, ensuring that the bottling operation (and eventually the company’s other plants) are linked to procurement, supply chain and financial enterprise resource planning (ERP) systems, thereby maximizing knowledge across the business. “We went from not a whole lot of automation to a huge amount,” says Luke Coleman, Moosehead brewery automation and controls specialist.
Not every company has to implement a huge amount of automation. A first step in evaluating smart manufacturing is to determine factory pain points—the manufacturing problems weighing down the organization’s value proposition to customers and affecting its competitiveness. (See Sidebar: “Healing the Pain Points,” right). “Making this assessment should be a cross-functional team directed by a project champion,” says Denise Swink, CEO and chairman of the board of SMLC and an independent consultant. This integrated team should be peopled with experts from operations, IT, procurement, quality control, supply chain and engineering to study and discuss the manufacturing complications the organization currently confronts.
“Perhaps the company isn’t managing the plant’s performance to an integrated set of Key Performance Indicators (KPIs) for a crucial step in their operations; but instead, [it] is managing just to productivity and cost,” Swink explains. “Once the team agrees to identify and manage to an integrated set of KPIs then data prioritization, targeted sensor infusion, and the selection of smart systems and intelligent machines can help you do that.”
This optimized plant environment better manages such factors as energy, water, quality, safety and yield to achieve desired outcomes. “Let’s say, for a certain customer, the issue is quality,” Swink explains. “You can manage the plant to improve quality by taking a minor hit, say, on energy or water costs involving this customer. These are informed decisions… It all depends on the company’s particular goals, which in most cases are driven by its customers.”
Moosehead launched its smart process with a cross-functional team in charge. Wayne Arsenault, VP of operations and human resources, says the group initially benchmarked competitors of similar size to discern possible performance improvements, only to learn that a sensor here and there would not provide the broad, competitive advantages the company sought. “Simply tweaking the plant was not an option; we needed to make a fundamental shift in how we utilize our assets, which required a multi-year investment in the physical transformation of our bottling line,” he explains.
This transformation involved the use of sensors on the conveyor line to determine a jam in the flow of bottles. “We use photo cells similar to the sensor preventing a garage door from coming down on your car to see when bottles are bunching up,” Coleman explains. “The operator receives a red or yellow light on the dashboard telling him there is a problem at a particular juncture along the conveyor line, and [he] can then slow it down. Previously, we’d have to shut down the entire system and have someone physically inspect the problem.”
Sensors accompanied by real-time data analytics have long been used in military and commercial helicopter engines to predict failure risks, notes Phil Shelley, former CTO at Sears Holdings and current president of Newton Park Partners, a data analytics firm. “Obviously, the failure here can be catastrophic. By embedding sensors and real-time computing techniques to analyze vibration and other performance metrics, vital repairs can be done before it is too late.”
At GE, massive physical machines like wind turbines are embedded with highly sophisticated sensors to determine their internal health. The sensors produce real-time data on water temperature, oil pressure, vibration and microscopic metal fragments in the ambient environment. Predictive data analytics systems model this wide-ranging data to determine the machine’s operational efficiency. “Obviously, knowing weeks, if not months, in advance that a wind turbine will need to be out of commission for a repair is a lot better than finding out today,” says Don Busiek, general manager of manufacturing operations management software at GE Intelligent Platforms.
Sensors also drive quality enhancements and waste reduction. Shelley cites the use of sensors in the injection injection-molding business, where the porosity of plastic can cause a product failure. “By using sensors and algorithms to optimize pressures, temperatures and other parameters in the injection-molding machine, you can ensure that enough plastic is injected at the right pressure to eliminate the risk of cavities,” he explains. “This improves the quality of the molded product and reduces waste, and it all happens in real time.”
All this sounds too good to be true; and in a way, it is. Companies don’t simply flip a switch and become “smarter.” Challenges abound, chief among them is a dearth of “skilled workers or ‘smart’ people who know how to use all these IT-driven machines,” Bernaden says.
Swink cites another impediment. “One of the most daunting tasks in any manufacturing operation is to really know and understand the data you already have, use it effectively to make more informed decisions, and then determine what else you should collect and know about,” she says. “The compartmentalization organizationally and process-wise is a huge obstacle in doing this.”
Some organizations are holding back for the time being on intelligent machines until their workforce and organizational structures are ready. “Talking with a CEO of a $60 million metal-parts fabricator recently, I asked him why he didn’t buy one of the new, automated welders that could increase his output with higher-quality products,” Bernaden says. “He said he couldn’t find someone with the combined programming skills and welding skills necessary to operate it.”
He adds, “Not many welders have IT skills, too. But that’s what it’s going to take to program, operate and maintain these smart machines in smart factories.”
Nevertheless, there are plenty of reasons for companies to make the leap now, despite the hurdles. According to a December 2013 survey of manufacturers by the American Society for Quality, only 13 percent have implemented smart manufacturing within their organizations. However, of the ones that did, 82 percent gained greater efficiency, 49 percent reported fewer product defects and 45 percent increased customer satisfaction.
The payoff may be well worth it.
Photo by The National Archives UK