Top 10 Manufacturing Data Sources Every Plant Should Connect

Start with the Tier 1 essential sources: PLC, SCADA, MES, and ERP. These four sources provide the foundational data for production monitoring, OEE tracking, quality analysis, and cost visibility. PLCs deliver real-time machine states for availability and performance metrics. SCADA adds process parameters for quality correlation. MES provides production order context, WIP tracking, and serial-level genealogy. ERP brings financial and inventory context. Together, these four sources enable plant-wide operational dashboards, and connecting them typically covers 80% of analytics use cases. Once these are established, add CMMS for maintenance analytics, QMS for structured quality data, and energy meters for sustainability reporting. Vision systems and IIoT sensors are advanced additions that require more infrastructure but deliver high-value capabilities.

Legacy PLCs and sensors without native IP connectivity can be connected through several approaches. The most common method is deploying an industrial gateway or edge device that bridges serial protocols (RS-232/RS-485, Modbus RTU) to modern IP-based protocols like OPC UA or MQTT. Alternatively, if SCADA is already in place, connect the analytics platform to the SCADA historian rather than directly to each PLC. For sensors with 4-20mA or 0-10V analogue outputs, use an I/O module with Ethernet connectivity to digitise the signal. iFactory's pre-built connectors support over 300 industrial protocols including Modbus RTU, DF1, Profibus, and ControlNet, making legacy integration achievable without replacing existing automation infrastructure. The key is to minimise disruption to existing control systems while extracting the data needed for analytics.

The best integration approach depends on your ERP and MES architectures, but a pattern emerges across successful implementations. First, establish a data warehouse or lake as the intermediary layer rather than point-to-point connections. Extract ERP data through standard APIs (REST, SOAP, ODBC) or IDoc messages for SAP. Extract MES data through REST APIs or direct database access. Map common entities (production orders, materials, work centres) between the two systems, reconcile status updates, and load into the analytics layer with timestamps for temporal correlation. Batch synchronisation at 15-minute intervals is usually sufficient for operational reporting. For near-real-time use cases, implement event-driven streaming where ERP order releases and MES production completions trigger immediate updates. iFactory provides pre-built connectors for SAP, Oracle ERP, Microsoft Dynamics, Siemens Opcenter, Rockwell MES, and 15 other platforms to accelerate ERP/MES integration.

Refresh frequency should match the decision timeframe of the use case. Real-time sources like PLCs and vision systems should update at sub-second to 5-second intervals for operator dashboards, alarm detection, and real-time SPC. IIoT sensors and energy meters refresh every 1-10 seconds for trend monitoring and peak detection. SCADA data should refresh at 1-5 second intervals for process visualisation. Batch sources follow a different cadence: MES refreshes every 1-15 minutes (production completions, quality results), ERP every 15 minutes to 1 hour (cost postings, inventory updates), and CMMS and QMS every 15 minutes to 1 hour (work orders, quality records). Operator input via mobile forms should be submitted immediately at the time of observation. The key principle is to align refresh frequency with the speed of the operational decision: faster decisions need fresher data, while strategic analyses can tolerate longer refresh intervals.

Each data source requires tailored quality checks. For PLC and SCADA data, implement range validation (temperatures within feasible bounds), rate-of-change checks (reject spikes), timestamp consistency (ensure clock synchronisation), and completeness monitoring (expected tag updates per minute). For MES data, validate order status transitions (cannot go from Completed to In Progress), quantity consistency (good + scrap + rework = total), and timestamp ordering (start before end). ERP data needs reconciliation checks (cost debits equal credits), duplicate order detection, and currency/unit consistency. CMMS data should be checked for work order completeness (mandatory fields populated), duration reasonableness, and asset hierarchy consistency. Vision system data requires image capture rate monitoring and pass/fail ratio reasonableness. Operator-entered data needs mandatory field enforcement, reason code validation against a defined list, and time-window checks (entries cannot be backdated beyond shift start). Implement automated data quality dashboards for each source with alerts when quality drops below thresholds.