Service Life & Reliability · 05

What do L70 and L90 mean?

LED sources do not fail suddenly — they dim gradually. The L-value notation describes the threshold at which a source is considered to have reached end of functional life, and the B-value tells you what fraction of a luminaire population falls below that threshold at any given point in time. Together they give a precise and verifiable description of service life that a vague "lifetime" figure cannot match.

Updated 18 June 2026 · approx. 5 min read

Share

The L-value — a lumen maintenance threshold

The notation L70, L80, L90 describes the percentage of initial luminous flux that an LED source retains at end of the rated service period. L70 means that, at the stated number of hours, the source is still producing at least 70% of the flux it produced when new. L90 means at least 90% remains.

The choice of threshold — 70%, 80%, or 90% — depends on the application and the standard being applied. For general indoor commercial lighting, L70 is commonly used. For road lighting to EN 13201, the maintained illuminance threshold may impose a tighter requirement: if the initial design was based on a specific overdesign factor (margin above minimum maintained values), dropping to 70% of initial output may push the road surface below the compliant maintained illuminance level.

This is why L70 claims in a road lighting context should always be viewed in conjunction with the maintained illuminance calculation for the specific installation — not as a standalone figure.

In practice

  • A claim of L70 > 100,000 h without a B-value defaults to B50 — meaning up to 50% of units may be below L70 at that hour count.
  • L70B10 is a considerably more conservative and useful procurement metric: only 10% of the population falls below L70 at the stated hours.
  • For critical infrastructure procurement, specify both the L-value and the B-value explicitly. Do not accept a claim with only one of the two.
  • L-values are derived from LM-80 test data via TM-21 projection — see the 100,000 hours article for a full explanation of that methodology and its limits.

The B-value — the population dimension

A single LED source tested in a laboratory follows a smooth lumen depreciation curve. A population of thousands of luminaires installed over a wide geographic area does not: individual units vary due to manufacturing tolerances, operating temperatures, thermal management quality, and incidental environmental differences. The B-value captures that statistical spread.

B10 means that 10% of the population has fallen below the L-threshold. B50 means 50%. The hour figure in an L/B notation is therefore the time at which that fraction of the population crosses the threshold — not the time at which every unit simultaneously fails.

From a maintenance planning perspective, B10 is the figure that matters: it marks the point at which failures begin to appear in meaningful numbers across an installation. If you are planning group relamping or group replacement based on service life, the B10 figure is more operationally relevant than B50.

Reading a complete notation

A properly stated lumen maintenance claim looks like one of these:

L70B10 > 60,000 h After 60,000 h, 90% of units retain ≥70% initial flux
L80B10 > 50,000 h After 50,000 h, 90% of units retain ≥80% initial flux
L90B10 > 36,000 h After 36,000 h, 90% of units retain ≥90% initial flux

These three claims describe the same hypothetical LED source at different threshold levels. Note that as the L-threshold rises (i.e., less depreciation is allowed), the hour figure falls — because it takes fewer hours to reach a smaller depreciation margin.

The test and projection methodology behind the values

L-values and B-values are derived from measured data using IES LM-80 (the test standard) and IES TM-21 (the projection standard). LM-80 measures actual lumen output of LED samples at defined temperatures over at least 6,000 hours. TM-21 projects the measured decay curve forward — capped at six times the test duration — to calculate where the L-threshold will be crossed.

This means that any L70B10 > 100,000 h claim based on 10,000 hours of LM-80 data is projecting to ten times the test duration — well beyond what TM-21 endorses as a reliable projection. The figure may be printed in a data sheet, but it carries considerably less certainty than the headline implies.

What this means for procurement

A practical procurement checklist for lumen maintenance:

  • Require both the L-value and the B-value to be stated explicitly (e.g., L70B10, not just L70).
  • Ask for the LM-80 test report, not just the derived claim. The test duration and sample size are stated in the report and are material to the projection's reliability.
  • Check that the TM-21 projection does not exceed six times the LM-80 test duration. If a 100,000 h figure is claimed, the test duration should be at least 16,667 hours to be within the standard's limits.
  • Verify that the maintained illuminance calculation for the installation was performed using the L/B figure as the lamp maintenance factor — not a higher assumed value.
  • For road lighting to EN 13201, consider whether L70 is actually the appropriate threshold for the specific lighting class and design margins used.

The full methodology behind the projection

How far can you trust a 100,000-hour claim? The answer is in the LM-80 test report.

Understanding the LM-80 and TM-21 methodology gives you the tools to evaluate a lifetime claim — not just accept it.

Service Life & Reliability

100,000 hours — what it actually means

LM-80 test duration, TM-21 projection limits, and why most long-hour claims cannot be independently verified.

Summary

L-values describe the lumen maintenance threshold — the percentage of initial flux that remains at end of rated service period. B-values describe what fraction of a luminaire population falls below that threshold. A complete notation combines both: L70B10 > 60,000 h means that 90% of units will still be producing at least 70% of initial flux at 60,000 hours.

The hour figure is derived from LM-80 test data via TM-21 projection, and is only reliable when the projection does not exceed six times the test duration. Requiring explicit L/B values — rather than accepting a standalone hour figure — gives a considerably more precise and auditable basis for comparing competing luminaire products.

Evaluating an L/B claim in a tender response?

We provide the underlying LM-80 test report — not just the derived figure.

Ask us for L70B10 and L80B10 values for VALDUR, stated with the corresponding LM-80 test duration and the TM-21 projection cap applied.