Customer Story

On the road to Industry 4.0, one success leads to another

Questions that have long remained unanswered are no longer mysteries thanks to Saint-Gobain NorPro’s commitment to analytics

The promise of Industry 4.0 is very much top of mind at Saint-Gobain NorPro. The digitization of manufacturing, robotics, the Internet of Things, the Internet of Systems – a fusion of the virtual and real worlds – these are concepts embraced in the company’s Stellar Factory initiative, through which Saint-Gobain NorPro is identifying and implementing best practices.

For well more than a century, Saint-Gobain NorPro has been servicing the petrochemical, chemical, refining, environmental and gas processing industries, providing an array of engineered ceramic media and shapes. As quality manager, it’s Jim Lamar’s job to ensure that in an increasingly complex manufacturing environment, the company is fully applying the power of analytics to deliver the best possible product to its customers.

Lamar, based in Bryan, TX, is helping create a culture within Saint-Gobain NorPro that builds on past successes by approaching data analysis in an ever-more structured manner. Designed experiments are at the core of that initiative, and they’re paying dividends.

Case in point: Lamar and his team were called upon to address an issue with a primary supplier’s raw material. Designed experiments were conducted both within Saint-Gobain NorPro and at the supplier’s facility. Two major findings emerged: first, that the quality of the raw material was the only useful factor in predicting the quality of the final product, and second, that there were two raw-material quality parameters in the supplier’s product that contributed to the problem. The interaction of these two parameters was the key to resolving the issue. “Through this process,” Lamar says, “we answered questions that in the past nobody knew how to even approach. Folks are excited about these results: OK, let’s do more. Let's build on this to achieve even bigger successes.”

A predictive model solves a persistent production issue

“For the past 20 years, I've been a full-time JMP user,” Lamar attests. Recently, Lamar and his team used the statistical analysis software to create a predictive model to address an issue that had vexed the company for decades: a quality characteristic of one of its products. “Nothing we checked in 20 years seemed to apply,” Lamar says. “We'd tweaked every knob on the process, and nothing worked. We tried every raw material variation, and nothing worked.”

Experiments set up in the Saint-Gobain NorPro plant indicated that the problem originated with a particular raw material. “We contacted the supplier,” Lamar recalls, “and we said, ‘It seems to be related to your raw material, but we can't find anything that actually correlates.’ It wasn’t something that we could measure."

Lamar’s team then ran more experiments using the raw material in question as the main variable, “and we still found that none of the operating conditions in our process had any effect whatsoever on final-product quality.” The team persisted, and in time, the raw material was confirmed as the root cause of the problem. The source was two previously unknown raw-material quality parameters. “I said to our supplier, ‘We need you to make sure these two things line up. They're interrelated. The value of this one means nothing to us; the value of the other one means nothing to us. But the two together mean everything to us.’”

One of the parameters was a feature of the raw material, and the other could be controlled through processing. Lamar explains that the traditional raw-material specification approach is to define specific limits for each parameter independently. In this case, the acceptable limits for the second parameter depended on the value of the first. “That interrelationship was the thing that nobody had understood in the past,” Lamar says. “Nobody had the data in the past. The profiler at the bottom of the multivariate multi-regression system gave me the answers that I needed.”

Scripts that access years of data

Lamar and his team used JMP to generate an interactive profile that allowed the supplier to input its raw-material quality measurements and then control the other factor to the level needed to ensure that the final product would render an acceptable raw material. “We saved the formula as an HTM file,” Lamar says. “JMP gave me the ability to hand them what they needed. We took it over on a thumb drive and showed them how it operates.”

“We not only solved the problem,” he continues, “we solved the problem to the point where our customer came to us and said, ‘You guys have done so well on this, we'd like to tighten the specs. The old specs don’t reflect what you can do today.’”

Lamar avows that were it not for JMP, no other analytical software would have been up to the task. “I don't have any other tools that would allow me to do what I did with the interactive profiler, and none of the other tools offer the capability of storing that as a Shockwave file that I could just hand to the supplier and say, ‘Here, use this. Your specification is this index off of this profiler.’ I don't know how we would have solved it without that.” He further attests to the scripting capabilities of JMP. “My scripts pull data directly out of our enterprise resource planning software in seconds. Things that used to take us hours and hours and hours to do, I can pull directly from, and am provided with years of data that I can analyze.”

Lamar deploys JMP to regularly, and rapidly, analyze key quality parameters and to conduct comparative analyses: “What does 2018 look like versus 2017? Do I see any trends? Can we show evidence of continuous improvement?” These are questions that Lamar and his team can now easily address.

“The scripts are already built to pull the data, and other scripts are built to do the analysis of the data. JMP can answer your question in a matter of seconds. A customer comes to visit with a question, and I no longer have to say, ‘I'll get back to you with an answer.’ I'll say, ‘Let's look.’ And I can put it on the screen,” sharing data files, journals and more. Moreover, those customers are themselves using JMP. “If my customers use JMP – and many of them do – then it's certainly to my benefit to use it as well.”

For cutting to the chase: ‘JMP^® is huge’

Management, too, is looking for quick answers, and Lamar notes that decision makers don't want “a pile of statistical indices. They want to be able to look at a graph or a chart of some kind and see that the answer is X. Don't bother them with 10 pages; show them the graph at the end.” JMP affords exactly that. “JMP is huge.”

NorPro’s team is comprised of research engineers, industrial engineers, chemists, lab technicians and others, many with advanced degrees, “but nobody with the word ‘statistics’ in their title.” That may well change. But in the meantime, with the assistance of JMP – its easily accessible efficiency, probing analytical capabilities and sophisticated visual tools – Saint-Gobain NorPro is nonetheless nurturing an environment that advances, success by success, into the promise of Industry 4.0.