How Intel's Using the Smart Factory to Improve the Supply Chain

Significant change is coming to the manufacturing arena. As artificial intelligence, the industrial IoT and the smart factory begin to transform production, Intel Corp. is gauging the impact such technologies will have on manufacturing and the supply chain -- and the humans that oversee these complex operations.

The benefits of smart factories are many: autonomous production, hyper-agility, real-time analytics and operational transparency, according to Intel. But transitioning to the factory of the future takes more than just technology. Industry 4.0 requires the co-evolution of workers and manufacturing operations, according to an aptly-named Intel study.

“We wanted to uncover pain points, desires, concerns, and expectations of these individuals as they and their companies pursue the promise of the intelligent factory,” said the paper’s co-author Dr. Irene Petrick, market innovation director for the Internet-of-Things Group at Intel. “The supply chain was among these pain points.”

For example, as a component manufacturer, Intel’s ability to meet customer demand is of paramount concern. “We already hear ‘I can’t forecast’ and, combined with something like machine downtime, it’s hard to meet demand,” Petrick said. “There’s also little flexibility if you want to change one customer’s product [run] over to another’s. A factory driven by software—which is a large part of the intelligent factory—is much more flexible.”

The key to achieving true intelligent factory status is to integrate every process and system into a holistic “system of systems” that is run by real-time data and IIoT connectivity, Intel found. The chip maker is already using the IIoT to improve yield and increase efficiency in its own facilities:

Big data analytics helps detect the potential failure of critical testing units before they wrongly categorize “good” devices as “bad”
Image analytics speeds up the once hours-long manual task of segregating true rejects from marginal units
Machine learning helps visualize and automate the control of equipment performance and anticipate maintenance needs before failures

Similar benefits – increased flexibility and better quality control – can also be realized throughout the supply chain. “It’s essential in a supply chain that the systems are interconnected and that they can be coordinated in near real-time,” Petrick said. “With more data in hand, companies can be more proactive at any point, including logistics and inventory management. If I am trying to move materials to a particular location or production facility, I want to know where those materials are in the warehouse.”

More than technology

However, Intel’s research deliberately extended beyond manufacturing systems to the people in the factories and the C-suites. Companies are at different rates of “digital intensity” – the degree to which computer-based technologies do the work for humans. Not surprisingly, most people were a long way from the vision of the intelligent factory.

Most workers believe that technology solves problems. The most frequent glitches in a manufacturing setting include:

26 percent – information challenges related to difficulties getting needed information for work tasks.
24 percent – equipment maintenance and upkeep including unplanned downtime, reactive rather than proactive management, difficulties diagnosing problems, and maintenance costs.
19 percent – communication challenges related to the lack of effective coordination across the factory (e.g. between teams, sites, or functions).
18 percent – safety hazards such as air quality, temperature, noise levels, and ergonomic issues (e.g. lifting heavy object) are top of mind.
17 percent – equipment not a good fit for work whether it is the complexity of changeovers, age of equipment, or not using it for intended purpose.

At the same time, workers across the spectrum are leery of technologies such as AI and the reliability of intelligent systems. Moreover, managers are notoriously resistant to change, Petrick said, and employees are worried about their job and the ongoing need for their skill set.

Fifty-six percent of the obstacles raised related to the culture and leadership of the company—whether workers were concerned with the slow pace of change at their factories; leaders concerned with resistance to change from their workers; or workers recounting C-suite-level distrust of new technology.

Although many respondents saw the advantages of a smart, data-driven manufacturing environment, they were also concerned that data would reflect badly on their decision-making. “That’s part of developing a digital culture at an organization,” said Petrick. “You must communicate how [employees’] jobs will change and how the system will help them. Developing trust at all levels will help deal with that.”

Agents of change

Like it or not, change is inevitable. Leading manufacturers are rapidly moving down the path toward AI-driven and IoT-enabled systems. IoT technology spending is forecast to reach $772.5 billion this year according to IDC — an increase of 15 percent over 2017.

“If we want manufacturing to truly benefit from the rise of the intelligent factory, we need to get workers and leaders to trust it,” Petrick said.

Most workers acknowledge that evolution is necessary for their companies to stay competitive. They can play a role in effecting change, Petrick said. “From the factory floor to the C-suite, the majority believe they should and do influence the technology decisions in their company. If employees play a hands-on role in that change they are more likely to buy in to technologies that demonstrably solve their problems.

“Our work here suggests that the vision [of interconnected, smart factories] should consider the co-evolution of manufacturing environments and the people that work within them,” she added.

In addition to improving its own performance, Intel is using smart-factory research in its product development. “If you are going to be selling components you should understand the systems where they reside and the problems those system resolve,” said Petrick. “We felt it was important to understand the operating environment where our components reside.”

“As we envision how things are connected,” she added, “we also see how our components might complement or supplement [smart system connectivity]. We see that we cannot develop a single solution or a single technology. There are many aspects that have to come together.”