Inside an electronics factory in China as late as early 2019, workers and smart machines were building circuit boards with chips in a multi-step process using a variety of machines from different vendors. None of the machines on the line communicated with the others to enable data analysis that could have greatly improved productivity and profitability.
Arch Systems CEO Andrew Scheuermann, who began working with the Chinese factory in the second half of 2019, described a line with a lot of discrete stations: a printer printing solder paste; another machine putting the paste on the circuit board; a robot arm placing the chips.
Each machine in the factory is from a different vendor. Each machine was doing its job very well. All of it was “Industry 3.0,” he said. “In some factories where you have built a system around a $3 million metal stamping machine, you may still be Industry 2.0.”
Arch Systems, which provides machine-data-sensing software compatible with new and legacy machines, has partnered with Flex, a global supply chain and manufacturing company, to create a platform called ArchFX. ArchFX integrates these individual machines, harvests data, and runs data analytics, Scheuermann said.
“Once you get all that data together for the first time, you get a lot of insight,” he said. “One of the first things we do is look for bottlenecks in throughput, look at key performance indicators, and machine uptime. For example, when you have all the data in a central place you can see that machine No. 4 is the one always having problems, and that is your bottleneck. You’re not seeing that problem a month later after someone looked at all the pen-and-paper records. You’re seeing it immediately.”
The circuit-board maker discovered that, for example, when a part reached machine No. 10 for inspection, it was not up to specs and had to be thrown out. Data analysis of the entire system revealed that the problem is caused by machine No. 2 not applying enough solder, Scheuermann said.
“Machine No. 2 is not putting enough solder down but the process is not being measured until machine No. 10,” he said. “The company spent all the money to do steps 3 through 10, and then had to throw the part away. By analyzing the full system data, we can now detect a problem at machine 10 and solve the root cause at machine 2 for all future parts.”
Despite advances in automation and data analytics, despite the clear advantages of the digital factory, 49 percent of U.S. manufacturers with a global factory footprint have not crossed the digital divide, according to a smart factory survey that Deloitte and the Manufacturers Alliance for Productivity and Innovation (MAPI) conducted last year.
Meantime, compound annual labor productivity growth—a robust 3.6 percent from 1987 to 2006—dropped to 0.7 percent from 2007 to 2018, according to the Deloitte/MAPI survey and U.S. Department of Labor statistics.
Total manufacturing sector productivity declined 0.7 percent last year, as output decreased 1.2 percent and hours worked decreased 0.5 percent, according to the Bureau of Labor Statistics.
Digital factory initiatives, such as ArchFX, could be the answer to that slowing labor productivity.
That’s because one key exception to stalled labor growth was the 51 percent of manufacturers transforming themselves to digital factories, according to the 2019 Deloitte and MAPI smart factory survey of 500 manufacturers.
These companies, which the Deloitte/MAPI survey labeled Group B, achieved compound annual productivity gains of 3.3 percent from 2015 to 2018.
In fact, the higher productivity in manufacturers adopting smart factory initiatives is what pulled overall labor productivity into positive numbers, the survey said. Productivity for Group A, the 49 percent that that did not adopt smart factory initiatives, likely declined a compound 2.3 percent per year, the Deloitte/MAPI survey estimated.
As manufacturers in Group A begin to adopt digital factory initiatives, these manufacturers are likely to see productivity gains of 4 percent per year for several years, the survey said. These gains will taper off to about 1.38 percent after five years—the same gains that Group B, the early adopters, is experiencing now, the survey predicted.
“Eighty-six percent of our survey respondents believe smart factory initiatives will be the main driver for factory competitiveness in the next five years,” said John Ferraioli, consulting managing director in Deloitte’s supply chain and network operations offering and the data leader in Deloitte’s digital supply networks practice.
“These investments in smart factories lead to greater visibility and greater insight about performance,” he added.
In the Deloitte/MAPI survey, digital/smart factory initiatives led to a 10 percent average increase in production output, an 11 percent average increase in factory capacity utilization, and an average 12 percent increase in labor productivity over the previous three years. Manufacturers reported operational and financial benefits after adopting these use cases: quality sensing and detecting, factory asset intelligence, plant consumption and energy management, command center, synchronization and real-time asset tracking, and smart conveyance.
ArchFX’s manufacturing clients typically see a 3-5 percent increase in efficiency as a result of better decision making, Scheuermann said.
For example, “A large consumer plastics and packaging manufacturer in Mexico last year was at risk of losing out on tens of millions of dollars in incremental revenue by not addressing core issues and modernizing its operations,” he said. “Through our work with them, the company has seen a 9 percent overall equipment effectiveness (OEE) improvement.”
More manufacturers are realizing the potential, said Jean-Philippe Provencher, VP for manufacturing strategy and solutions at PTC: “At PTC, we believe the enterprise is catching up in terms of using the data. Implementation of digital threads is accelerating, and customers are more and more looking at enterprise deployments for their IIoT initiatives. Customers are more and more aware of the hidden value of their data and are looking to leverage it at scale.”
Although some benefits require crunching months of data, manufacturers usually can get actionable insight within weeks and realize return on their investment within three to six months, Scheuermann said. Longer-term benefits include efficient asset deployment within and among factories, and the ability to act faster on data that is more up to date, he said.
For example, one electronics manufacturer in the U.S. last year depended on weekly data from its coordinate measuring machines to make decisions about moving those machines around among manufacturing plants and whether to buy more machines.
Before using ArchFX, the company needed a month to collect all the data and another month to generate the report. The result was the July report was based on data from May.
Now with ArchFX, the data is gathered and analyzed quickly and automatically. “They’re looking at data from an hour ago or less,” Scheuermann said. “They’re looking at data from each machine aggregated at the machine level, the plant line level, the plant level.”
In the past, the high cost of retrofitting or replacing an expensive machine to make the machine compatible with Industrial IoT technology prevented some manufacturers from moving to a digital factory, he said. Companies were reluctant to invest $1 million and wait three years for a return on investment.
ArchFX’s compatibility with legacy machines and quicker ROI helps overcome that barrier. For the smart machines, ArchFX accesses the data from existing sensors, Scheuermann said. For legacy machines, the platform adds sensors and/or an Industrial Internet of Things device to collect data, often for the first time.
“Historically, there has been so much money involved if you had a lot of legacy machines,” he said. “We’re trying to break down the door. We’re dramatically dropping the cost to retrofit these older machines.”
Using such a system, a manufacturer may realize that a factory now running 1,000 machines could accomplish the same work with 850 machines, he said. Or that manufacturer might see the situation as unused capacity and take on more orders, he said.
Manufacturers also are able to increase output from the same factory line, he said. For example, one factory line was regularly producing 30 units per hour but had achieved 32 to 36 units per hour in the past.
With the ArchFX system, “You see the bottlenecks in real time,” he said. “Right now, your bottleneck is between steps four and five. You see you need to add another person to step four. Suddenly things start moving. Or, you see that at step seven, someone—the person or the machine—is constantly making a mistake. You see that you either need to retrain the person or recalibrate the machine.”
Based on that data analysis, the company was able to boost hourly throughput from 30 to 34, a 13 percent increase, he said.
More out-of-the box systems may be part of the answer for increased deployment of inclusive digital factory programs to smaller companies and to those reluctant to cross the digital divide.
Increasingly, more generic blueprints are becoming available that can be tweaked to meet the needs of smaller manufacturers that haven’t yet ditigized.
For example, GE now has a catalog of over 300 digital twin blueprints for equipment, including pumps, aircraft engines, transformers and turbines, GE Digital’s Dan Lohmeyer said.
Larger manufacturers make up many of the early adopters of ArchFX because they have more machines and can better absorb the cost, Scheuermann said. But similar to GE’s catalog of blueprints, ArchFX has a growing library of systems available.
“Someone with one factory line wouldn’t have been one of our first customers,” he said. “As we work with large enterprises, we’re building a ‘Connector Library.’ A customer today wouldn’t necessarily have to pay the up-front costs to get the data infrastructure.”
ArchFX also encourages smaller manufacturers to share data in a multi-tenant cloud so that, for example, a factory with only three printers using ArchFX could gain insight from a pool of users, Scheuermann explained.
“We have multiple steps and multiple ways to work with machines out of the box,” he added. “Plug and play: that’s a key enabler of what we do.”
Ten years after beginning a digital factory journey and evolving from 25 percent to 75 percent automated, productivity at Siemens’ manufacturing plant in Amberg, Germany, is up 1,400 percent.
“Yes, it’s a crazy-big number and it’s a sum of a number of things,” said Alastair Orchard, VP of digital enterprise for Siemens Digital Industries Software. “It is now a very high-speed factory. We now make a million items per month in that factory, one every second.”
But productivity is “absolutely not the only improvement,” he said. Other KPIs, including the number of products made per month, the error rate, the variety of products and time needed to introduce new products, also are significantly improved. Overall, Siemens focused on productivity, quality, flexibility and efficiency and on making these changes without sacrificing security.
Siemens began the digital factory process in 2010 as part of the German government’s push to move to Industry 4.0, Orchard said.
Siemens did not cut the factory’s workforce or square footage. As workers retired, the factory replaced them with digital natives.
“As a country, we felt increasingly under price pressure from China,” he said. “If we continue making very high quality, very expensive products slowly, our manufacturing is no longer going to be competitive. We’re going to be eaten by these rising economies. To protect the overall economy of the German state, we have to massively increase productivity of all our manufacturing plants and protect the jobs of people who work in those companies. The idea is not to become more productive with fewer people but to do much, much more with the same number of people.”
“That was the goal of Industry 4.0 in Germany,” he added. “Each of those workers is trained on digital technologies. They’re radically more productive than they were before. Their jobs are safer. The company, and therefore the whole Germany economy, is more competitive on the world stage.”
In Germany, manufacturing comprised 20.4 percent of GDP in 2018, according to Trading Economics. In the U.S., that number in 2018 was 11.6 percent, according to the Bureau of Economic Analysis.
Before the digital transformation, the factory “had an extremely inflexible operational model,” he said. “Delivery times were often ... a few weeks. We had stock problems, supply chain issues. We used to operate this factory like most other factories. For example, in a car factory, you put a chassis in at one end of the factory and then that chassis is going to go on a fixed journey through the factory and meet people and machines in a fixed order and pop out the other end a car. If you want to build a hybrid car, you have to build a new factory.”
Now, the plant is reconfigured monthly with 60 percent of machines moved based on the demand profile of what the plant will be making, Orchard said.
Before, the factory could make five different products. Now, in any 12-month period, it makes 1,300 different products and has the capability to make 9,000, he said.
For Siemens, this transformed digital manufacturing plant not only shows what the company can do for itself, but also what it can do for its customers, Orchard said.
“We wanted to create far more choices for our customers,” he said. “Instead of building other single-purpose factories, we wanted to unlock the flexibility within this factory. We are moving toward the holy grail of Batch Size One.”
Before using High QA’s quality-management software, ballooning a drawing—adding numbers to the part print to correlate with sometimes hundreds of numbers on the dimensional data sheet—could take three to five days because it involved eight steps and different inspection software for each of them, Tyler Schildt, lead quality engineer at Hypro, said in a webinar.
That slowness leaves manufacturers choosing between accuracy and speed.
“With our previous technology, our staff spent too much time on manual processes. Everything had to be inputted manually,” he said about the process at Hypro, which has seven facilities in the Midwest. “It allowed for many errors. The usability made it harder to train new users. Editing was very difficult. We needed quality planning to lead the quality control process.”
Instead of ballooning the drawing and then making and inspecting the part, the company first made the part. “Ballooning took so much time that the coordinate measuring machine team had already completed the inspection prior to having the balloon plan shared with them,” Schildt said.
Using High QA’s Inspection Manager, Hypro’s ballooning process now involves only two steps. “We input the project, then immediately auto-balloon the drawing,” he said. “Within minutes, we have a fully ballooned drawing that’s ready for review and other automated steps, such as in-process inspection, inspection plans, tool assignments and more.”
The speed and trusted data are game changers. Because of the need to deliver parts on time, “more than 50 percent of the market is doing the process upside down:” measuring the part and then ballooning the drawing and matching results, High QA CEO Sam Golan said in an interview with Smart Manufacturing.
“The guy who measures the part cannot wait,” he said. “The owner of the shop needs to deliver the part or he will miss his delivery window.”
With Inspection Manager, Golan said his firm helps customers achieve good price, higher quality and on-time delivery.
Inspection Manager starts with auto-ballooning and extracting geometric dimensioning and tolerancing (GD&T) information from the drawing. It then builds an inspection plan based on that data, Golan said, adding that “manufacturers become almost completely paperless.”
Another game changer is eliminating interpretation, he said. “When you balloon a drawing manually and extract quality requirements manually, this is where interpretation starts,” Golan said. “The wrong interpretation will impact part quality, which is associated with higher cost and missed delivery times.”
Imagine a simple not-yet-ballooned drawing with four holes. The Coordinate Measuring Machine checks the holes and gives results for each numbered hole. “On the original drawing, you don’t know which hole is one, two, three or four,” Golan said. Someone has to figure out which result is which for each hole—and usually that’s not four but hundreds or even thousands of holes, Golan said.
Many customers estimate up to 15 percent errors in those situations, he added.
Connect With Us
