Summit MES: A Connected Manufacturing Execution System

The Problem

Summit Manufacturing Co. operates three precision-machining plants supplying components to the automotive and aerospace industries, but production visibility still relied on paper travelers, end-of-shift spreadsheets, and manual machine checks by floor supervisors. Downtime was typically discovered only after a machine had already stopped, and root-cause analysis for defects often happened days later, based on operator memory rather than actual sensor data.

The lack of real-time data made capacity planning across the three plants largely guesswork, and Summit's largest automotive customer had begun requiring documented quality traceability that the paper-based system couldn't reliably produce. Two plant managers had proposed off-the-shelf MES products in the past, but each quote assumed a level of machine standardization Summit's mixed fleet of older and newer CNC equipment didn't have.

The Solution

TechNova started with a plant floor audit across all three sites, cataloguing machine age, existing PLC interfaces, and network infrastructure to design a sensor retrofit plan that worked with Summit's actual equipment rather than requiring a fleet-wide hardware replacement. Over 400 vibration, temperature, and cycle-count sensors were installed and connected via MQTT to Azure IoT Hub, with a Kafka pipeline streaming telemetry into TimescaleDB for the time-series analysis the predictive maintenance models needed.

The MES application itself, built in React with a Python backend, gave floor supervisors a live production dashboard tracking OEE in real time and replaced paper travelers with digital work-order routing that followed each job through the plant automatically. Predictive maintenance models, trained on several months of baseline sensor data collected during the retrofit phase, began flagging anomalous vibration and temperature patterns well before failure thresholds were reached — surfacing alerts directly to maintenance staff tablets rather than requiring someone to notice a problem on a dashboard.

We rolled the system out one plant at a time, using lessons from the first deployment to refine sensor placement and alert thresholds before the second and third plants went live, which meaningfully shortened each subsequent rollout.

Results

Within two quarters of full deployment across all three plants, unplanned downtime fell 44% as maintenance shifted from reactive to predictive, and scrap and rework dropped by over a quarter as defects were caught and traced to root cause far earlier in the production process. Overall equipment effectiveness improved by nineteen points plant-wide, and supervisors now detect line issues in minutes rather than the hours it previously took to notice a slowdown through manual checks.

The quality traceability the system now produces automatically also let Summit meet its largest automotive customer's new documentation requirements without adding headcount, and the plant comparison view has become a standard input into Summit's quarterly capacity planning process.