The promise of a ‘digital lean’ manufacturing strategy

The promise of a ‘digital lean’ manufacturing strategy

Combining established “lean” manufacturing principles with advanced digital technologies could provide the foundations for a new era of U.S. manufacturing growth, experts say.

One study by consulting firm Bain and Co. found that, while traditional lean practices can deliver a 15% reduction in operating costs, a combined "digital lean" approach can produce twice the savings.

Lean Manufacturing systems are based on continuous improvement statistical measurement, and just-in-time inventory management. Toyota developed them in the post-war years, and they have been widely adopted in the decades since across many industries, including automotive, apparel, consumer products, heavy equipment, and technology.

Digital manufacturing weaves together technologies such as AI, IoT, and fifth-generation wireless (5G). Combined, these technologies amplify the benefits of lean practices by giving managers more accurate, real-time data about their operations.

Lean practices also act as a check on the uncritical embrace of digital tools—costly mistakes codified in industry catchphrases like "pilot purgatory" and "random acts of digital."

“Ensuring we are solving a well-defined problem, one supported with data, is a fundamental part of lean,” says Rebecca Morgan of Fulcrum Consulting, a Cleveland-based firm that works with manufacturers across many industries. “The more we have the right data at the right time and in the right place, the better we can understand and solve the problem,” she adds. “Digitization helps us do that.”

Digital-lean transformation

Lean manufacturing has always been focused on eliminating waste of material, time, and talent. It looks to reduce costs in every part of the manufacturing process, from excess inventory to unnecessary movement of materials, people, or equipment.

Digital lean takes this up a notch. IoT sensors enable managers to track production in real time and adjust capacity to prevent overproduction. They also reduce waiting time by using digital twins to prevent production bottlenecks.

One common target of lean processes is supply chain management. The pandemic revealed how fragile just-in-time processes could be, as disruptions in supply and sudden surges in demand left manufacturers unable to fill orders. The result was empty shelves. Supply chain vulnerabilities remain a problem, as seen in the microchip shortages that hobbled U.S. car manufacturers, and the recent Suez Canal blockage that caused production problems worldwide.

While new digital tools can’t anticipate the effects of extraordinary events, they can help companies prepare for problems before they occur. A recent McKinsey survey cites the case of an Asian consumer packaged-goods company that, before COVID-19, had built a digital twin of its supply chain. When the pandemic arrived, the company ran simulations of plant shutdowns and interruptions in deliveries of raw materials, helping the company to minimize disruptions and better plan for similar scenarios in the future.

Lean machine learning

Other advanced technologies can help manufacturers adapt lean principles when responding to meet new needs. At the start of the pandemic, U.S. hospitals found themselves critically short of tens of thousands of ventilators. Brooklyn-based Nanotronics, which uses machine learning to build robotic industrial microscopes, turned its capabilities to the task of designing and building an inexpensive, over-the-counter CPAP machine, typically used by people with sleep apnea, and winning FDA approval to use it as a respirator for patients who don’t require hospitalization.

Machine learning algorithms devised a way to reduce the number of parts in the equipment and then trained robot-assembly machines to do the work with minimum time and material, bringing down the cost to nearly a quarter of a common CPAP machine.

Other machine learning techniques can also augment traditional lean practices to help plant operations run more efficiently. Factories often have to solve a complex scheduling problem: How to coordinate the production of different machines and workstations so that parts are made and delivered to the right place at the right time, without delays or wasted parts.

San Francisco–based Pathmind, which models industrial applications using AI, used a technique called deep reinforcement learning (DRL) to speed the production of cars for a European automaker.

DRL determines the best sequences of actions by trial and error in simulations that can run through years of experience in a matter of hours. The European automaker used DRL to coordinate the movement of autonomous guided vehicles for carrying sensor-equipped materials and components from one workstation to the next. Guided by 5G, the vehicles can be maneuvered to avoid collisions and prioritize urgent payloads, making the process more efficient—a classic lean result.

“It’s like choreography, but with AI and Roombas,” says Chris Nicholson, Pathmind’s CEO, referring to the popular robot vacuum cleaners.

Traditional lean practices give manufacturers the tools to eliminate waste and steadily improve their production processes, but the productivity gains can level out over time. As the Pathmind example illustrates, digital tools like machine learning and wireless sensors can build on those practices and magnify their impact. Indeed, it’s becoming difficult to get the advantages of lean without them.

As Nicholson puts it, “Lean manufacturing requires digital transformation.”

Fuente: The promise of a ‘digital lean’ manufacturing strategy