Commercial claims for LED-based products in terms of lumen maintenance are fully based on TM-21 extrapolations using LM-80 data. This paper indicates that there may be a risk in doing this as TM-21 only relies on the behavior of the average LED degradation, instead of taking into account the degradation of all individual LEDs. Therefore, we propose a more profound statistical analysis in order to make the appropriate step from TM-21 extrapolation to lumen maintenance on a product level. This is needed as some commercial claims are based on 10 years of warranty and some service bids provide periods of 20 to 25 years of operation. This paper reviews the different approaches currently available to perform lumen maintenance extrapolations. We propose a new method to analyze and extrapolate LM-80 data using a more profound statistical approach.@en

Per today, commercial claims for LED-based products in terms of lumen maintenance are fully based on TM-21 extrapolations using LM80 data. There may be a risk in doing this as TM-21 only relies on the behavior of the average LED degradation, instead of taking into account the degradation of all individual LEDs. A more profound statistical analysis is required to make the step from TM-21 extrapolation to lumen maintenance on product level. This is needed as the commercial claims are used as input for service bids up to periods of 20 to 25 years of operation. This paper describes the different approaches currently available to perform lumen maintenance extrapolations. For that, we have analyzed several LM80 data sets from a statistical point of view.@en