Changing the direction of the luminaire: A strategy to improve lighting energy efficiency in offices

. Light environment’s non - visual effects influence people’s health and work efficiency. However, considering non-visual requirements in addition to traditional visual requirements may significantly increase lighting energy consumption. This study utilized simulation software to explore energy saving potential of changing the direction of the luminaire. A model of a single-person office with the luminaire attached to the ceiling right above the workstation was built in ECOTECT. Vertical eye-level illuminance and horizontal work-plane illuminance were calculated with luminaires of different luminous fluxes and elevation angles from downward vertical (0°-180° at an interval of 10° on both sides) using RADIANCE. Furthermore, six cases of different lighting requirements and light correlated color temperatures were considered. Based on the illuminance-versus-luminous flux coefficients obtained from simulation results, luminous fluxes were calculated to fulfill both visual and non-visual requirements under different elevation angles in all cases. It was found that compared to traditional lighting design with the luminaire facing vertically downwards, turning the luminaire at an elevation angle of 50° reduced the required luminous flux by up to 22.7%, which would benefit energy savings. Therefore, changing the direction of the luminaire has the potential to improve office lighting energy efficiency when considering both visual and non-visual requirements.


Introduction
Since the discovery of intrinsically photosensitive retinal ganglion cells (ipRGCs) as the third type of photoreceptor cells on the retina which maintain the circadian clock of mammals for daily synchronization [1], the non-visual effects of the light environment have gained increasing attention. Studies have found that light impacts sleep quality [2], both subjective [3] and objective alertness [4], mood [5], and task performance [6].
As people spend large amounts of time working in offices nowadays, the light environment in offices, with its non-visual effects on the well-being and performance of workers, is of great importance. Besides traditional visual requirements, non-visual requirements for the indoor light environment have been put forward. The WELL Building Standard states the equivalent melanopic lux (EML), which is a non-visual conversion from visual illuminance developed by Lucas et al. [7], for non-visual evaluations. Considering work areas, a 200 EML threshold is proposed on the vertical plane at eye level of the occupant for at least the hours between 9:00 and 13:00 at 75% or more of workstations [8].
A large proportion of building energy consumption comes from lighting. Therefore, strategies should be explored to reduce the lighting energy usage. Studies have been focused on lighting control strategies of dimmable lighting systems [9][10][11]. Nevertheless, * Corresponding author: linbr@tsinghua.edu.cn previous studies only included visual demands on the horizontal work plane when considering lighting qualities. The non-visual requirements with the compliance values measured at the eye level vertically, which differ greatly from the visual requirements, were not considered. It has been found that even with the use of dimmable lighting systems, the additional non-visual requirements of the EML still significantly increase lighting energy consumptions comparing with the traditional consideration of the sole visual requirements, especially when the daylight is insufficient and illumination predominantly depends on electric light [12].
Therefore, new strategies are needed, especially for spaces mainly depending on electric light for illumination, to reduce lighting energy consumption when both visual and non-visual requirements are considered. A strategy of changing the direction of the luminaire was proposed in this study. As the lighting energy usage is proportional to the luminous flux outputs of the luminaire [13], the relationship between illuminance (at vertical eye-level and at horizontal work plane) and luminous flux with luminaires of different elevation angles from the downward vertical was obtained by lighting simulation. Then, several typical cases of lighting requirements (both visual and nonvisual) and lighting conditions with luminaires of different correlated color temperatures (CCTs) were considered, and the required luminous fluxes under different elevation angles in these cases were calculated. The results were compared to find the optimal angle in each case with the least luminous flux required. Finally, the patterns of the optimal angles were deduced and the corresponding luminous flux reduction compared to traditional lighting design (with the luminaire facing vertically downwards) was calculated, which validated the strategy's potential for energy saving and brought new possibilities to lighting design.

Methods
An overview of the methods is shown in Fig. 1. An office lighting model was built. The vertical eyelevel illuminance and horizontal work-plane illuminance were calculated with luminaires of different luminous fluxes and different elevation angles using simulation software. It can be deduced that the average lighting illuminance on a particular plane is linear with the luminous flux of the luminaire. Then the linear regression coefficients of vertical eye-level illuminance versus luminous flux and horizontal work-plane illuminance versus luminous flux under different elevation angles were obtained. Six typical cases of lighting requirements and lighting conditions were considered. The required luminous fluxes under different elevation angles in these cases were finally calculated. The calculation results were compared to explore the optimal angles with which the least amount of luminous flux was need and the corresponding luminous flux reduction was obtained. Detailed explanations of the methods are presented below.

Characteristics of the office lighting model
A model of a single-person office measuring 4.0 m wide 3 m deep 2.8 m high was built in ECOTECT. The workstation was located at the center of the office and the work plane was 0.75 m above the floor. A chair with a worker sitting on it was placed beside the workstation. A "FlouroRecessedDroppedDiffuser" luminaire, which represented the common type of luminaire implemented in office spaces, was placed at the center of the ceiling right above the workstation. Surface reflectance settings are shown in Table. 1. The reflectance values of room surfaces and furniture are within the recommended range of the China's Standard for lighting design of buildings (GB 50034-2013) [14] and European Standard for lighting of work places (EN 12464-1:2021) [15]. Measurement grids were set up on the work plane (1.3 m 0.6 m with 0.05 m spacing between measurement points) and at the eye-level of the worker (0.15 m 0.06 m with 0.01 m spacing between measurement points) (Fig. 2). The average illuminance of the measurement grids was calculated by the well-proven Radiance software.

Simulation settings of various luminous fluxes and elevation angles
Luminaires of different luminous fluxes and different elevation angles from the downward vertical were adopted in the simulation. A preliminary simulation was conducted to choose the appropriate luminous flux values. The result showed that under the traditional lighting design in which the luminaire faces vertically downwards, the 1800 lm, 3000 lm, and 4300 lm settings enabled the horizontal work-plane illuminance to exceed 300 lx, 500 lx, and 750lx respectively. 300 lx is the required illuminance on the horizontal work plane for a general office in the standard GB 50034-2013 and 500 lx is the requirement for a high-grade office [14]. 750 lx is a higher illuminance value but within the restricted band in the standard EN 12464-1:2021 [15]. Based on the preliminary simulation, the luminous fluxes were set to 1800 lm, 3000 lm, and 4300 lm, which respectively represented the low, middle, and high level of light output of the luminaire (Fig. 1). For the elevation angles of the luminaire from the downward vertical, 0°-180°at an interval of 10°on both sides were all considered to ensure a thorough study (Fig. 3).

Calculation of the required luminous fluxes
The linearly fitted relationship between the average lighting illuminance on a particular plane and the luminous flux of the luminaire can be deduced. The luminous flux arriving at the plane can be divided into two portions: (i) direct luminous flux, and (ii) indirect luminous flux, which is contributed by the light that goes through at least one reflection in the room before reaching the plane. The luminous flux arriving at the plane can be described by Eq. (1), where Φ and Φ' are the initial luminous flux of the luminaire and the luminous flux arriving at the plane, kd is the proportion of direct luminous flux from the luminaire, and ki is the proportion of indirect luminous flux from the luminaire. The average lighting illuminance E can be described by Eq. (2), where A is the plane area.
For the same type of luminaires, kd is determined by the positional relationship between the luminaire and the plane. ki is determined by reflection-related parameters including the reflectance values of room surfaces and elements, position and angle of incidence in each reflection, and the number of reflections. A is a constant for a particular plane. Therefore, the relationship between E and Φ is linearly fitted with the coefficient of (kd + ki) / A. As the calculated vertical eye-level illuminance and horizontal work-plane illuminance in this study are actually the average illuminance on the reference plane of measurement grids, the linear regression coefficients of vertical eye-level illuminance  versus luminous flux and horizontal work-plane illuminance versus luminous flux under different elevation angles can be obtained.
Then a workflow is proposed to calculate the required luminous fluxes to fulfill both visual and nonvisual requirements under different lighting conditions with luminaires of different CCTs (Fig. 1). Visual requirements are set using the horizontal work-plane illuminance, while non-visual compliance can be evaluated using the EML. EML is a non-visual conversion developed by Lucas et al. [7]. It is the product of the vertical eye-level illuminance and a conversion factor which is determined by the spectral power distribution (SPD) of the luminaire in this research. Considering the linearly fitted relationship mentioned above, Eq. (3) and Eq. (4) can be derived as follows: (4) where Eh and EML are the horizontal work-plane illuminance in visual requirements and EML value in non-visual requirements, kh and kv are the linear regression coefficients of horizontal work-plane illuminance versus luminous flux and vertical eye-level illuminance versus luminous flux, Φh and Φv are the required luminous fluxes to fulfill visual and non-visual requirements, and R is the non-visual conversion factor. Considering that Eh and EML are set according to the requirements in different cases, kh and kv under different elevation angles can be obtained from the simulation results, and R is derived based on the CCTs of the luminaires, Φh and Φv can be deduced. Then the larger value of Φh and Φv is selected as the final luminous flux required.

Cases of lighting requirements and lighting conditions
Six typical cases of different lighting requirements and lighting conditions with luminaires of different CCTs ( Fig. 1) were considered. The 300 lx and 500 lx horizontal work-plane illuminance requirements were set with reference to the standard GB 50034-2013 in which a 300 lx threshold for a general office and a 500 lx threshold for a high-grade office were prescribed [14]. The 750 lx requirement was proposed as a higher level case but within the restricted band in the standard EN 12464-1:2021 [15]. The 200 EML requirement was set according to the WELL Building Standard [8]. As for the CCTs of the luminaire, 4000 K and 5500 K were chosen to represent lighting conditions with different non-visual conversion factors (R). This study used the conversion factors measured by Y. Zeng et al. with 0.69 for the 4000K luminaire and a higher value of 0.99 for the 5500 K luminaire due to a larger proportion of shortwavelength components [12]. Then the required luminous fluxes under different elevation angles in six cases were calculated adopting the workflow proposed in section 2.3.  Under the same luminous flux, the horizontal workplane illuminance is always higher than the vertical eye-E3S Web of Conferences 396, 01108 (2023) https://doi.org/10.1051/e3sconf/202339601108 IAQVEC2023 level illuminance regardless of the elevation angles, which indicates that the light provided by the luminaire is more efficient for the horizontal work-plane illuminance than the vertical eye-level illuminance. Furthermore, the changing trends of illuminance values are different, with the maximum value occurring at 0°f or the horizontal work-plane illuminance and at 50° for the vertical eye-level illuminance under all of the luminous flux settings.

Required luminous fluxes under different elevation angles
Taking  Adopting the workflow proposed in section 2.3, the required luminous fluxes under different elevation angles in six cases were calculated (Fig. 7). The optimal angle, with which the least amount of luminous flux is needed, can be derived from the calculations in different cases. The optimal angles differ with cases, with 50°for the cases of 200 EML + 300 lx + 4000K, 200 EML + 300 lx +5500 K, and 200 EML + 500 lx +4000 K, 40°f or the cases of 200 EML + 500 lx + 5500 K and 200 EML + 750 lx + 4000 K, and 10°for the case of 200 EML + 750 lx + 5500 K. Considering the 200 EML + 300 lx + 4000K case as an example, the required luminous flux is 6334 lm under the traditional lighting design in which the luminaire faces vertically downwards (0°), but it decreases to 4900 lm under the optimal angle, which brings a 1434 lm (22.7%) of luminous flux saving. Therefore, compared to the traditional lighting design, changing the direction of the luminaire will bring luminous flux-saving benefits when considering both visual and non-visual requirements.

The optimal elevation angles under different requirements and CCTs
It has been mentioned in section 3.2 that the optimal elevation angles vary with cases of different lighting requirements and CCTs of the luminaires. In fact, the optimal elevation angles under different cases follow some patterns.
The maximum value of kh and kv occur at 0°and 50°, indicating that 0°and 50°are the most efficient angles for visual and non-visual requirements respectivlely. As kh and kv both change continuously with the angles, the optimal angles will always fall between 0°and 50°, though differ with various lighting requirements and CCTs.
It can be easily proposed that whether the optimal angle falls closer to 0°or 50°relates to which one of the visual and non-visual requirements (Eh and EML) is dominant, as the higher luminous flux value is selected. The required luminous fluxes to fulfill visual and nonvisual requirements (Φh and Φv) are described in Eq. (5) and Eq. (6), which are derived from Eq. (3) and Eq. (4). (5) Φv = EML / (kv × R) (6) When Φv

Φh = Eh / kh
Φh , which can also be expressed as kv / kh EML / (Eh × R) based on derivation from Eq. (5) and Eq. (6), the non-visual requirement takes the dominant role. When kv / kh EML / (Eh × R), the visual requirement is dominant. Fig. 8 illustrates Φh (fulfilling the visual requirements of 750 lx, 500 lx, and 300 lx) and Φv (fulfilling the non-visual requirement of 200EML with luminaires of 4000 K and 5500 K) under elevation angles of 0°-50°. In most cases, Φv is larger than Φh, which can be attributed to the fact that kh is always higher than kv. However, Φh exceeds Φv sometimes. Taking the case of 200 EML + 750 lx +5500 K as an example, under the elevation angle of 50°, Φh (5374 lm) is larger than Φv (3415 lm). This is due to the fact that the values of Eh and R are relatively higher compared to other cases, and the distance between kh and kv is relatively small compared to other angles between 0°-50° (Fig. 6) which results in a higher kv / kh value.
The positons on which the optimal elevation angles fall in different cases can be deduced. The fitting curves of Φh and Φv are shown in Fig. 8. Φh increases monotonically between 0°and 50°, while Φv decreases monotonically. Therefore, the optimal angles occur where the distance between Φh and Φv reaches the minimum value. The following is an explanation in detail. If the fitting curves of Φh and Φv intersect at a point between 0°and 50°(200 EML + 500 lx + 5500 K, 200 EML + 750 lx + 4000 K, 200 EML + 750 lx + 5500 K, and the intersection points are marked with the cross shapes), the first angle beside that point is the optimal angle. Whether it's the angle at the left or right side of the intersection point depends on which one results in a lower required luminous flux. An example is the case of 200 EML + 500 lx + 5500 K. The fitting curves intersect at a point between 40°and 50°. The Φv point scatters higher than the Φh point at 40°and the Φh point scatters higher than the Φv point at 50°. The two higher points are selected and compared, with the higher point of 40°s cattering lower, which mains 40°enables a lower luminous flux (3466 lm) compared to 50°(3583 lm). Therefore, the optimal angle falls on 40°. If the fitting curves don't intersect between 0°and 50°(200 EML + 300 lx + 4000 K, 200 EML + 300 lx + 5500 K, 200 EML + 500 lx + 4000 K), the non-visual requirement totally dominants the case and the optimal angle falls on 50°. The star shapes in Fig. 8 mark the optimal points where the optimal angles are reached and the required luminous fluxes are the lowest in the studied cases. Then the patterns that the optimal angles follow can be described and explained, which are also shown in Fig.  8. In half of the six cases the optimal angle falls on 50°d ue to the difficulty of fulfilling non-visual requirements. When the CCTs of the luminaires remain the same, the increase in visual requirements pushes the optimal angle to move closer to 0°by pushing the intersection points towards 0°. In addition, the increase in CCTs moves the optimal angle towards 0°under the same lighting requirements also by pushing the intersection points. It can be concluded that the harder fulfilling the non-visual requirement is compared with fulfilling the visual requirement, the closer to 50°the optimal angle falls on.

Luminous flux reduction compared to the traditional lighting design
It is known that the lighting energy usage is proportional to the luminous flux outputs of the luminaire, no matter the proportionality is linear or nonlinear [13]. The reduction in luminous flux brought by changing the direction of the luminaire benefits lighting energy savings. Fig. 9 shows the luminous flux reduction brought by the optimal angles compared with the traditional lighting design (with the elevation angle of 0°) under different cases. The reduction rate varies with cases, with the maximum reduction rate of 22.7% occurring in the cases of 200EML+300lx+4000K, 200EML+300lx+5500K, and 200EML+500lx+4000K. In contrast, the minimum reduction rate, which occurs in the case of 200 EML + 750 lx + 5500 K, is only 3.2%.
The following provides an explanation for the difference in reduction rates. As is shown in Fig. 8, the non-visual requirement dominates the required luminous fluxes when the elevation angle is 0°. Φv decreases monotonically between 0°and 50°. Based on the deduction of the optimal angles mentioned in section 4.1, it can be found that the closer the intersection point of the Φh and Φv curves is to 50°, the less luminous flux is required compared with that under 0°. Further considering the 'lower one of the higher points' method proposed in section 4.1 in choosing the angle on the left or right side, it can be finally concluded that the optimal angle's greater distance from 0°leads to a higher luminous flux reduction rate. This can be demonstrate in Fig. 9 with the optimal angles and corresponding luminous flux reduction rates marked on it, which means that the strategy of changing the direction of the luminaire is more meaningful in cases when the nonvisual requirement outweighs the visual requirement in the determination of energy usage to a large extent.
Nevertheless, the higher luminous flux reduction rate doesn't always mean the lower luminous flux value, as the reduction rate only reflects the luminous flux saving potential of changing the direction of the luminaire. Other factors such as the CCTs of the luminaire are not considered. Taking the cases of 200 EML + 750 lx + 4000 K and 200 EML + 750 lx + 5500K as an example, the reduction rate in the latter case (3.2%) is lower than in the former case (21.5%), but as for the values of the minimum required luminous flux, the latter (4275 lm) is also lower than the former (4974 lm) (Fig.  9). This can be attributed to the higher non-visual conversion factor in the lighting condition with higher CCT, which significantly reduces Φv. In fact, comparing the cases of the same lighting requirements and different CCTs considered in this study, it can be found that under the traditional lighting design (0°) and the same lighting requirements, simply increasing the CCT of the luminaire from 4000 K to 5500 K brings a 30.3% reduction in the required luminous flux. The reduction rate brought by increasing the CCT is higher than the maximum reduction rate brought by changing the direction of the luminaire (22.7%). Therefore, increasing the CCT of the luminaire should be considered in prioriry from the prespective of energy savings by reducing luminous fluxes.
But still, changing the direction of the luminaire can be regarded as an effective strategy to improve lighting energy efficiency in offices, as it significantly reduces the increase rates of energy usage brought by adding non-visual requirements based on visual requirements. Considering its application over a typical working day, as the 200 EML threshold at the eye level of the workers is proposed between 9 :00 and 13 :00 in the WELL Building Standard [8], the adjustment of the elevation angle can be conducted in relation to time of day. The working hours are 9 :00-17 :00, so the optimal elevation angle should be applied during 9 :00-13 :00 to fulfill both visual and non-visual requirements with the least amount of luminous flux, but the traditional lighting design can be applied during 13 :00-17 :00 as it's the most efficient way to light up the horizontal work plane and fulfill the visual requirement when the non-visual requirement is removed. The following is an example of the 200 EML + 300 lx + 4000 K case. Considering here that the electric power is linear with the luminous flux [12], it can be calculated that under the traditional lighting design with a fixed elevation angle of 0°, the additional 200 EML non-visual requirement during 9 :00-13 :00 increases the required energy usage of the whole working day by 136.3% compared with considering the sole visual requirement of 300 lx during the whole working hours. However, by adopting the the optimal elevation angle during 9 :00-13 :00, the increase rate decreases to 94.1%. Therefore, the strategy of changing the direction of the luminaire can be applied in architectural practice to save energy.

Limitations and expectations
This paper proposes a new strategy of changing the direction of the luminaire to improve lighting energy efficiency while fulfilling both visual and non-visual requirements. However, the study was limitted to a typical single-person office with the luminaire placed at the center of the ceiling. Offices with multiple workstations [12] or open-plan offices which are popular in today's workplace design [16] were not considered in the study. In these office settings, the positions of the luminaires are different from those in single-person offices. It can also be expected that changes in the direction of one luminaire will impact the vertical eye-level illuminance for staff at more than one workstations. Therefore, future research is needed to explore the strategy's application in offices with various layouts.
In addition, the measurement grids at the eye-level of the worker were set up vertically in this study, which only simulated the worker's view direction facing forward. Although gazing forward is common in today's office work as staff spend a lot of time on computerbased tasks which require them to look forward at the screen, the vertical measurement grids cannot cover all the activities in the office. For example, reading the text on the paper is still needed and this requires the worker to look down at the horizontal work plane. Furthermore, the visual focus of the worker is switched between the keyboard and screen when typing, but the fixed vertical grids cannot represent the changes of the view direction. The eye-level illuminance should be considered with workers' specific activities and movements in future research.

Conclusions
Recently, the non-visual effects of the light environment have gained increasing attention. However, the consideration of non-visual requirements in addition to the traditional visual requirements may significantly increase lighting energy consumption. This study explored the potential of reducing energy consumption by changing the direction of the luminaire. The vertical eye-level illuminance and horizontal work-plane illuminance were simulated to obtain the relationships of vertical eye-level illuminance versus luminaire's luminous flux and horizontal work-plane illuminance versus luminaire's luminous flux under different elevation angles. The required luminous fluxes under different elevation angles in six cases considering three types of lighting requirements and two types of lighting conditions were calculated. After comparing the required luminous fluxes under different elevation angles in these cases quantatively, the conclusions were summarized as follows : (1) The light provided by the luminaire is more efficient for the horizontal work-plane illuminance than the vertical eye-level illuminance. The linear regression coefficients of horizontal work-plane illuminance versus luminous flux is always higher than the coefficients of vertical eye-level illuminance versus luminous flux under different elevation angles. Furthermore, 0°is the most efficient angle for horizontal work plane illuminance, but for vertical eye-level illuminance, 50°is the most efficient one.
(2) The optimal angle, with which the least amount of luminous flux is needed to fulfill both visual and nonvisual requirements, falls beween 0°and 50°. Taking a step further, it falls where the distance between the required luminous fluxes to fulfill visual and non-visual requirements reaches the minimum value. The harder fulfilling the non-visual requirement is compared with fulfilling the visual requirement, the closer to 50°the optimal angle falls on.
(3) Changing the direction of the luminaire reduces the luminous flux required to fulfill both visual and nonvisual requirements, with the reduction rate reaching 22.7% in the cases of 200 EML+300 lx+4000 K, 200 EML+300 lx+5500 K, and 200 EML+500 lx+4000 K, thus reducing the lighting energy usage. In addition, the strategy of changing the direction of the luminaire is more meaningful in cases when the non-visual requirement outweighs the visual requirement in the determination of energy usage to a large extent.
The most efficient elevation angles of the luminaire for horizontal work plane illumiance and vertical eye-level illuminance are different. The strategy of changing the direction of the luminaire, with its potential to improve lighting energy efficiency while fulfilling both visual and non-visual requirements, can be applied in architectural practice.