Challenge

Digital manufacturing systems at Murata’s Vantaa, Finland site generated far more data than could be managed with existing tools. Barriers to data access and time-consuming data integration processes caused delays and limited the value that could ultimately be extracted from the company’s digital factory. 

 

Solution

Murata Finland made JMP® statistical discovery software its universal data integration tool. Now used by operators, technical experts, data scientists and management alike, JMP helped automate the site’s entire data analytics workflow and improve data quality.

 

Results

Having an end-to-end workflow tool like JMP has released domain experts from tedious data preparation tasks to instead focus on solving engineering problems that create real business value. Key outcomes include barrier-free real-time data access; the standardization of statistical best practices; the shift from firefighting to proactive improvement; and significant improvements in data quality. Says Data Integration Manager Philip O’Leary: “JMP is an enabler. You can see how they've actually engineered the software to meet your needs – it’s second to none.” 

As we enter a new era in connectivity, innovations in sensor technologies have shifted the horizons for human health and mobility. Leading electronic component manufacturer Murata has held a place at the vanguard for more than seven decades, and today, the company’s ground-breaking sensor technologies can be found in devices as diverse as pacemakers and autonomous vehicles. Advancements in MEMS sensors are largely responsible for making these devices both safer and more convenient.

At Murata’s facility in Vantaa, Finland, specialists, scientists and engineers pursue the design, development and manufacture of 3D MEMS-based accelerometers, inclinometers and gyro sensors for wide-ranging safety-critical devices. There, the company has developed a state-of-the-art, integrated manufacturing environment in which to produce these technologies in a way that fulfills the uniquely high reliability and quality assurance standards that come with safety applications.

Data Integration Manager Philip O’Leary leads a team tasked with discovering and revealing insights from the mountain of digital manufacturing, test and equipment data generated every day in Vantaa where the availability of integrated data supports AI and machine learning activities. The team’s remit is to not only continuously monitor process, equipment and product status, but also to manage the analytics workflow and support the deployment of hands-on data tools to domain experts across the site.

“Murata’s analytics culture has changed dramatically through the years,” O’Leary says. The digitization of what was previously consigned to handwritten paper run cards was a huge innovation, he concedes, albeit not a solution in of itself to the company’s key data challenges. Data management and access, he explains, continued to be of paramount concern even after the transition to digital architecture.

“Even with [a digital factory], we still hadn’t solved the problem of finding the right information when we needed it in real time,” he says. “Though we might have been able to see associated data collection operations and measurements, once a process was finished, it was consumed by the database. There was so much value in this data that was not being utilized … We had all the data; we just didn't have access to it.”

An evolving analytics landscape requires new and better tools

In the early days of digitization, it was taking O’Leary and his team up to a week to build a data set from a variety of sources using Excel. The extended lag time meant that any data set was already a week old by the time it was ready for analysis. The same need that had once motivated the transition from paper to Excel now motivated the transition from Excel to Minitab.

“Upgrading to Minitab gave us more insight and reduced the amount of time from a week to a number of days,” O’Leary says. “But Minitab didn’t have great data acquisition functionality and we were still always running a couple of days late.” Contending with fast-growing data sets and increasingly sophisticated manufacturing equipment, O’Leary and his team again sought to upgrade.

That, he says, was when the need for more robust, high-capacity data applications and customization led Murata Finland to JMP® statistical discovery software.

“Once we got access to JMP, we found that we could now automate how data sets are made. And make them in such a way that every morning when we come in, there will be a fresh new data set giving us the history of the previous three months’ worth of production,” O’Leary explains.

“With the introduction of JMP – and the way in which we can now gather data in a more straightforward manner – we’re getting information so much faster. We literally went from collecting data over weekends to collecting data within an hour or so, and that has since been further refined down to a few minutes. Moreover, we do it hundreds of times every day.”

“JMP is by far the best way to collect data – or use data that was previously collected – and automate a lot of the analysis people have to do on a day-to-day and week-to-week basis,” he adds. “Most of our weekly reports are now automated, and by centralizing the distribution of data, we’ve made them available to everybody. JMP enables people to access databases and datasets that they otherwise wouldn't have had access to.”

This level of access is critical, O’Leary says, in ensuring that domain experts use statistics as a tool to augment the impact of their expertise. For example, engineering change notifications are now based on a combination of expertise and analysis where previously, decisions would have been made based on expertise and intuition.

Having an end-to-end workflow tool like JMP has released domain experts from tedious data preparation tasks to instead focus on solving problems and introducing proactive improvements that generate real business value. Statistical experts on O’Leary’s team provide custom support in the form of JMP scripts that can be deployed sitewide to automate repeat analyses. “This saves [engineers] a tremendous amount of their time,” he says.

Universal access and standardization ensure widespread adoption of best practices for data analysis 

Another major efficiency is universal, continuous data access. Murata runs JMP on its server so that the system is continuously backed up such that even in the event of a power outage, the site would still continue to collect and share data. A single script, O’Leary adds, calls up a dashboard whereby anyone in the organization can click into any data set. This system – which the team calls a Standard Data Indexer – completely dispenses with data siloes that could otherwise seriously limit the business value Murata might be able to extract from its data. 

Standardization goes hand-in-hand with this expanded access. “The data structure is such that regardless of which product dataset you open, it's going to be identical from product to product,” he explains. “Some persons may be specialists on one product but when a colleague is out on holidays, any specialist can help monitor another product. Everyone has access to everybody else’s datasets.”

Though increased speed and expanded access are significant benefits to Murata’s new systems, O’Leary says perhaps even more important is the way JMP has enabled the company to standardize best practices for data collection and analysis. Standardization has contributed to better reproducibility – and therefore quality – in addition to making data-driven approaches easier to scale.

“We now have very stringent ways in which analysis is done,” O’Leary explains. “We have a reference data set that we can use as a comparison before actually implementing changes. And this is just one example.”

Automation in JMP flags and repairs data sets for unprecedented data quality 

The tool has also provided a much-needed solution to inevitable data quality issues. Not only does JMP automatically collect, process and integrate data from multiple sources, it also uses scripts to flag data issues like duplicates or missing data. Users can then review the flags and take the appropriate steps to recover errant data points.

“With JMP, we can automatically assess our data quality and make improvements, including reconstructing and repairing data sets,” O’Leary says. “We've gone from manually writing scripts and running them one at a time to being in the situation that we're in now, where scripts run automatically, and data sets are self-repairing. You don't see problems anymore. I would argue that our data quality is the best that it's ever been.” And recent advancements in artificial intelligence and machine learning – a top-priority area of investigation for Murata – will only reduce data quality issues further.

The ability to automate data collection, management and integration in JMP was the tool’s key selling point for Murata, O’Leary recalls. “It was the automated manner in which you could provide data for analysis that most attracted us to JMP initially,” he explains. “But what’s really amazing is that we didn’t even know that it was also a superior data analysis tool” – something it didn’t take them long to find out.

In the hands of domain experts, JMP provides a quick and easy solution for analysis on a daily basis. The fact that a single tool can be both a means for statistical experts to automate sophisticated data processes and an approachable solution for domain experts, he says, makes JMP a uniquely effective workhorse. 

The democratization of analytics helps domain experts reduce variation 

In process development, statistical methods are packaged into an automated analysis so that teams can run an experiment comparing the performance of a new design against the current version. JMP produces what O’Leary calls a “whole family of analyses” indicating whether parts are statistically the same and if not, whether they are better or worse in regard to variation and process capability. “This is relatively sophisticated analysis, but even entry-level technicians can do it,” he adds.

“With a small investment of time and effort [in learning about the software] in a structured way, once people start to use it, they don't go back. They use it more and more.” Now, O’Leary explains, anyone at Murata can have a JMP license, whether they’re a production operator on the shop floor or upper-level management.

The value of JMP goes well beyond the software 

With users at such a wide variety of statistical skill levels, O’Leary has encouraged everyone to take advantage of the full spectrum of resources JMP offers. Newcomers to statistics, for example, benefit from hours of online training modules via Statistical Thinking for Industrial Problem Solving – a free course developed by JMP to make statistics more accessible to non-statisticians.

STIPS, he says, is “so brilliant that we make it compulsory for some of the roles within our company. You must get yourself certified. And you do it during the workday. Having people complete STIPS gives me the assurance that there’s nothing in JMP they haven’t at least seen. That’s the thing about JMP – you may not use a lot of it, but at least you’ll know what's there.”

Regardless of statistical skill level, O’Leary adds, users can tap into the JMP Community to share ideas, engage in discussions and ask questions. “I might get 3 responses to a question I posted in the Community 5 minutes earlier,” he says. “The fact that you have access to developers, colleagues and fellow users – all at different levels – certainly makes the JMP experience a superior experience.”

These relationships also affect future releases of the software itself. Says O’Leary: “We eagerly await all the newest versions and actively participate in the Early Adopter program as well. There’s so much value not only in the Early Adopter program, but the amount of support we get from our JMP colleagues as a whole.”

“JMP is an enabler. You can see how they've actually engineered the software to meet your needs – it’s second to none.”

The results illustrated in this article are specific to the particular situations, business models, data input and computing environments described herein. Each SAS customer’s experience is unique, based on business and technical variables, and all statements must be considered nontypical. Actual savings, results and performance characteristics will vary depending on individual customer configurations and conditions. SAS does not guarantee or represent that every customer will achieve similar results. The only warranties for SAS products and services are those that are set forth in the express warranty statements in the written agreement for such products and services. Nothing herein should be construed as constituting an additional warranty. Customers have shared their successes with SAS as part of an agreed-upon contractual exchange or project success summarization following a successful implementation of SAS software.