Connecting the Factory to the Field: A Big Data and Six Sigma Case Study

8/12/2024

BY: ANTHONY WELSH

It's no surprise that advancements in Industry 4.0 (I4.0) have opened new opportunities for  Six Sigma practitioners. As data sources get larger, competition grows, and customer expectations become more demanding, we require new tools to tackle the challenge. Machine learning and predictive analytics are an interesting and effective avenue that many have been reluctant to pursue–me included! To show how taking the leap into this technology can reap significant rewards, let’s review a case study demonstrating how these tools helped process improvers find signals in large, initially disconnected datasets. 

UNDERSTANDING THE LANDSCAPE – DEFINITION OF TERMS

Six Sigma, or what I like to call OpEx (Operational Excellence), is a data-informed approach to process improvement. Through the application of proven tools and critical thinking, practitioners discover key insights about the y=f(x) of their process: first, ensuring that it is stable, predictable, and consistent; then, learning what inputs they can control to get outputs that are on target and with little variability.  

Artificial Intelligence (AI), Machine Learning Algorithms, and Predictive Analytics are all interconnected I4.0 technologies that deserve their own articles to fully define, but in the broadest sense, they involve using computers to find intricate relationships within large complex datasets. These tools can identify patterns, correlations, and outliers that may not be apparent while using traditional tools like process sampling, multiple regression, or design of experiments (DOE).

PROACTIVE QUALITY IMPROVEMENT

The marriage of Six Sigma, Big Data, and Machine Learning facilitates a shift from reactive to proactive quality improvement. In the past, an unforeseen quality problem could make it out into the field. A spike in warranty issues would trigger a project to resolve the problem. This approach requires the lagging indicator of the field spike to identify the problem. Rather than addressing warranty issues only after they arise, manufacturers can now anticipate and mitigate potential problems by understanding the nuanced connections between production metrics and field performance. This proactive approach enhances product quality and contributes to consumer satisfaction and brand loyalty. Utilizing these big data-enabled models empowers Six Sigma practitioners to predict potential warranty issues in the field based on factory performance metrics, ultimately leading to more effective preventative measures. Ultimately, if you only use lagging indicators to monitor quality, you'll never get ahead of your problems.

DEMONSTRATING THE POWER OF SIX SIGMA WITH CONSUMER APPLIANCE PRODUCTION 

An old colleague of mine, Jim Oskins, MBB currently working at Minitab, has utilized some of these approaches to find actionable information from large and complex datasets.  While working for one of the world's leading home appliance makers, Jim was part of a team of Six Sigma Master Blackbelts trying to expand upon the success of a program saving $30M annually by adding predictive analytics tools to the mix. When asked about the program, Jim provided these insightful responses 

Q: Why do you think these tools were met with some initial skepticism?
A: Probably a multitude of factors, varying by individual, but here are some noticed trends through various discussions.

• Why start something new when we already have our hands full solving problems we know how to solve with methods that have proven to be effective on past projects?
• An expressed fear that the data analytics tools are too advanced (or, reframed, a lack of belief in our own skills or methods)
• Disbelief in data science trends compared to proven Six Sigma and OpEx methods.
• It sounds too similar to regression, which requires a lot of skill to understand
• It’s hard enough to get basic users to become proficient in what you expect them to use
• To some extent, leaders may not believe they have the right resources to deliver with new methods

Q: What enablers were required up front for this initiative to be successful?
A: We needed management committed to the vision and willing to approve the necessary resources. As mentioned above, fear of the tools or lack of confidence in the data can be a stumbling block at implementation. So, in our case, hiring outside consultants who brought that experience and confidence with them was helpful to get things going. Although, as you can imagine, that did come at a high cost.  

It’s impossible to make a model with no data, so you need some proven measurement systems and good data to start. In our case, this came from both customer complaints databases and factory databases. You need a way to connect them, such as the product serial number. 

Q: Can you describe some of the program’s accomplishments? What were you able to achieve?
A: Our goal was to identify and prioritize projects that directly impact emerging customer issues. Keep in mind that linking field failures to factory data doesn’t solve the problem, it provides insight, which can then be acted on proactively.  These methods won’t replace your quality team.  An example success story might look like:

• The data analytics model found that Failure “A” appears to emerge when factory data inputs “x1, x23, x87” behave a certain way.
• Assign a cross-functional team to ask the right questions using this signal.
• Quality Engineer asks a Design Engineer: How could “x87” contribute to failure “A”?  Design Engineer: “I thought that might happen. It would be caused by A or B scenario.”  

The signals from the model can encourage reprioritization, new thinking, and questions that become paths of investigation.  This leads to learning that wouldn’t have happened otherwise.

Examples like the one above were helpful because instead of waiting half a year for service incidents to pile up, leading to a quality improvement project, field failures could be linked to factory data faster… and factory data that suggested a field failure was possible would be proactively improved. Finding problems and asking good questions sooner prevents wasting money and resources due to poor quality, and protects your brand reputation. 

Q: What steps could a small or medium-sized manufacturer in Michigan take to achieve something similar?
A: My companies were indeed all giants, and quality cost them millions or billions of dollars, I suppose! But you can still use these concepts and tools to find issues faster, link field failures to factory data, arm cross-functional teams with the right questions to lead to answers, and prevent failures in the future by monitoring those key inputs found with these methods. Small and medium size companies probably cannot hire the top tier of consultants due to cost, but you can take advantage of Minitab’s lower cost Predictive Analytics.  You can teach yourself, you can utilize Minitab’ s training (instructor led or e-learning) and you can utilize Minitab’s analytics consultants.

I agreed. More resources exist than ever before to help small and midsize companies leverage predictive analytics and Industry 4.0 technologies. MMTC can help companies find the technologies and software that meet their needs and their budgets. 

What comes next?

Have you seen any other examples of Industry 4.0 applied to Six Sigma or Operational Excellence work? As manufacturing professionals our overall goals remain the same—to design and implement good processes. However, the tools available are ever-changing and can help us become even more efficient and effective in achieving these goals. Check out MMTC’s upcoming Six Sigma courses or contact us today for personalized assistance. 

Since 1991, the Michigan Manufacturing Technology Center has assisted Michigan’s small and medium-sized businesses to successfully compete and grow. Through personalized services designed to meet the needs of clients, we develop more effective business leaders, drive product and process innovation, promote company-wide operational excellence and foster creative strategies for business growth and greater profitability. Find us at www.the-center.org.

Categories: Advanced Manufacturing, Industry 4.0, Innovation, Six Sigma, Smart Manufacturing, Technology