We successfully implemented real-time shop floor data integration between our MES system and Oracle Fusion Supply Planning. Our challenge was eliminating the 24-hour lag in production data that caused forecast inaccuracies and frequent stockouts.
Our manufacturing facilities use IoT sensors on production lines feeding data to a Siemens MES system. Previously, we exported daily batch files to update supply plans, but this delay meant planners worked with outdated information.
We built a REST API integration that streams production completions, WIP status, and material consumption events directly into Supply Planning. The integration leverages Oracle’s Supply Chain Planning REST APIs to update supply plan parameters in near real-time.
Key implementation approach:
// Pseudocode - MES to Supply Planning data flow:
1. IoT sensors capture production events (completion, scrap, downtime)
2. MES system aggregates data and triggers webhook to middleware
3. Middleware transforms MES format to Oracle SCM data structure
4. REST API calls update Supply Planning work orders and inventory
5. Supply plan automatically recalculates based on actual shop floor status
// Integration runs every 15 minutes with event-based triggers for critical updates
After three months in production, we’ve reduced stockouts by 34% and improved forecast accuracy by 28%. Planning cycles shortened from daily to every 4 hours. Would love to share detailed architecture and lessons learned.
This is exactly the type of integration we’re exploring. How did you handle the data transformation layer between MES and Fusion? We’re struggling with mapping our shop floor data model to Oracle’s supply planning entities. Did you build custom middleware or use an integration platform?
What’s your approach to handling data quality issues from IoT sensors? We’ve had problems with sensor noise and false readings that would corrupt our supply plans if pushed through automatically. Do you have validation rules in the integration layer?
How does this integration impact your Supply Planning calculation performance? I’m concerned that frequent updates might trigger excessive plan recalculations and slow down the system.
Excellent question - data quality was our biggest concern initially. We implemented three validation layers. First, the MES system itself filters obvious sensor anomalies using statistical process control rules. Second, our OIC integration includes business rule validation that checks for physically impossible values like negative quantities or completion rates exceeding 100%. Third, we maintain a staging table in Oracle that holds data for 5 minutes before final commit, allowing automated reconciliation checks against expected ranges. Any record failing validation triggers an alert to operations and gets quarantined for manual review. We also built a dashboard showing data quality metrics and rejection rates by production line.
I’ve implemented similar architectures for three clients. The performance concern is valid but manageable. Oracle’s incremental planning engine only recalculates affected portions of the supply plan. The key is properly configuring planning parameters and using collection-based planning rather than full replans. Set your planning time fences appropriately and use planning exception messages to highlight significant changes. Also consider using Oracle’s event-driven planning features in 24A which are specifically designed for this use case. You might want to batch minor updates and only trigger immediate replans for material exceptions or critical constraint changes.