By Russ Banham
It wasn’t long ago when “quality control” consisted of expert personnel whose task was to measure, weigh, touch, and scrutinize finished goods to discern evidence of possible defects. While many manufacturers still rely on such expertise, more are turning to automated “smart machines” in the Industrial Internet of Things (IIoT) for these appraisals.
Machine intelligence is essential for success in the digital future. According to a 2017 study by management consultancy Bain, the IIoT market will generate approximately $85 billion annually by 2020. To remain competitive in today’s rapidly-changing business environment, manufacturers must digitally transform how they make and distribute products.
A 2017 report by Deloitte affirms this view. “The concept of adopting and implementing a smart factory solution can feel complicated, even insurmountable,” the report stated. “However, rapid technology changes and trends have made the shift toward a more flexible, adaptive production system almost an imperative for manufacturers who wish to either remain competitive or disrupt their competition.”
A smart machine in a manufacturing context is a piece of factory equipment embedded with IoT sensors that calibrate and communicate performance issues over the internet to manufacturing control rooms and even other machines to drive faster and smarter decisions. The benefits include instant alerts of a machine that is wearing down and in need of maintenance or repair; self-correcting machine adjustments to address variations in tolerance; and real-time insight into a manufacturing slowdown at a major supplier to rapidly shift production to other suppliers.
“The primary value of smart machines is their ability to produce products of the highest quality,” said Dean Bartles, president of the National Tooling and Machine Association, a trade group representing the precision manufacturing industry. “The sensors inside the machines basically do what seasoned quality professionals have long done, only much, much better.”
High product quality is good news for a manufacturer’s top and bottom lines, as it results in more satisfied customers and fewer defects and production scrap (leftover materials that add to costs). “With smart sensors measuring vibration, temperatures, moisture and dimensions, you’re able to tell pretty quickly when something is off,” said Bartles, an industrial engineer and former vice president and general manager at three General Dynamics manufacturing plants.
“The machine itself tracks the `drift’ in the tooling as it occurs,” he explained. “Some smart machines can even intervene on their own to bring the ideal configurations back into line—self-correcting without the need for human intervention.”
Self-correction capability, for example, could be found in an injection-molding machine, in which specific molten materials like metals and plastics are injected into a mold for use in fabricating a part or finished good, everything from surgical devices and electrical circuit boards to more prosaic toys and automotive intake manifolds. Once the mold is made, products are produced on a constant, uninterrupted basis. However, the process isn’t perfect; defects like blistering, cavities, and contamination by foreign materials are common problems.
By embedding visual, temperature, and weight sensors into the molding machine, imperfections can be identified as soon as they become evident. If the defect is determined to be the presence of cavities—caused by an inadequate volume of metals or plastics being pumped into the mold—the data analytics will direct a computer inside the machine to increase the volume to the correct level.
Companies like Marvin Windows and Doors are at the outset of developing tomorrow’s smart factories today. At a company plant in Oregon that manufactures different-sized wood pieces, Marvin has installed computer numerically controlled (CNC) machines that use laser sensors to give operators an inside look at a board’s knot and grain structure before cutting it into smaller pieces of wood. The sensors help ensure maximum yield, in this case, the largest piece of wood possible from a single block of lumber.
“The sensor inputs data into a computer inside the machine that analyzes the visual image,” said Jim Macaulay, CFO of Marvin Windows and Doors. “Based on this information, the machine knows the optimal cut to make, increasing production yield with less human intervention.” Previously, Marvin had relied on the eyes and experience of shop foremen to identify possible defects.
Internet-connected sensors have also been embedded inside other factory equipment at Marvin plants to measure temperature, vibration, moisture, and other conditions, Macaulay noted. If a motor inside a machine exceeds a specific temperature threshold, this information travels over the internet to a central location for corrective actions. By connecting the factory equipment together in what is known as the Industrial Internet of Things (IIoT), a problem can be self-corrected—one machine picking up a troubled machine’s production tasks. “As a result, we’re able to reduce the chance that one machine’s failure will result in a production stoppage,” said Macaulay.
Precision Is Paramount
Other manufacturers are beginning to seize similar value. “By using sensors to measure and report on diverse conditions, and then collecting all this information in one place for analysis using algorithms, manufacturers are able to draw rapid conclusions on remedial actions,” said Alex Reed, cofounder and CEO of Fluence Analytics, a manufacturer of industrial monitoring solutions that produce continuous data streams.
The algorithms ferret out correlations and non-correlations in the diverse data produced by different sensors, indicating machine wear and product quality issues well before they result in batch failures. While many smart machines can self-correct a problem, more complex manufacturing processes still require human beings to intervene.
“It’s not as straightforward as it often is portrayed to be, though we are definitely headed in a direction where the use of AI [artificial intelligence] and machine learning will direct a correction in a machine based on the findings of the analytics,” Reed said.
Both he and Bartles are members of the Smart Manufacturing Leadership Coalition (SMLC), a nonprofit group comprised of major companies like Rockwell, General Motors, and Owens Corning that works to develop the world’s first smart manufacturing open technology platform. The hope is that midsize and smaller manufacturers can use the open source platform in developing their IIoT strategies.
Abetting these aims is the increasing sophistication of sensors at lower price points. “In our work, we use spectroscopic, infrared and optical sensors to determine viscosity at the molecular level,” said Reed. “For example, we’re able to discern the composition of a material like a polymer. If the composition is off even a little, it can result in [product] failure, waste, and production downtimes. Nowadays, virtually everything in the supply chain begins with sensors.”
Bartles shares this opinion: “Sensors are the modern equivalent of a `red flag,’ giving you insights into possible manufacturing hiccups so your supply chain doesn’t fall behind schedule. More and more OEMs [original equipment manufacturers] are receiving sensor-produced data over the internet from their key suppliers’ machines. This information is extremely insightful for decision-making purposes.”
He recalled how this remarkable capability contrasts sharply with his earlier career at General Dynamics. “Like other manufacturers, we’d receive daily status updates from our suppliers on part counts. But suppliers sometimes don’t tell you the truth,” Bartles said. “Ideally, you want real-time accurate information. This way you know if you need to turn the knob off on one supplier that’s having trouble [in order] to turn the knob on at another supplier.”
Smart machines also offer a more advanced way to trace and track the product quality of all suppliers linked in the supply chain, ensuring each component of a finished product’s specifications has been validated for quality specifications, from the raw material through varied production stages to customer delivery.
No Turning Back Now
Given these myriad benefits and the wider profit margins that can accrue, the production experts anticipate growing demand by manufacturers of all sizes for smart machines down the line.
“We’re not yet at the point where this is widely adopted and delivering massive value; in many manufacturing environments, work still needs to be done by quality control experts,” said Reed. “However, these individuals tend to hail from older generations and are soon to leave the workforce; this puts the onus on companies to invest in the IIoT sooner rather than later. There’s no question that manufacturing is migrating from qualitative assessments by people to quantitative assessments by machines.”
Like everything in business, early movers and their fast followers generally have a leg up on competitors. While smart factories may seem like a fantasy torn from a Buck Rogers novella, they’re how most things will be made now and into the future.
Russ Banham is a Pulitzer-nominated financial journalist and best-selling author.