Mixed-Color LED Lighting with Circadian Benefits Y. J. Tam, V. Kalavally, J. Parkkinen Department of Electrical and Computer Systems Engineering, School of Engineering, Monash University, Jalan Lagoon Selatan, 4615 Bandar Sunway, Selangor, Malaysia Contact: vineetha@monash.edu ABSTRACT In addition to providing vision, light is known to have several physiological, neurobehavioral and circadian-related effects. These non-visual effects of light such as shifting of circadian rhythm, alertness and hormonal production, are generally mediated by the melanopsin-containing intrinsically photosensitive retinal ganglion cells (iprgcs). We compute the melanopsin response for different times of the day as well make a comparison of the effect of seasonal changes. We optimise a mixedcolor LED lighting system for high melanopic-lux in addition to high Colour Rendering Index and Luminous Efficacy. These findings have important implications for future lighting architectural design for maintaining alertness and improving the wellbeing of humans. INTRODUCTION Recently, there has been a discovery of a specialized, novel type of photoreceptor in the human eyes, the melanopsin-containing intrinsically photosensitive retinal ganglion cells (iprgcs), which mediates a number of non-visual light responses, in addition to the conventional rods and cones [1]. These newly found non-rod, non-cone photoreceptors that contain the photo pigment melanopsin have sparked a growing interest in incorporating these non-visual effects of light into building design [2, 3] as well as lighting design for the health and wellbeing of humans. In addition to improving sleep and alertness as well as circadian related effects, there are several other biological effects of light such as pupil dilation, suppression of melatonin production, increases of heart rate and core body temperature [4] and even reduced depression and eased pain [3]. According to [5], the melanopsincontaining iprgcs are more sensitive to short wavelength light than the conventional rods and cones. These iprgcs have a sensitivity curve which peaks near 48nm as compared to human photopic and scotopic vision curves which peak around 555nm and 55nm respectively. Researchers have identified several properties of light exposures which are important in affecting the human circadian system. In addition to displaying a different sensitivity to different light spectrum, the human non-visual system also responds differently to different light intensity, duration, timing, pattern of light received at the eye as well as prior light history [6,7].These are also the major factors which determine the entrainment of the circadian cycle [5]. Together with an increased understanding of light effects on the human well-being, studies about this effect have also increased. For example in [8], a small scale model was used and one section of the office was used as an experimental area. In [2], an indoor modelling method is used. Given that multicolour LED lighting systems will be favoured over white LED systems in future for their larger colour gamut and spectrum tunability, such lights can be designed and utilized to provide benefits taking into consideration both the visual and non-visual effects of light. This paper presents an LED-based light source with its spectrum optimized for high quality white light and high melanopic lux through the computation of visual and non-visual responses to light from human photoreceptors. SEASONAL AND DAYLIGHT DEPENDENT MELANOPSIN RESPONSE The measurement of changing light spectra in a room in the presence of both artificial light (fluorescent) and sunlight was done for both Finland and Malaysia to observe the effect of seasonal and daylight changes on the melanopsin response. Generally, a room with florescent lighting with window has been selected to observe the influence of daylight on the melanopsin response. The room 443
was divided into four zones as shown in Fig.1(a) and a number of measuring points were selected for each zone which were then averaged and 4 as shown in Fig.1. Fig. 1. Room zoning (Finland and Malaysia) The photopic and melanopic luxes are computed for zones 1-4 using methods outlined in Ref [4]. The results for summer and winter day time and night time in Finland are illustrated in Fig. 2. Fig. 2. Effect of illuminance on melanopsin response for different times of a day and seasons in Finland From Fig. 2, it can be seen that the melanopic lux values are dependent on the illuminance with higher melanopic lux value calculated for higher illuminance. From the bar plot for summer daytime, it can be seen that zones 3 and 4 which are near the window received higher amount of daylight and hence illuminance which results in a higher melanopic lux value. Daylight spectrum would better stimulate the response of melanopsin containing photoreceptors as daylight is the natural cue for the synchronized entrainment of the circadian rhythm. As such, for the case of winter and night time where the contribution from sunlight is minimal, the melanopic lux values calculated are lower. The summer night time melanopsin and photopic luxes are lower than winter night time responses due to the presence of sunlight during summer nights in Finland. 444
Fig. 3. Radar chart for comparing the melanopsin responses for each zone at different times of a day and different seasons The effect of daylight and seasons on the melanopic response in the 4 zones of the room can be illustrated using radar charts shown in Fig. 3. The melanopic lux values calculated are much higher for summer days compared to the rest. During night time, the contribution of light source is mainly from indoor artificial lighting which is fluorescent lamp in this case. As for winter, the relatively low melanopic lux values calculated is due to the minimal contribution of sunlight which also explains the moodiness and depression that people experience during winter time [9]. In Malaysia with tropical weather conditions and uniform weather throughout the year, the spectral measurements were taken at 2 hours interval starting from 8am to 6pm. Zones 3 and 4 corresponding to areas which are near the window has the higher melanopic lux values calculated as shown in Fig. 4. Besides that, it can also be seen that the melanopic lux value calculated for each zone is quite consistent with changes in the daylight illuminance with the highest value recorded around 2pm. This is because the contribution from sunlight is at maximum around that time. 8 6 Malaysia Zone 1 8 6 Malaysia Zone 2 melanopic lux photopic lux (illuminance) 4 4 2 2 8112141618 8112141618 8 Malaysia Zone 3 8 Malaysia Zone 4 6 6 4 4 2 2 8112141618 8112141618 Fig. 4. Effect of illuminance on melanopsin response for different times of a day in Malaysia 445
Fig. 5. Radar chart for comparing the melanopsin responses for each zone at different times of a day The effect of time of the day on the melanopic response in the 4 zones of the room in Malaysia is shown using radar charts shown in Fig. 5. One interesting finding is that the melanopic lux values calculated for Finland is lower than the photopic lux whereas it is the inverse for Malaysia. This can be attributed to the differences in geographical location as well as the position of sun with respect to the two countries. Another explanation can be seen from the light spectrum measurements. The spectrum measured at Finland has Spectral Power Density (SPD) which are less peaky near the peak sensitivity of melanopsin response which is around 446-47nm. On the other hand, the spectrum measured in Malaysia has much higher SPD values around that region. Another possible explanation is the differences in the type of fluorescent lamp used for indoor lighting. Overall, it can be seen that the melanopsin response is strongly dependent on daylight and seasonal changes as observed from the case of two different countries which brings the need to be able to design lighting systems with an optimum spectrum maximising both visual and non-visual effects of light. OPTIMISATION OF MIXED-COLOR LED FOR HIGH MELANOPIC LUX After analysing the daylight and seasonal dependence of the melanopsin response in humans, it seems possible that with careful design of artificial lighting in a room, we can compensate for lower melanopic and photopic luxes. We thus proceed to optimise a mixed-color LED light source for high melanopic and photopic lux in a room. We ignore the sunlight spectrum in the room and carry out the optimisation based on five colors of LEDs (royal blue, cyan, lime green, orange and dark red), combination of which is reported to be capable of generating high visual quality white light [1]. Apart from melanopic lux, color quality metrics such as CRI and luminous efficacy were included as objective functions in the optimisation process. The final optimised spectrum will thus cover three important aspects of light which are visual performance (CRI), energy efficiency (luminous efficacy) and non-visual responses (melanopic lux). Steepest descent hill climbing method [11] was used to find the solutions which will maximise all three objective functions. The multi-objective problem is formulated as follows: maximise f 1 (x), f 2 (x) and f 3 (x), where f 1 (x) is CRI, f 2 (x) is luminous efficacy and f 3 (x) is melanopic lux. The corresponding solution or decision vector in this case is x = [i 1, i 2, i 3, i 4, i 5 ] where i represents the intensity for each LED. Steepest descent hill climber is a stochastic search method whereby the algorithm is initiated with some starting LEDs intensities and the objective functions are evaluated for this light. A random 446
perturbation from to 1 is then added to the intensity of the initial point to create a new spectrum. Next, the objective functions which correspond to this new spectrum are calculated. If it shows an improvement as compared to the initial point, these parameters and its corresponding objective functions value are stored in a matrix. For every data stored, there is an associated angle value which is calculated between this new point and the initial point. The new point which gives largest inclination angle will be chosen for every iteration. This is then repeated until it reaches the maximum iteration value set at the beginning. As the algorithm runs, it will eventually reveal the Pareto Front along the process until reaching the local maxima. Fig. 6. 2D plot showing the comparison between each objective function as well as relationship between melanopic lux and Correlated Colour Temperature (CCT) For each simulation, although the optimal fronts obtained are not exactly the same, they show similar pattern and shape. From the results for 1 simulations summarised in Table.1, it can be seen that on average the optimized spectrum obtained using Steepest descent hill climbing method was quite consistent with an average CRI of 92, luminous efficacy of 348 lm/w and melanopic lux value of 62. Besides that, it can be noted that the optimized spectrum has melanopic lux value which is as high as those for daylight. This is a positive finding as artificial lighting based on mixed color LED can now also achieve similar level of melanopsin response as natural daylight in addition to providing high quality white light. No. i 1 i 2 i 3 i 4 i 5 K (lm/w) CRI Melanopic-lux delta uv CCT 1.69.35.57.75.1 357 9 58.313 4.1E-3 3453 2.69.42.58.67.32 341 92 62.256 5.4E-3 3582 3.62.39.55.86.1 359 9 58.113 4.6E-3 3283 4.73.31.61.73.1 359 9 58.616 4.2E-3 3489 5.76.52.62.92.4 353 9 71.283 4.6E-3 3445 6.56.43.47.6.4 332 93 55.351 5.4E-3 3542 7.7.54.45.61.1 342 93 64.137 4.4E-3 3929 8.6.51.42.57.19 334 94 58.81 5.1E-3 388 9.64.44.53.77.3 354 91 6.415 5.E-3 346 1.76.62.54.78.4 345 92 73.488 5.E-3 3731 Table 1. The optimized characteristics of the LED spectrum based on 1 runs of simulation 447
RESULTS AND CONCLUSION In conclusion, we have analysed and discussed the daylight and seasonal dependence of melanopsin response as observed for two different countries in different zones of a windowed room. There is noticeable difference in the melanopsin response due to the differences in geographical locations. Besides that, we have also identified an optimum spectrum that can be generated using a set of commercially available LEDs which satisfies the criteria for high quality light with high melanopsin response. In future studies, the effect of melanopsin response on various human activities will be studied and the design of the light source will be tailored to suit these activities. The simulation results will be verified using psychophysical tests. It is hoped that this study would provide a stepping stone for future research to develop better lighting in terms of visual performance, energy efficiency and non-visual responses. References [1] Hattar et al. (22), Melanopsin-Containing Retinal Ganglion Cells: Architecture, Projections, and Intrinsic Photosensitivity, Science 8, Vol. 295 no.5557 pp.165-17. [2] M. Andersen et al. (213).Modelling non-visual effects of daylighting in a residential environment.building and Environment.[Online].vol.7. pp. 138-149. [3] M. Anderse et al. (212), A framework for predicting the non-visual effects of daylight Part 1: Photobiology based model, Lighting Res. Technol., vol. 44, pp. 37-53, Feb. 212. [4] Lucas et al. (213). Measuring and using light in the melanopsin age. [Online]. Available: http://personalpages.manchester.ac.uk/staff/robert.lucas/lucas%2et%2al%214.pdf [5] M. L. Amundadottir et al. (213, Apr). Modelling non-visual responses to light: unifying spectral and temporal characteristics in a single model structure. In Proc.Of CIE Centenary Conference Towards a New Century of Light.[Online]. pp. 11-11.Available: http://infoscience.epfl.ch/record/18674/files/amundadottir_op16_cie213_epfl.pdf [6] Cajochen et al. (2), Dose-response relationship for light intensity and ocular and electroencephalographic correlates of human alertness, Behavioural Brain Research, Boston, USA, 2, pp. 75-83. Available: http://www.chronobiology.ch/wp-content/uploads/publications/2_12.pdf [7] Chang et al. (211), The human circadian system adapts to prior photic history, J Physiol., vol. 589, pp. 195-112, 211. [8] Fostervold K.L. and Nersveen J.: Proportions of direct and indirect indoor lighting The effect on health, well-being and cognitive performance of office workers. Lighting Research and Technology, Vol. 4, pp. 175 2, 28. [9] Molin et al. (1996). The influence of climate on development of winter depression.[online].vol.37. pp.151-155. [1] Linke et al. (214), Simulating of LED sum-spectra for the best color rendering index along the black body curve a reverse engineering attempt", in the Proceedings of CIE 214 Lighting Quality & Energy Efficiency. [11] Thorseth, A., Optimization of light quality from color mixing light-emitting diode systems for general lighting. 212: p. 82781O-82781O. 448