Back to results list
Show full item record
Please use this identifier to cite or link to this item:
|Title:||System level lifetime prediction and carbon footprint qualification for led lighting products||Authors:||Ibrahim, Mesfin Seid||Degree:||Ph.D.||Issue Date:||2021||Abstract:||With the advancement of technology and the increased demand for better quality of life, artificial lighting has become a central part of everyday activity. The introduction of high-power white Light-emitting Diodes (LEDs) is one of the key innovations that has revolutionized the lighting industry in the past few decades due to multiple benefits in terms of high reliability, long lifetime, low energy consumption, and small size when compared to the traditional lighting sources. Nowadays, LEDs are being used in various applications including general lighting, traffic, and display lighting, medical, and communication. Although LEDs are also known for their higher reliability, it has brought another challenge for manufacturers in obtaining sufficient failure data to assess the reliability and estimate the remaining useful lifetimes (RUL) of such products in relatively short lifetime testing periods. In addition, LED products are a complex optoelectronic assembly manufactured from many components (such as LED chips, electrical drivers, encapsulant materials, heat-sink components, and so on). Because of the interaction of the different components, LED systems are also known to have a large number of failure modes and failure mechanisms. This makes the system level reliability assessment and lifetime prediction of LED lighting products challenging. On the other hand, pertaining to the longer time of operation of LEDs, greenhouse gas (GHG) emissions take place for an extended period which contributes to climate change. However, little is known about the environmental life cycle impact, including the carbon footprint of LEDs within its whole lifecycle. Currently used reliability assessment methods have been found prone to prediction errors and experimental tests take a very long time to collect reliability information. In order to address this research gap, we used stochastic process models, the Bayesian network method, and the cradle-to-grave system boundary life cycle assessment (LCA) approach. The first part of this research focused on developing an accurate and effective degradation model for performance degradations of LEDs. The required degradation data were acquired through the thermally stressed accelerated degradation test (ADT) and the Gamma process model, with the capability of handling temporal variability and measurement dynamics, applied to analyze the degradation path. Two methods were employed to estimate the model parameters, maximum likelihood estimation (MLE), and method of moments (MM). In terms of model parameter estimation, the MLE approach has clear superiority over MM due to its iterative algorithm. The long-term lifetime prediction results based on 45% of full lifetime lumen maintenance data showed that the Gamma process approach proved to have much better prediction accuracy compared to the NLS regression-based industry standard. Similarly, the Wiener process was also demonstrated where the parameter estimation was carried out with a likelihood function and the lumen maintenance lifetime prediction was validated in comparison with the NLS regression approach. In addition, the Bayesian model was also proposed to model the lumen maintenance lifetime of LED packages and lamps. In the analysis, the Markov Chain Monte Carlo (MCMC) sampling-based Metropolis Hasting algorithm was used to estimate the model parameters in the form of probability distributions. .
In the second part of the study, a novel experimental setup was designed to enable the estimation of the product/system level lifetime of LED lamps based on the degradation of the main subsystems/components (LED module, driver, and Diffuser and Reflector). The performance degradation data for the subcomponents were modeled based on the Gamma process and Weibull distribution, while the LED lamp level analysis was based on the Bayesian network method. The degradation and prediction results showed that LED modules contributed a major part in the lumen degradation of LED lamps followed by drivers, with the least effect from the diffuser and reflector. The proposed approach was found to be effective in evaluating and addressing the long-term reliability assessment concerns of high-reliability LED lamps and in fulfilling the guarantee of high prediction accuracy in less time and a cost-effective manner. The final part of the study focused on addressing the environmental impact of LEDs that serve and emit GHG for an extended period of time. Environmental lifecycle impact assessment methods, Eco-indicator 99 and CML-2001, were used in the LCA and carbon footprint qualification to estimate the potential contribution of LEDs to global warming by quantifying all the significant GHG emissions and removals over its lifecycle cradle-to-grave, in the context of Hong Kong, according to the ISO 14067 standards requirements. The results revealed that the use phase has a major environmental impact followed by the manufacturing phase, which is mainly due to energy consumption and the source of energy production. The scenario analysis results showed that prospective sources of energy production (greener) would provide a 61.22% reduction in the global warming potential in lighting sources for Hong Kong compared to the current energy mix
Light emitting diodes
Hong Kong Polytechnic University -- Dissertations
|Pages:||xiv, 242 pages : color illustrations|
|Appears in Collections:||Thesis|
View full-text via https://theses.lib.polyu.edu.hk/handle/200/11276
Citations as of May 22, 2022
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.