Why Alerts Are at the Heart of IoT Monitoring
Collecting data without alerts is like building a control room without an operator. The data is there, but no one is constantly monitoring it. The value of industrial IoT monitoring lies in proactive notifications: the right message, to the right person, at the right time.
An effective alert drastically reduces the MTTR (Mean Time To Repair): the failure is detected in a matter of seconds (rather than hours, when an operator happens to walk by the machine), the technician is notified immediately, and the diagnostic process begins on the technician’s smartphone before they even head to the site.
The Structure of an Effective Industrial Alert
Types of Alerts in Industrial Monitoring
| Alert Type | Trigger | Priority | Example |
|---|---|---|---|
| High Threshold | Value > configured threshold | High | Discharge pressure > 4.5 bar |
| Low Threshold | Value < configured threshold | High | < tank level at 15% |
| No Communication | Site offline > X min | Critical | Gateway has been offline for 5 min |
| Equipment fault | Alarm log active | High | Pump fault (Modbus coil) |
| Rapid variation | ΔValue/ΔTime > threshold | Medium | Sudden pressure drop |
| Preventive Maintenance | Hour meter reached | Low | 1,000 hours of pump operation |
| Abnormal Consumption | Deviation from historical data | Medium | Energy +20% compared to previous week |
Best Practices for Configuring Thresholds
Thresholds based on actual data, not theoretical values Analyze 30 days of historical data to identify the normal operating range. Set the alert threshold at 2 sigma above the mean—not at the manufacturer’s limit.
Adjusted Confirmation Delays
- Critical alarm (safety pressure, critical level): 5 to 10 seconds.
- Process measurement (temperature, flow rate): 30 to 60 seconds.
- Trend (energy consumption): 5 to 15 minutes.
Deadband for Analog Values A pressure that fluctuates between 3.98 and 4.02 bar around the 4-bar threshold should not trigger a series of alerts. The deadband (e.g., 0.1 bar) prevents this "flapping" phenomenon.
Back to Normal Alert When the alert condition is no longer met, a "Back to Normal" notification automatically closes the incident in the log and notifies the team. It confirms that the response successfully resolved the issue.
Notification Channels
| Channel | Delay | Cost | Suitable for |
|---|---|---|---|
| Mobile push | < 1 s | None | Critical alerts, teams with smartphones |
| < 5 s | None | Detailed alert report, escalation | |
| SMS | < 30 s | Variable | Escalation, technicians without smartphones |
| Webhook | < 1 s | None | CMMS integration, Slack, Teams |
| Automatic call (IVR) | < 60 s | High | Critical on-call duties, final escalation |
The Eziwan Approach
The Eziwan Cloud includes a comprehensive alerting engine: thresholds by variable and by site, configurable confirmation timeouts, deadband, automatic escalation, acknowledgment via the mobile app, an incident log, and PDF alert reports.
For more information: remote site monitoring, industrial remote maintenance, and remote site maintenance.