<![CDATA[Induction heating machines are critical production assets, yet their performance is often monitored manually through local HMI panels. This practice can delay Overall Equipment Effectiveness (OEE) reporting and introduce inaccuracies. Existing real-time OEE architectures for Odoo-based MES platforms have not been demonstrated on induction heating equipment. In such installations, standalone HMI panels remain disconnected from centralized OEE systems, leaving operators dependent on delayed and error-prone manual reporting. This paper presents the design and implementation of a real-time OEE monitoring system integrating Odoo 18 MRP with an Omron NB HMI via direct Modbus TCP communication, without intermediate middleware. The system comprises three custom Odoo modules: mrp_oee_custom (OEE engine and REST API), mrp_oee_modbus (Modbus TCP I/O and state machine), and mrp_shopfloor_custom (shop floor monitoring). The core mrp_oee_modbus module uses a background polling thread, with a cron-based fallback, to read six HMI registers and write OEE values to 24 output registers. This enables operator-level OEE display without browser access. System reliability was evaluated through a 26-day uptime test (130 operating hours) and 12 fault-injection scenarios. Communication performance was assessed by analyzing 26,040 Modbus TCP transactions from three Wireshark capture sessions. Functional correctness was verified through 14 Odoo and 10 HMI black-box scenarios, while OEE calculation accuracy was validated across five production runs. The results showed: uptime of 99.69%, an MTBF of 32.4 hours, an MTTR of 6 minutes, a 100% auto-recovery rate, mean Modbus TCP latency of 5.92 ms with 0.00% packet loss (P99 = 33.4 ms, jitter:4.99 ms), and 100% functional test pass rate. OEE calculations deviated from manual calculations by a maximum of 0.005 percentage points (mean deviation: −0.001 pp). This error is substantially below the 1-percentage-point resolution commonly used for OEE-based production decisions and therefore has negligible practical impact on monitoring accuracy.]]>
