Thinklytics

Benchmark Electronics · Manufacturing · San Antonio, TX · 12 weeks

OEE monitoring automated across 6 plants

A precision parts manufacturer with six plants relied on spreadsheets to track Overall Equipment Effectiveness, causing a two-day delay in reporting downtime events. We developed an automated OEE monitoring system that uncovered $3.7 million in annual recoverable downtime and cut unplanned stoppages by 31 hours each month.

Challenge

The manufacturer relied on spreadsheets to track OEE, causing a two-day delay before downtime showed up in reports. Plant managers couldn’t act quickly because the data was always outdated. They suspected $3M to $5M in lost production annually but lacked accurate measurements to confirm or address the issue.

Approach

We connected machine data from six plants using OPC-UA, then calculated OEE in real time. We created a Tableau dashboard that highlighted downtime patterns and identified root causes. We also developed predictive maintenance alerts based on anomalies in vibration, temperature, and cycle time to catch failures early.

Outcome

We cut OEE data delays from two days to 15 minutes, enabling faster decision-making. By analyzing data across six plants, we uncovered $3.7 million in avoidable downtime annually. In the first three months, unplanned stoppages dropped by 31 hours each month. Predictive maintenance alerts caught four major equipment failures within six months, preventing costly shutdowns.

How We Got OPC-UA Connectors Talking and Data Moving Fast

Here’s the deal: those six plants were all running different machine control systems, four in total. So, we rolled up our sleeves and built OPC-UA connectors for each one. That let us grab machine data directly and keep it consistent across the board. The kicker? We ditched that painful 48-hour lag from manual data pulls. Everything just clicked after that.

How we slashed downtime and saved big with predictive maintenance

Real-time OEE tracking gave us some good data, but the real win was the predictive maintenance alerts. They flagged odd spikes in vibration, temperature, and cycle times way before things got serious. Thanks to that early warning, we fixed problems before they became full-blown breakdowns. Over six months, that helped us avoid four major shutdowns and saved between 8 and 16 days of surprise downtime.

Getting Plant Managers to Use the System, No Headaches, No Hassle

Plant managers are swamped, right? They don’t have time to sift through messy reports. So, we built a dashboard that’s super simple. Each machine gets a health score from 0 to 100. We threw in some color coding, red, yellow, green, so they can instantly see which machines are acting up. No guesswork. No stress. Just clear, quick info.

Results

  • $3.7M Recoverable downtime identified annually
  • 31 hrs/mo Unplanned stoppages reduced
  • 48 hrs to 15 min OEE data lag
  • 12 wks All 6 plants live

We knew some downtime was recoverable, but we didn’t have a clear number on the cost. It’s about $3.7 million a year. Since we started using the predictive maintenance alerts, we’ve prevented four major failures. The system pays for itself every month.

VP of Manufacturing Operations, Precision Parts Manufacturer

In the client's terms

The question

Which of this work should a person still be doing?

The situation

OEE was tracked in spreadsheets across six plants with a two-day lag, so plant managers were always acting on a picture that had already changed. Lost production was suspected at $3M to $5M a year and could not be confirmed.

What we found

Nobody could act on the number because nobody could act on a number that was two days old. The measurement problem and the management problem were the same problem, and adding analysis on top of a two-day lag would not have touched either.

What it produced

OEE lag fell from 48 hours to 15 minutes. Cross-plant analysis identified $3.7M a year in recoverable downtime. Unplanned stoppages fell by 31 hours a month in the first quarter, and the alerts caught four major equipment failures in six months.

The roadmap

Machine data from all six plants was connected over OPC-UA with OEE calculated in real time. A Tableau view surfaced downtime patterns and root cause rather than only the headline figure. Predictive maintenance alerts were built on vibration, temperature and cycle-time anomalies so a failure is caught while it is still a signal.

The work behind this result

Related case studies

Frequently asked questions

How current does OEE data need to be?

Current enough to act on the shift it describes. Benchmark Electronics tracked OEE in spreadsheets with a two-day lag across six plants, so managers always acted on a picture that had changed. Cutting the lag to 15 minutes exposed $3.7M a year in recoverable downtime.

What does predictive maintenance catch that OEE tracking does not?

Failures before they happen. Alerts on vibration, temperature and cycle-time anomalies at Benchmark caught four major equipment failures over six months and avoided somewhere between 8 and 16 days of unplanned downtime, on top of the 31 hours a month of stoppages removed.

How do you connect machines running different control systems?

One connector per system. The six plants here ran four different control systems, so an OPC-UA connector was built for each, which made machine data consistent across the estate and removed the manual data pulls.