Walter A. Shewart, often referred to as the “father” of statistical process control, introduced the world to SPC and his creation of the control chart in the late 1920s. Of course, SPC has appreciably evolved since then, with real-time data analytics and Cloud-based software becoming an integral part of the package.
If Shewhart were alive today, he might be surprised by how little humans, as opposed to computers, read and interpret control charts these days. Alternatively, he might be impressed by how modern engineers have chosen to build upon his legacy. By moving from handwritten control charts and locally-hosted spreadsheets to data-driven and centrally-hosted SPC software, companies tend to enjoy increased visibility, efficiency, and quality, not to mention cost savings.
Steve Wise, vice president of product design at InfinityQS, is familiar with this transition. After receiving his degree in industrial statistics in the mid-1980s, Wise began teaching SPC to operators at Boeing by having them draw control charts with long sheets of paper and colored pencils.
On a phone call with Quality, Wise discussed the ways in which SPC has changed and stayed the same over the decades.
Same backbone, different approach
Wise’s first job out of college was teaching SPC on the factory floor. “I had a training manual of how to use a calculator” he says. Excel didn’t exist yet, and operators were expected to draw the control charts themselves.
“The mission was to teach operators how to do R-bar over D. 2 and how to get sigma and the mean of each of their plot points, and how to plot the points properly, to get them at the right spots on the graph,” he continues. “And after all the math, plotting, and colorizing was done, eventually they got around to looking at what the data looked like. No one was really analyzing the data so much, because it was just so much work to get the pictures of the data, and the goal was to get the pictures.”
Though Shewart’s control chart remains the spine of modern SPC, it’s now tucked into the background of smart software, like InfinityQS’ Enact. The purpose of the control chart is the same today as when Shewhart created it: to create a signal or not create a signal, which tells the operator to either do something about a new piece of data or do nothing. The big difference, Wise notes, is that computers are reading the charts and sending the signals to the appropriate people, so that the users themselves don’t have to create, read, or even understand a control chart.
“Being proficient in the mathematics behind these charts, what’s creating signals and what’s not creating signals—there’s no need at all for in the general user today,” Wise explains. “They don’t have to have any knowledge whatsoever.” As a consequence, an SPC software provider like InfinityQS must work harder to assure customers who are making expensive decisions with their data that they have the right people behind the scenes to ensure that the information they’re receiving is correct.
“We hope that we’ve built that image over time: That customers can rely on us to take care of the math part and the integrity of the data,” Wise says. “If there is anything wrong with the data, as far as how it gets analyzed and the triggers it creates for operators to do something or do nothing with the data, everything stops and that takes priority,”
The data boom
Wise adds that in the late 1980s and 1990s, SPC providers were trying to educate the markets on the benefits of using real-time data to make decisions; but it wasn’t until the marketspace began turning from mostly analog to digital in the new millennium that the message stuck and took on a new meaning.
“Between going from Excel to databases to relational databases to Cloud-based systems, data just became more ubiquitous with how we make our decisions,” Wise explains. “We didn’t have to really teach the values of SPC as much as ‘How can we make use of this data that is just flowing all around us?’”
People in the business of supporting factory floors with equipment and tools began turning “smart,” Wise adds, as they incorporated smart gages and smart instruments that could collect the data for them and stick it somewhere else.
“This data was everywhere,” Wise continues. “So our challenge became: ‘We have all these piles of data all over the place; how do we bring it all together?’ We as well as our competitors, and anyone who was trying to provide tools to analyze data, had to figure out a way to get it all in one spot, so it became meaningful.”
As the definition of SPC has expanded to include automatic data analysis and reporting, Wise says that customers’ expectations also have changed.
“When you call it SPC, you’re really calling it data management or big data or analytics, because what you’re really doing is collecting data and analyzing it and making decisions on it,” Wise explains. “And as we’re competing in this marketspace, it’s just a matter of who can do a better job with providing useful information to our customers.
So, that has changed quite a bit,” he continues. “We don’t have to sell the message of, ‘If you just train your operators how to read these control charts, things will get a lot better. I know it takes time away from them doing their job, but it’s really important.’ We don’t have to make those statements anymore.”
The digital revolution
When Wise worked at Boeing in the ‘80s, the company relied on physical files and file cabinets to store their data. “And once that data became a year old, someone would come in and empty these files into cardboard boxes and tote them off into another storage facility,” Wise recalls, “because if an airplane fell out of the sky, someone would need retrieve all of that data.”
Now, all of a company’s data can be stored in a single Cloud repository; and the ability to crunch the numbers, provide predictive analysis, and give immediate feedback about what the data is saying also has become infinitely easier, Wise notes.
“And with customized reporting and dashboards, once you log in to the system, it knows who you are and, on any given day, what operations you’re responsible for, and what’s being fed into those processes,” Wise explains. “This is not SPC per se; but because these are all types of data that are flowing on the factory floor, those who want to make the SPC piece of data flows more meaningful need to integrate SPC in with the rest of this stuff. What’s being created at any given operation, what’s being measured—all of that can be analyzed using SPC.”
Wise estimates that about six or seven years ago, the technology of cloud computing finally caught up with ideas that had been circulating within InfinityQS for at least a decade.
“When it really began to change, at least for InfinityQS, is when the Cloud became a reality,” Wise says. “Once these Cloud-based systems came out—where the system is not relying on the company’s internal IT group to create an Intranet—that created the ability for anyone to log on to a device and plug into these streams of data from anywhere in the world.”
Previously, a company’s internal IT group had to create their own internal database at each site, and each had to power their database with their own computers, hard drives, and other hardware. Replacing IT teams and hardware with Cloud-based software was a no-brainer from a cost savings standpoint, Wise says, and the vision was already there. Consequently, for InfinityQS, it was just a matter of building the tools to take advantage of the new technology.
Four years ago, Wise says, InfinityQS began writing code for Enact. Today,the Cloud-native platform serves aerospace, automotive, medical, packaging, general manufacturing, and other industries across the globe.
“Easy is hard”
In the digital age, and especially in digital manufacturing, the user-friendliness of one’s product can be enough to make or break a sale. Operators don’t have time to waste on factory floors; so if a smart SPC software can collect, monitor, and report on data for them, then usually the investment is worth it.
When building the Enact product, Wise says the team was faced with a choice: “Do we create a wall of charts for people to look at more efficiently?” After digging deeper, they decided that they didn’t need charts to be front-and-center anymore. The operators didn’t need to look at them.
“You can look at them with a few clicks, if you really want to,” Wise adds. “But the front and foremost screens of our system don’t have control charts on them, because the control charts are doing their job in the background. Even the ability to aggregate and analyze and crunch the numbers—that’s not necessary for the users of these systems anymore, because the computers are doing all of that work for them.”
Instead, smart SPC software frees up operators and quality managers to do their jobs more efficiently. “Someone still needs to respond if the system sends a signal that says, ‘You need to go over here and check this particular line pressure on this machine running that part,’” Wise explains. “Someone still needs to go out there and do that work, and they still need to make sure that the proper sampling strategies are in effect.”
Cobbling together the right sampling strategy so that it’s not too expensive, but still provides enough data to give the users what they need, is part of the training that InfinityQS providers for the end-user.
“When the signals are flying, it’s still up to the users of the data to make sure that the data doesn’t become stale, that they’re still responding in the right amount of time,” Wise asserts. “Otherwise, if you just ignore the signals, SPC is nothing but producing control charts, and that’s all it stands for.”
In Enact, data models are stitched together to go from raw materials to finished goods; and as signals go out from the SPC tool, the Enact system automatically assigns whatever problem is occurring to the right place—even if that place is at receiving inspection, halfway around the world.
“We’ve ingrained that into the process models, the work flows, of how data come together to produce these finished products,” Wise explains. “Experts build the models, and those models are stitched into the computer.” Therefore, when the signals go out, they automatically go to the right person and the right place—even if that person is way upstream in the process or in another country.
“We realize that no one person has all the knowledge of how ingredients get transformed into finished goods,” Wise says. “But now, with this process model logic in Enact, you just need to have that knowledge as you cobble these models together; and after that, the system does the work for you.”
Wise relates this to how, when he was a college student in 1982, Apple introduced their first personal computers on college campuses, including his own.
“They came out with these Macintoshes at the student centers and said, ‘Hey, come look at this computer, click the mouse, we’ll give you a free T-shirt,’” Wise remembers. “And so I got an Apple T-shirt that had the colorful Apple logo on the front, and on the back the slogan was, ‘Easy is hard.’”The phrase is equally apropos to modern data management, Wise notes.
“It is really hard on the back-end to make it comes across as easy on the front end," he says. But that is exactly what he does.