We recently completed implementation of real-time OEE dashboards across our three manufacturing sites using Delmia Apriso’s Dashboard Configuration Manager. The goal was to provide plant managers and operators with instant visibility into production efficiency metrics.
Our setup involved configuring multiple data feeds from shop floor equipment to populate OEE calculations automatically. We created role-based dashboard views showing availability, performance, and quality metrics refreshed every 30 seconds. The multi-site monitoring capability allows our corporate team to compare performance across facilities in real-time.
Key implementation steps included setting up data collection points at each production line, configuring the OEE calculation engine with our specific downtime categories, and designing intuitive dashboard layouts that highlight exceptions. We also integrated alerts for when OEE drops below threshold values.
The results have been impressive - managers can now identify and respond to production issues within minutes rather than discovering problems hours later through shift reports. Would be happy to share our configuration approach and lessons learned.
How did you approach the multi-site comparison views? Are you using a centralized dashboard or separate instances? We have clients asking about this capability but concerned about network latency between sites.
Great question. We tested various refresh intervals and found 30 seconds struck the right balance. The performance impact was minimal because we implemented data aggregation at the collection layer. Rather than querying raw machine data constantly, we have edge processors that calculate running OEE components and push updates only when values change significantly. The dashboard queries these pre-aggregated values. We also configured dashboard caching for less critical metrics that update every 2-3 minutes. Our database load actually decreased compared to the old batch reporting system.
Let me address both questions with our complete implementation approach.
Multi-Site Architecture:
We deployed a centralized Apriso instance with site-specific data collection modules. Each facility has local edge servers that aggregate production data and push to the central system every 15 seconds. The corporate dashboard pulls from this centralized database, while site-specific dashboards can operate independently if network connectivity drops. Network latency hasn’t been an issue because we’re transmitting aggregated metrics (small payloads) rather than raw machine data. We use dashboard filters to switch between individual site views and comparative multi-site displays.
OEE Dashboard Configuration Details:
Data Feed Setup:
Configured three primary data streams per production line:
- Machine state signals (running/idle/down) from PLC outputs
- Production counters (actual vs target) from MES work order tracking
- Quality rejection data from inspection stations
These feed into Apriso’s OEE calculation engine which we configured with custom formulas matching our specific business rules.
Downtime Categorization:
We implemented a two-tier approach after pilot testing. Operators select from 8 high-level categories (changeover, material shortage, equipment failure, quality hold, etc.) via touchscreen interfaces at each line. Maintenance and engineering teams can add detailed sub-codes later during root cause analysis. This keeps operator burden minimal while preserving analytical depth.
The system auto-categorizes micro-stops under 2 minutes as performance losses rather than availability hits, which better reflects our actual production constraints.
Dashboard Layout Strategy:
Created three dashboard types:
- Operator View: Large OEE gauge, current production count, active alerts. Updates every 30 seconds. Minimal clutter.
- Supervisor View: Multi-line comparison, Pareto charts of loss categories, trend graphs. Mix of 30-second and 2-minute refresh rates.
- Management View: Site comparisons, shift-over-shift trends, financial impact calculations. Mostly 5-minute refresh with drill-down capability.
Alert Configuration:
Set threshold alerts at 75% OEE (warning) and 65% (critical). Alerts route to mobile devices via Apriso’s notification system. We configured escalation rules - line supervisors get immediate alerts, plant managers receive notifications if issues persist beyond 15 minutes.
Real-Time Data Feed Optimization:
The key technical implementation was configuring data buffering and smart polling. Rather than constant database writes, we buffer production events and write in micro-batches every 5 seconds. Dashboard queries hit materialized views that refresh on this same 5-second cycle. This architecture supports our 30-second dashboard refresh while keeping database transactions manageable.
Results After 6 Months:
- Average response time to production issues dropped from 45 minutes to under 8 minutes
- Overall OEE improved 7.3% across all sites (from 68% to 75.3%)
- Eliminated 15+ hours per week of manual report generation
- Maintenance team identifies recurring issues 3x faster through trend analysis
Key Lessons:
- Start with simplified downtime categories - you can always add granularity later
- Invest time in data feed validation before going live - garbage in, garbage out
- Role-based dashboard design is critical - operators need different views than executives
- Build in data quality checks - we flag suspicious patterns like 100% OEE or impossibly high counts
- Train users on interpreting the metrics, not just viewing them
Happy to share our Dashboard Configuration Manager settings or discuss specific technical challenges you’re facing.
The downtime categorization must have been challenging. How granular did you go with your availability tracking? We’re debating between high-level categories versus detailed reason codes. Would love to hear what worked for your operators in terms of data entry burden versus analytical value.