Belden’s Connected Solution: Laying the Foundation for Predictive Maintenance in Manufacturing
Belden’s 95-year-old Richmond plant has stood the test of time. Every day, a mix of new equipment and 70-year-old machines turns out wire and cable to meet the growing demand in the US market.
However, the plant faced several challenges which put its success at risk, including:
- Costly downtime due to unplanned maintenance and hard-to-find replacement parts
- A retiring workforce of knowledgeable employees and a shortage of skilled workers to take their place
- A lack of machine connectivity and visibility
- A substantial number of older and obsolete machines in operation
The bottom line? The Richmond plant needed to find a way to eliminate unplanned downtime, reduce its reliance on skilled labor, and reliably produce more wire and cable.
Richmond’s forward-thinking vision
While the Richmond team routinely scheduled time to assess and repair the equipment, there were still occasional, surprise failures resulting in lengthy production delays. Given the age of some of the machines, difficulty locating replacement parts could delay repairs. To take the next step in their digital transformation journey, the team recognized the need to transition to predictive maintenance.
The predictive model optimizes the maintenance process. It uses trends from data insights to identify potential equipment issues before they occur. This approach provides advance warning about the need to order and prepare replacement parts.
The team had already started to update the plant. They had replaced some older machines and were using third-party software and a “homegrown” solution to track equipment performance — but these solutions only provided a limited view of the overall operation and couldn’t be scaled. To help reduce downtime and increase efficiency, they needed to connect and gather data from more machines. Plant leaders turned to data engineering consultants from Belden’s Customer Innovation Center (CIC) for help.
Belden: A guiding partner toward automated operations
Like many brownfield plants, the Richmond factory floor has a myriad of machines and devices of different vintages and makes. To connect equipment and extract, converge and contextualize the data wouldn’t be easy. But ripping and replacing all legacy equipment wasn’t an option.
The CIC team — armed with deep expertise and a clear focus on required business outcomes — designed a custom network infrastructure that could efficiently and securely connect and capture OT data from every machine. They acted as a true partner throughout, providing consistent guidance and support to the Richmond team.
“It’s really about evaluating what was already there. Taking those older systems and using our in-house solutions to provide a way to get those legacy systems into a more Industry 4.0 domain.” — Belden CIC solution consultant
Early steps and learnings
Digital automation consultants from the CIC and Richmond plant leaders identified operational goals. The consultants interviewed operations leaders and plant controllers to collect data on the potential impact of improved visibility and maintenance and helped calculate a target ROI for the pilot. The information was used to build an internal business case for C-suite approval.
Designing this proof of value was a critical and focused process. It ensured that the pilot was tightly aligned with key business objectives.
Based on the CIC team’s evaluation, the Richmond team decided to run the pilot on high-volume, high-margin production lines that had heavy backlogs. Any capacity gained would deliver extra revenue. The lines also included a variety of older equipment. “The idea was that if we could successfully implement it here, we would be able to copy and paste the process and setup to other areas of the facility fairly easily,” says Richmond’s Director of Quality.
The overall scope was to:
- Build a robust, secure network infrastructure and connect it to equipment
- Install sensors on motors and mechanical equipment to start gathering data for predictive maintenance
- Analyze data to identify trends and indicators of drift outside of tolerances that suggest pending failures, and take avoidance measures
The pilot was a success. The original OEE target was 56.5 percent. To date, the deployment has helped boost OEE to 62 percent, and the number is rising.
In eight months, the Richmond team collected around 300GB of data from 150 sensors. They detected at-risk components such as abnormal vibration readings due to belt alignment issues. And some of the data could be used for other use cases. As part of phase two, the team is working with Amazon Web Services (AWS) to enhance the system and contextualize sensor readings for more insights.
One critical component for the pilot’s success? Collaboration. The CIC worked with people across the company: supervisors, operators and plant leaders, as well as IT and OT teams throughout the design, build and implementation phases. Feedback has been positive, and adoption is high.
Future outlook and Belden’s role
Plant leaders are determining what data is needed before scaling the pilot across the plant and other facilities. They plan to:
- Expand OEE initiatives and collect more data for predictive maintenance throughout the rest of the facility by mid-2024
- Develop next-level analytics for predictive maintenance
- Use a single repository for both performance and quality data for better informed decision-making
- Build a system to automate production scheduling that leverages collected performance and maintenance data to optimize product routing and scheduling
What’s next? As one Richmond plant leader notes, “We have a proven model for collecting basic performance and maintenance-related data. We are just starting to see the potential of what is next and where this will ultimately take us.”
Learn more about how the experts at Belden’s Customer Innovation Centers help guide customers on their digitization journey.