Gelston Howell, senior vice president of marketing with contract manufacturer Sanmina Corp., says factories and warehouses are already being transformed by the latest developments in automation – specifically, robotics, guided vehicles and machine learning.
Call it the age of the “digital factory,” if you’re comfortable with a term that can be maddeningly vague. Howell sees it as the application of technology to streamline operations, control costs, boost traceability and create operations that can respond to rapid changes in consumer demand and component availability.
The capabilities come in the nick of the time. The booming economy, while great for manufacturers and merchandisers, has also led to chronic electronics component shortages that threaten to curb profits. And the very nature of contract manufacturing presents a series of complexities that get tougher to overcome as demand for product increases.
Sanmina, one of the world’s largest contract manufacturers of electronics, launches or transfers an average of 10 to 15 products per day, in highly regulated industries. In the medical sector, it must grapple with documentation related to the qualification of installations (IQ), operations (OQ) and performance (PQ). Recently the company produced a 4,000-page PQ document for a single medical product.
For the automotive industry, Sanmina is being called on to craft ever more sophisticated components, such as complex electromechanical subassemblies, in high volumes. It relies on electronic “travelers” that might employ several dozen robotic cells and some 150 sensors to perform optical inspections in real time. The components themselves are equally complex, with a modern-day printed circuit board assembly containing up to 10,000 components.
While some factories are becoming fully mechanized “lights-out” facilities, most still require the presence of humans, who interact closely with automated systems. It’s the human’s job to remove faulty parts from the line for rework or disposal. Sanmina can verify each operator on the line, and ensure that those individuals are trained on the latest version of authorized operating procedures.
Often that’s where glitches occur. According to Howell, among the top causes of recalls are incorrect parts or operator error.
Of course, all parts are verified before they go into production, and are duly entered into the database to ensure future traceability. That’s why the occurrence of a faulty part is extremely rare in operations that employ the latest technology for scanning and verification. Howell says the rate of failure comes “very close” to the holy grail of Six Sigma, which allows for no more than 3.4 defects per million opportunities.
For all the buzz about machine learning, it's still in its infancy on the production line. But manufacturers can come close to it with the use of high-speed, real-time data analytics. The tool came in handy when one customer requested large volumes of a Class III medical device. Out of the box, the product contains a sensor that ensures accuracy within a range of plus or minus 10 percent. Sanmina was able to squeeze that down to plus or minus 5 percent, through the use of calibration procedures supported by analytics.
Expect the role of humans to continue to diminish as automation takes on a greater share of tasks on the factory floor. “Lights-out” facilities are a compelling option for complex, highly regulated items such as certain medical devices, which require full traceability throughout the supply chain.
Full automation is less attractive when output levels can't justify the investment. Examples include CAT scan equipment, MRI machines and ATM units.
“The volume is not there and the product is too complicated,” says Howell. “If you have thirty to four hours of labor going into a highly complex system, it doesn’t make sense to automate it.”
The mix of people and machines in the factory is constantly changing, he says. But the long-term trend is toward fewer people on the production line. Over the past five years, Sanmina’s revenue has risen steadily, while its headcount has remained fairly flat.
With unemployment dipping below 4 percent, one might assume that it’s getting harder to find the necessary human talent to staff factories and other nodes of the supply chain. Training looks to be a particular issue.
Howell says that’s not necessarily the case. Over the past five to eight years, the technology has become easier to deploy. It removes the need to hire large forces of human programmers and other highly trained engineers or analytics experts.
For the future, Howell sees further innovations in automation taking place within supply chains. Already Sanmina has in place a cloud-based manufacturing execution system (MES). But the company is careful not to over-invest in technology before it’s needed. Decisions are made in line with what executives really want – say, to receive push alerts whenever yields are running below set limits.
Today, Sanmina offers 42Q, an MES that it calls “a factory in a box.” The tool offers visibility based on key process parameters, accessible from any location. But the technology still must be deployed in a considered manner, Howell says, and not just because a CEO is dazzled by automation.
“Five to seven years ago, technology was the barrier,” he says. “It’s not the barrier anymore. It’s about deciding what you want to do, and whether it provides value to management.”