6 reasons to automate root cause analysis
Last updated on February 5th, 2024
Root cause analysis (RCA) is important to automotive manufacturers for several reasons. Firstly, it serves as a powerful tool for enhancing product quality by identifying and rectifying the fundamental causes of defects and manufacturing errors. This, in turn, reduces costs associated with rework and warranty claims while improving customer trust and satisfaction.
RCA also contributes significantly to process optimization, helping manufacturers identify and mitigate process bottlenecks, inefficiencies, and part quality concerns within their operations. Root cause analysis promotes continuous improvement, enabling manufacturers to innovate and maintain a competitive edge in a rapidly evolving industry. This is especially relevant today as the industry shifts towards newly innovative components for electric vehicles and advanced electronic systems. Manufacturing newly designed products poses a higher risk of producing defects, as there is less known about the process.
Root cause analysis is an essential part of the manufacturing process that is not going away. And RCA, like many things in the industry, is getting an upgrade with new Industry 4.0 technology. The option is now available to automate the process of root cause analysis.
What’s the difference between manual and automated root cause analysis?
For the purpose of this article, let’s define “manual” root cause analysis as it pertains to investigating the cause of parts failing an end-of-line test or having some major defect.
Manual root cause analysis is the process of using methods and/or manual data analysis to determine the root causes of a failure in the manufacturing process. The process may look different depending on the expertise of the teams performing it, and it can include a combination of:
- Investigative methods like The 5 Whys, or Failure Mode and Effect Analysis (FMEA)
- Visualizations such as Fishbone diagrams or Pareto charts
- Brainstorming sessions
- Manually gathering data from different sources
- Compiling data in Excel
- Analyzing data in tools like Minitab or Tableau
Automated root cause analysis is the process of identifying the factors which have caused a failure or nonconformance through advanced statistical methods and/or machine learning.
Automated root cause analysis is commonly performed by a single software solution that ingests process and test data. After initial factors are identified, some of these manual methods may also be used to further the investigation.
Why manual root cause analysis should be automated
1. Automating manufacturing data collection saves time
Gathering data for manual root cause analysis can be time-consuming and expensive. Manufacturers must collect data from various sources, including sensors, manufacturing software, production logs, and quality control reports. Ensuring the accuracy and completeness of this data is crucial.
Today, although machines are collecting data automatically, when it comes time to analyze it, engineers may be physically running from machine to machine with USB sticks, pulling flat files of data. They then compile the data using a business intelligence tool, such as Minitab or Tableau, and in some cases use even more basic tools, such as Excel.
The process of collecting data in this way, whenever there is a problem to be investigated, wastes considerable time. When a critical defect occurs, it may mean that production is shut down until it is resolved, and the more time it takes to resolve, the more money is lost.
When using a software solution that automates root cause analysis, manufacturing and test data is connected to the system once, and after that can be accessed and analyzed in real time. This automates the data collection part of the investigation and means that the data needed to investigate the failure is easily accessible any time a failure occurs.
2. Multifaceted root causes of defects can be identified more easily
Failures in the automotive manufacturing process often have multiple contributing factors. When a complex part such an engine is assembled and fails an end-of-line performance test, where does one even start to look for the source of the problem? Many different factors could contribute to the malfunction, such as:
- Mechanical: Caused when parts of a manufacturing machine or robot are wearing down or working improperly and cause a defect. This could be a single defect caused by one machine, such as a torque tool fastening a bolt at the wrong angle. It could also be a combination of mechanical factors that are not out of spec on their own, but add up to create a cumulative defect, such as variably sized cut pieces adding up to an out-of-spec stack height.
- Human intervention: Any station involving human intervention or manual assembly may be subject to mistakes being made or inconsistencies between shift workers.
- Material issues: Source material or parts can vary in their initial dimensions or quality, leading to unexpected variation.
When root cause analysis is performed manually, the first step is to narrow down possible causes. This is done using brainstorming sessions, methods like Fishbone diagrams, the 5 Whys, Failure Mode and Effect Analysis, manual data analysis, or even an engineer’s gut feeling. Depending on the issue, this part of the analysis can be overwhelming, and feels like looking for a needle in a haystack.
Automating root cause analysis using a data-driven approach gives engineers and quality teams a head start on their investigation by using artificial intelligence to analyze the data and present the most probable contributing factors to the failure right away. This process takes seconds, and means that anyone can work on the problem, not just the most experienced team member who knows the line like the back of his/her hand.
3. Automated root cause analysis doesn’t require a team of experts
Conducting a thorough root cause analysis using manual methods requires a team of manufacturing experts with diverse skills, such as engineers, data analysts, and quality control specialists. Hiring and maintaining a skilled team can be costly. It can be challenging to find engineers who are proficient in both manufacturing and data analysis.
Occasionally, a manufacturer will employee a data scientist or an external consultant to use more advanced techniques such as building machine-learning algorithms to identify the causes of failures. This can be an effective but time-consuming process that needs to be re-created each time an issue requires investigation. An additional challenge of this approach is the lack of manufacturing context that data scientists or consultants may have about the specific problem being solved.
When root cause analysis is automated, the data analysis and machine learning expertise is contained within the software platform. No data scientists or data analysists are needed. Engineers can focus their time on solving problems on the shop floor, instead of working with Minitab and analyzing data. This makes managing a quality team much simpler, since there is no need to acquire and retain data scientists and analysts.
4. Automated root cause analysis removes the risk of human error
Humans are complex and prone to many types of error. Even the most seasoned engineer cannot compete with data-based automation when it comes to accuracy.
When a failure occurs, there is extreme pressure to solve the problem quickly. Managers and directors may be informed and set unrealistic deadlines on engineers, causing them extreme stress. We know that under stress, the cognitive function of humans decreases, and they are more likely to make mistakes. Even if they are not affected by stress, the quality of the RCA investigation could suffer due to time pressure if there simply isn’t enough time for the team to review all relevant data.
Employees may even act as obstacles in the way of root cause investigations. If a line operator fears blame or fault being placed on him/her being the cause of a failure, they may fail to come forward with critical information. When RCA is automated, there is more reliance on pure objective data than opinion. This can reduce bias and conjecture and keep the focus on solving the problem rather than pointing fingers.
Problems with teamwork can also contribute to an inefficient or inaccurate root cause analysis investigation. For example, a roller pin was changed a week ago, and the new team doing the investigation isn’t aware of that detail. Or the team leader, senior manager, and others might have their own methods for getting to the root cause and fail to collaborate effectively.
Automated root cause analysis is as accurate as the manufacturing data it analyzes and will always perform with a high level of speed and accuracy.
5. Automating root cause analysis can increase uptime
When a failure occurs, it is often necessary to stop production while the problem is investigated. Stopping or slowing down the production line to investigate issues can result in substantial financial losses and must be minimized as much as possible.
Automated RCA can quickly process immense volumes of data in real time. It can analyze usually complex relationships between various manufacturing parameters, components, and processes at a speed that would be impossible with manual analysis. This speed is crucial for quick decision making and timely corrective actions, reducing downtime and potential revenue losses.
By automating data collection and narrowing down the possible cause of a problem with machine learning, hours or even days of unplanned downtime can be avoided.
6. Automated root cause analysis can scale
The manufacturing landscape is increasing in complexity as time goes on. More complex parts are being designed and produced that require higher levels of precision, and more data is generated by the processes that create them.
Traditional approaches to root cause analysis may not keep up with the complexity and scale of modern manufacturing processes, leading to delays in identifying and addressing critical issues. An automated root cause analysis platform can scale effortlessly to handle increasing data volumes and complexities. It can adapt to new data sources, technologies, and manufacturing methodologies to ensure seamless and continuous improvement, no matter how the technology, tools, or staff members change over time.
Automate root cause analysis in manufacturing
You can accelerate root case analysis and solve issues on the shop floor faster with a predictive quality analytics solution such as LinePulse.
LinePulse starts by focusing on the target failure and all upstream data that is collected across the process. Next, the module identifies the most probable contributing factors to the failure. The system is flexible—engineers can apply their domain knowledge to improve the analysis by eliminating any high correlation but low causation signals from the list to improve the results. Unlike relying on a dedicated data scientist or diving through Excel themselves, the analysis can be done in hours and minutes, not days and months.
One of our customers measured the comparable time it took for them to perform data-driven root cause analysis manually. The cost was nearly 50k for one failure. With LinePulse, that analysis is more efficient and affordable—it helps investigate a wide variety of production issues in a fraction of the time.
Book a LinePulse demo today to learn how you can automate root cause analysis in your plant.
Share on social: