In short. Monitoring shows whether the system is ready before it is needed. A good design measures the right points, reports to a supervisory system, keeps history for diagnosis, and raises alarms that people can act on. The monitoring itself must not become a single point of failure or a security risk.
What to monitor
| Area | Examples |
|---|---|
| Sources | Utility voltage and frequency, generator status, fuel level, battery voltage |
| Transfer | ATS position, STS state, source availability, transfer counts |
| UPS and BESS | Load, battery health, state of charge, bypass status, alarms |
| Distribution | Breaker status, currents, voltages, power factor, harmonics |
| Protection | Relay health, trips, settings group |
| Support | Cooling, ventilation, room temperature, leak detection |
Readiness indicators
Specific signals indicate whether the system can respond: generator in auto, no latched alarms, start battery voltage OK, fuel above minimum, transfer switch healthy, UPS on line with battery charged. Present them on one readiness screen so the state is obvious.
SCADA architecture
- Field devices connect by industrial protocols such as Modbus or IEC 61850. See industrial power communications.
- Local controllers act autonomously; SCADA supervises and records.
- Redundant servers and networks where the site demands it.
- Time synchronisation so events from different devices line up. See time synchronisation.
Analytics that help
- Trends: battery impedance, generator warm-up, load growth.
- Capacity tracking: loading per path against its redundancy limit.
- Power quality: voltage and harmonic trends.
- Event reconstruction: sequence-of-events and disturbance records.
- Predictive alerts: based on slow changes, not just limits.
Design principles
- Control must not depend on SCADA. Equipment keeps working if the supervisory system is down.
- Alarms must be prioritised and actionable. See alarm and event management.
- Remote access must be secured. See OT network architecture.
- Monitoring power must come from a protected supply.
Worked illustration
A site’s dual path loads are each about 45 per cent of rating, tracked on a dashboard. Over a year the sum creeps to 58 per cent on one path. A threshold alarm at 50 per cent warns before the redundancy margin is gone, and the team balances load. Without the trend, a failover would have overloaded the surviving path. The analytics value was in watching the margin, not the failure.
What to watch, and why
| Signal | Early warning of |
|---|---|
| UPS battery voltage, temperature, internal resistance trend | Weak cells, ageing |
| Generator starting battery voltage and charger status | Start failure |
| Fuel level and quality indicators | Autonomy and fuel degradation |
| Path A and path B loading | Loss of redundancy by creep above 50 percent |
| Breaker temperatures and thermography | Loose connections |
| Power quality: THD, unbalance, voltage sags | Stress on equipment |
| Room temperature | Cooling issues |
Analytics that pay back
- Capacity headroom per path, per source.
- Trending of battery and generator readiness.
- Alarm statistics showing repeat events.
- Run-hour and fuel consumption tracking.
- Event reconstruction after any disturbance.
Architecture
Meters and relays feed a gateway to the SCADA or monitoring platform, with a local view that works if the central link fails. Time-sync the devices so events line up.
Operating practice
Define who watches the data, who is called on an alarm and what each alarm requires. Monitoring without response is decoration.
See SCADA architecture and alarm and event management.
Common mistakes
- Collecting thousands of points with no readiness view.
- SCADA required for basic control actions.
- Monitoring supplied from the system it monitors, so it dies with it.
- Uncontrolled remote access.
What to send us
Send equipment lists and any existing monitoring. Send your application details. See also SCADA and BMS monitoring.
Questions engineers ask
What should I monitor first?
Readiness of sources and transfer equipment, battery health and per-path loading.
Can SCADA control the system?
It can supervise and command, but critical control should be local so it survives SCADA failure.
How long should data be kept?
Long enough to diagnose events and trends; follow your records policy.
Is remote monitoring safe?
With proper network design, authentication and segmentation, yes. Treat it as a security design task.
This guide gives general principles. Cybersecurity and network design require specialist input.