Predictive maintenance is a buzzword in industrial settings—using data to predict equipment failures before they happen. Some vendors are now promoting similar concepts for LED lighting systems.

Is this useful, or is it solving a problem that doesn’t really exist? Let me share my perspective.

The Basic Concept

Traditional lighting maintenance is either:

• Reactive: Fix things when they break
• Time-based: Replace components on schedule regardless of condition

Predictive maintenance adds a third option:

• Condition-based: Monitor equipment health and intervene when indicators suggest impending failure

For lighting, this might mean tracking parameters like:

• Driver temperature over time
• Current draw variations
• Operating hours per fitting
• Lumen depreciation
• Dimming response characteristics

The promise: replace fittings before they fail noticeably, optimise maintenance scheduling, reduce emergency call-outs.

What Data Is Actually Available

Modern LED lighting systems can collect various data:

From DALI-2 Drivers

DALI-2 Part 253 specifies diagnostic data that compatible drivers can report:

• Operating hours
• Number of switching cycles
• Power consumption
• Internal temperature (some drivers)
• Fault status

This gives visibility into individual fitting operation that wasn’t previously possible.

From Lighting Management Systems

Platforms like Philips Interact, Signify, Osram Lightelligence, and others aggregate fitting data:

• Fleet-wide failure trends
• Energy consumption patterns
• Anomaly detection across installations

From Sensors

Occupancy and daylight sensors in smart lighting systems generate data about usage patterns, though this is less relevant for maintenance prediction.

What Prediction Is Actually Possible

Here’s where I get skeptical.

LED module failure: Modern quality LEDs have very long operational lives. L70 ratings of 50,000+ hours are common. Actual catastrophic LED failures are rare in quality products. Predicting gradual lumen depreciation is possible; predicting sudden failure is harder.

Driver failure: Drivers are more likely to fail than LED modules. They contain capacitors that degrade, components that can fail thermally. Some failure modes give warning signs (irregular output, temperature increases). Others don’t.

Connection failures: Loose connections, corrosion, mechanical damage—these are common failure causes. They’re not predictable from electrical monitoring.

Environmental factors: Failures caused by water ingress, physical damage, voltage events—unpredictable from normal operation data.

My honest assessment: the data can identify some problems, but many common failure modes aren’t predictable from operational telemetry.

Where Predictive Approaches Add Value

Despite my skepticism about prediction itself, the data collection can be valuable:

Fleet Health Visibility

Knowing which fittings have the highest operating hours, which have had fault events, which are running hot—this informs maintenance planning even without sophisticated prediction.

Prioritise inspection of the oldest, hardest-working fittings. That’s common sense, but data makes it actionable.

Early Anomaly Detection

A fitting suddenly drawing more power than its peers, or running hotter, might indicate developing problems. Catching this early prevents cascading failures or safety issues.

Failure Pattern Analysis

Across a large installation, data might reveal patterns:

• A particular product batch with higher failure rates
• Locations with environmental factors causing problems
• Installation issues causing similar failures

This informs warranty claims, product selection, and installation practices.

Optimised Group Relamping

For large installations, group replacement can be more efficient than spot replacement. Data helps determine when group replacement is optimal—replacing enough failing fittings to justify the mobilisation cost.

The Cost-Benefit Question

Predictive maintenance capability requires:

• DALI-2 or equivalent connected fittings
• A management platform to collect and analyse data
• Ongoing platform costs (often subscription-based)
• Staff time to review alerts and reports

For a small-to-medium commercial installation (say, 200-500 fittings), this investment is hard to justify based on maintenance savings alone. LED lighting requires relatively little maintenance compared to older technologies.

For large installations (thousands of fittings) or critical facilities (hospitals, data centres, 24/7 operations), the calculus changes. The cost of any failure is higher. The value of prevented disruption is greater.

Integration With Building Systems

The more interesting application might be integrating lighting health data with broader building management.

For buildings with sophisticated automation, lighting data is one input among many. Equipment health monitoring across HVAC, lighting, power, and other systems provides holistic facility visibility.

AI consultants Melbourne working on intelligent building systems often work with this kind of multi-system integration. Lighting becomes part of an overall building health picture rather than a standalone monitoring system.

But this requires investment across building systems, not just lighting.

My Practical Recommendations

For Most Commercial Buildings

Don’t chase predictive maintenance for its own sake. Instead:

1. Choose quality products with genuine warranties (5+ years)
2. Maintain installation records showing what’s installed where
3. Respond promptly to reported failures before they multiply
4. Plan periodic inspections (annual visual check of fittings)
5. Track warranty claims to identify problem products

This is good maintenance practice without expensive monitoring technology.

For Large or Critical Installations

Consider connected lighting with health monitoring if:

• You have thousands of fittings across multiple sites
• Failure has significant operational consequences
• You’re already investing in building automation
• Staff have capability to act on monitoring data

The value is in visibility and optimised response, not magic prediction.

For New Installations

Even if you don’t implement monitoring immediately, install DALI-2 compatible drivers and a control system that could support it. The capability exists if you want it later.

The Technology Arc

I expect maintenance-related data collection to become standard in commercial lighting over time. Not because it pays for itself through maintenance savings, but because:

• Control systems are collecting the data anyway
• Facility managers increasingly expect visibility
• Sustainability reporting benefits from operational data
• Building intelligence trends are driving integration

The question isn’t whether lighting will be monitored, but whether the monitoring will be used thoughtfully or just generate ignored alerts.

Conclusion

Predictive maintenance for LED lighting is a reasonable concept but often over-sold. The prediction capabilities are limited by what’s actually predictable. The real value is in visibility and informed decision-making.

Don’t buy monitoring technology expecting dramatic maintenance savings. But do consider how lighting data fits into your overall facility management approach.

James Thornton has been working in commercial lighting for 18 years and is based in Australia.