Influence of Opposing Exterior Window Geometry on the Carbon Emissions of Indoor Lighting under the Combined Effect of Natural Lighting and Artificial Lighting in the City of Shenyang, China

Abstract

According to relevant statistics, the electricity consumption for lighting in university buildings accounts for 20 to 40% of the total energy consumption of the buildings. Lighting energy saving is a key influential factor in achieving a low-carbon campus construction. The electricity consumption for lighting is simultaneously affected by the utilization of natural daylight and artificial lighting schemes. Currently, there is a lack of research regarding the dynamic quantitative correlation between the geometric design of external windows affecting the utilization of natural daylight and carbon emissions. Also, research on the dynamic synergistic impact between natural light utilization and artificial lighting on carbon emissions has not been observed. Hence, there is a lack of quantitative carbon impact prediction and guidance in the early design and actual operation of such spaces. This study took the professional drawing space of a university in the severe cold regions of Shenyang as a prototype. Daylight factor (DF) and spatial daylight autonomy (sDA) were determined using Rhino + Grasshopper and Ladybug + Honeybee for window geometry. DIALux evo simulation was used to analyze the carbon emissions of space operation, followed by correlation analysis and multiple linear regression analysis using SPSS to determine the degree of influence of each window design parameter on the carbon emissions. The window-to-floor ratio (WFR), window-to-wall ratio (WWR), windowsill height (H_ws), window width (W_w), and window height (H_w) had inhibitory effects on carbon emissions from daylight-responsive artificial lighting (C), and the influence of different orientations was different. Under the condition of an opposing window, the overall C trend of the professional drawing space was west < east< south < north, and the C of the morning period in each orientation was significantly lower than that in the afternoon period. Taking the frame structure system space with a floor-to-floor height of 4.2 m as an example, within the requirements of WFR and WWR, the C of the west-facing professional drawing classroom with 2.55 m for H_w, 0.75 m for H_ws, and 9.6 m for W_w was the smallest. To a certain extent, opening large windows and opening high windows can reduce the C of the space.

1. Introduction

Indoor lighting environments are composed of natural and artificial light. The key to achieving a good indoor lighting environment is to make full and balanced use of both. Natural light has widely known advantages, such as sustainability, high uniformity, and having a superior color rendering index. Introducing an appropriate amount of natural light can provide a better environment for building users, generate positive psychological effects, and enhance work efficiency [1]. Natural lighting can save on electricity in buildings and can achieve the dual goals of a high-quality lighting environment and a reduction in carbon emissions. Much research has been conducted on exterior windows and daylighting design. There are also numerous studies on the relationship between energy consumption and lighting environment quality associated with daylight and artificial light.

Currently, scholars have conducted extensive research on the impact of various factors, including different geographic regions and building types, on daylighting performance with respect to window geometric shapes. Susorova et al. [2] conducted research on the impact of window size, window-to-wall ratio, and room geometry on daylighting performance in office buildings across different climate zones, including cold, temperate, and hot climates. The aim was to optimize daylighting quality and uncover energy-saving potential. Zhao and Mei [3] evaluated the impact of 22 different design factors on the indoor daylight environment of a stadium in the mid-latitude region of China. The study identified date, latitude, window position, glass transmittance, building height, building depth, and window opening area as the most significant factors. The study proposed a simplified formula through linear regression analysis, assuming the stadium’s illuminance and illumination uniformity meet the standard requirement. The study enabled architects to calculate the required window area easily in the preliminary design stage. Katunský et al. [4] proposed an approximate method for calculating the daylight factor through a case study of two production halls in a textile factory in the eastern part of Slovakia. The results can be applied to similar production halls illuminated by natural light through windows located at the vertical positions on both sides.

With the development of low-carbon campus construction, the introduction of natural light in university buildings can achieve improved environmental quality and energy-saving and carbon reduction effects, while also promoting the physical and mental well-being of both faculty and students [5,6]. Rubeis et al. [7] changed the geometry and window-to-floor ratio and applied different self-adjusting lighting systems for an academic classroom with only one window. By evaluating the daylight factor (DF) and daylight autonomy (DA), they found that the geometry of the room and window had a significant impact on daylighting. Katunský et al. [8] renovated the attic classroom space of an old building by optimizing window shading, ensuring sufficient daylight, and harnessing energy-saving potential. Galal [9] studied the effects of building orientation on thermal and lighting performance using classrooms in Lebanon as an example. The study found that classrooms in mountainous and coastal areas have similar daylight performance when the long elevation of the classroom faces south. Atthaillah et al. [10] built a typical simulation model with double-sided (opposing) windows without shading devices. They applied the Honeybee plug-in via Rhino and Grasshopper to simulate the annual daylight environment. The study found that the design of windows on different elevations of a school building should not be the same to achieve better daylight performance. Lakhdari et al. [11] used a parametric approach to calculate how much the building envelope and building orientation could affect daylighting environment, thermal comfort, and energy efficiency, using a case study of a secondary school classroom in a hot and dry climate. The building envelope involved the window-to-wall ratio, wall materials, glazing types, and shading devices. The study also proposed a solution to balance daylight use and thermal comfort.

In recent years, scholars have embarked on studying the mechanisms of natural light and artificial lighting regulation, as well as the energy-saving potential of artificial lighting. This includes research into lighting energy consumption and overall building energy consumption. Danny et al. [12] conducted an analysis of electricity bills and indoor illuminance for workshops and classrooms with high-frequency dimming control. They used a simple predictable method to demonstrate that energy-efficient lamps with dimming control and proper lighting design can reduce electricity use and complete green building design. Yu et al. [13] used the lighting simulation software RELUX to quantitatively analyze the annual daylight energy-saving potential. They found that the geometry of the room, window-to-wall ratio, window transmittance, surface reflectance of the building and surrounding buildings, artificial lighting layout, and daylight control strategies all have a significant impact on daylight environment and total energy saving. Din and Kim [14] proposed a lighting design optimization and controlling system. They studied the optimal angle of the blinds according to the sun’s altitude and window orientation. This proposal can reduce energy consumption via maximizing daylight utilization without glare. Shishegar and Boubekri [15] investigated the impact of various daylight control devices on building energy consumption and lighting energy use in office buildings in hot climates. They tested open-plan office buildings with different orientations and window-to-wall ratios. It was concluded that daylight control devices in office buildings could significantly reduce lighting energy consumption. Pellegrino et al. [16] simulated office building energy use in Italy with different room depths, window-to-wall ratios, exterior shading devices, and window orientations using Daysim and EnergyPlus. They found that optimizing daylight design can reduce the total energy use of office buildings. Li et al. [17,18,19] conducted lighting evaluations, as well as lighting performance and energy-saving simulations, of office buildings in Hong Kong. Further research has also been conducted on energy saving using lighting control systems [20,21].

The literature review revealed a research gap, in that the direct correlation between the geometry of exterior windows and carbon emissions is still unknown. There are two reasons for this: firstly, a real-time dynamic model that combines natural light and artificial lighting has not been established. Secondly, campus building design is only partway through the process of achieving carbon neutrality. Optimizing natural lighting design in buildings, as well as establishing a dynamic model that can adjust and balance natural and artificial lighting, are crucial to achieving high-quality light environments. This can also achieve a better artificial lighting design scheme to reduce light energy consumption and thus reduce carbon emissions.

To achieve “carbon neutrality and carbon peak”, the current research focused on the coupling of natural and artificial lighting to explore the energy-saving potential of daylight. The present research investigated the relationship between the window geometry, artificial lighting layout, and lighting operation schedule. The present research studied professional drawing classrooms with windows in different orientations in severe cold regions. The research used the Ladybug and Honeybee plug-ins installed on Rhinoceros and Grasshopper to simulate daylight environments [22]. DIALux evo was applied for quantitative simulation of building energy consumption, lighting costs, and carbon emissions. The buildings with daylight control systems were input into the DIALux evo model. DIALux evo could dynamically correlate natural light utilization and artificial lighting. It provided quantitative simulation for building energy consumption, cost, and carbon emissions. It also calculated lighting illuminance. DIALux evo had higher adaptability, reliability, and accuracy, especially for large spaces. The simulation results were obtained through sampling and comparing lighting layout schemes. Correlation analysis, box-plot analysis, and multiple linear regression analysis were conducted to analyze window geometry (window height, windowsill height, window width, wall width between windows, window-to-floor ratio, window-to-wall ratio) and carbon emissions caused by artificial lighting in professional drawing classrooms during the daytime. The present paper established a simulation model to predict the impact of window geometry on carbon emission, taking a case study of professional drawing classrooms with windows in different orientations.

The research assumed that the lighting environmental quality met the standard. The simulation model provided quantitative guidance for reducing carbon emission in the early design stage, as well as operations for professional drawing classrooms in universities in severe cold regions.

2. Methods

2.1. Workflows

This study aims to examine the effect of opposing exterior window geometry on the carbon emissions of indoor lighting in university professional drawing classrooms. After proposing a hypothesis and setting goals, the sample space of professional drawing classrooms in universities was analyzed and the experimental prototype space was selected. Simulation modeling parameters were set based on the actual space measurement, and simulation analysis data were collected and organized. A predictive model was established using multiple linear regression. Finally, discussions and conclusions were drawn. The experiments were conducted step by step, as illustrated in the flowchart (Figure 1).

2.2. Location and Climate

Shenyang is located between latitudes 41°11′51″ and 43°2′13″ N, and longitudes 122°25′9″ and 123°48′24″ E [23]. It is one of the capital cities in China, located in the southern part of Northeast China and central Liaoning Province, in the warm temperate sub-humid region [24] (Figure 2a). According to GB 50176-2016 [25] and JGJ 26-2018 [26], Shenyang belongs to the severe cold zone C of China. Therefore, the shape and spatial layout of buildings in Shenyang are usually regular geometric forms, and the body shape coefficient of the building is small, which is conducive to the target needs of winter insulation design of buildings. At the same time, according to GB50189-2015 [27] and GB50033-2013 [28], the building window area is subject to the dual requirements of the upper limit of “window-to-wall ratio” and the lower limit of “window-to-floor ratio”, so as to achieve the dual goals of building energy conservation and uniform lighting in indoor space.

According to GB50033-2013 [28], Shenyang belongs to a Class III daylight climate zone, with an average total illumination of 35~40 klx (Figure 2a). According to IWEC Shenyang typical meteorological year data, the monthly average hourly global horizontal radiation was as low as 6.5 klx in December and 23.8 klx in June (Figure 2b). The lowest monthly average hourly total horizontal radiation was in December, which was 44.3 kWh/m²; The highest, in May, was 159.6 kWh/m² (Figure 2c). Shenyang, like other areas with thermal design needs, bases its buildings’ indoor lighting environments on the moderate use of natural lighting, which is combined with artificial lighting and affects lighting energy consumption and CO₂ emissions through artificial lighting operation. Space type, indoor natural lighting utilization, artificial lighting layout, and usage behavior are the four key factors affecting the quality of an indoor lighting environment.

2.3. Evaluation Criteria

In this research, we intend to establish the influence of the coupling of exterior window geometry and artificial lighting on carbon in the professional drawing teaching spaces of colleges and universities and ensure that the optical performance of the working interface of the three types of subspaces (drawing area, discussion area, and aisle area) of the space is always satisfied. In the case of simply meeting the standard values for indoor daylight illuminance, the comfort of the lighting environment is assessed through daylighting performance evaluation indicators for data selection. Daylight factor (DF) is the most widely used static daylighting evaluation indicator [28]. The dynamic daylighting evaluation indicators, such as daylight autonomy (DA) and spatial daylight autonomy (sDA), further reflect the impact of regional lighting climate on the lighting environment and the proportion of the area achieving a given illuminance level within the space [29,30]. Therefore, we selected DF and sDA as the evaluation indicators for daylighting performance.

The standard value of interior daylight illuminance is the illuminance value on the reference surface corresponding to the specified outdoor natural light design illuminance value and the corresponding daylight factor standard value. According to GB50033-2013 [28], DF = (E_n/E_w) × 100% (E_n: indoor illumination, E_w: outdoor illumination), the standard value of interior daylight illuminance in the drawing area and discussion area is 600 lx, and the standard value of interior daylight illuminance in the aisle area is 150 lx.

Daylight factor (DF) is defined as the ratio of the illumination generated by the direct or indirect reception of sky-diffused light from the assumed and known brightness distribution of the sky at a point in the indoor reference surface to the sky diffuse illumination produced by the sky hemisphere on the outdoor unobstructed horizontal plane at the same time [28]. According to the requirements of GB50033-2013 [28] and GB/T50378-2019 [31], the height of the exterior windowsill is 0.75 m (≥0.75 m), the design illuminance of exterior daylight of the class III daylight climate zone is 15,000 lx, the reference surface of the working area is taken from the horizontal plane of 0.75 m from the ground, the reference surface of the aisle area is taken from the ground, the standard value of daylight factor of the drawing area and the discussion area is set to 4.0%, and the standard value of daylight factor of the aisle area is set at 1.0%. The minimum daylight factor satisfaction rate ≥60% can be regarded as excellent.

Daylight autonomy (DA) is defined as the percentage of time during the year-round working time that the work surface can reach the specified minimum illumination requirement by natural lighting alone [32]. In this study, DA, as an intermediate indicator to obtain sDA, does not participate in the process of screening the model, and its value can be used as a reference variable for subsequent analysis.

Spatial daylight autonomy (sDA) is based on DA to analyze the proportion of each grid point in the plane area to meet a certain time ratio of illuminance. According to IES LM-83-12, when sDA_{500lx, 50%} is greater than or equal to 75%, the area can be considered to have a high daylight adequacy [33].

2.4. Sample Analysis and Selection

2.4.1. Sample Analysis

Taking the drawing classroom space of architecture majors in Chinese universities as an example, we investigated the drawing classroom space of architecture majors in a university in Shenyang and the top eight universities in China, and its spatial forms were mainly divided into two categories, namely: general small classrooms based on class and open large-space classrooms based on grade [34] (

Table 1. Spatial typology of professional classrooms in the top eight ranked universities for architecture discipline in China.

Serial Number	Name of Teaching Building	Classroom Type	Space Geometry
1	Drawing Room, Department of Architecture, Tsinghua University, Beijing, 1911	Undergraduates—Large Space Classroom Graduate Students—Studio
2	Drawing Room, Department of Architecture, Tongji University, Shanghai, 1907	Undergraduates—Large Space Classroom
3	Drawing Room, Department of Architecture, South China University of Technology, Guangzhou, 1952	Undergraduates—Large Space Classroom
4	Drawing Room, Department of Architecture, Xi’an University of Architecture and Technology, Xi’an, 1895	Undergraduates—Large Space Classroom Undergraduate Senior—Studio
5	Drawing Room, Department of Architecture, Southeast University, Nanjing, 1902	Undergraduates—Large Space Classroom
6	Drawing Room, Department of Architecture, Tianjin University, Tianjin, 1895	Undergraduates—Large Space Classroom
7	Drawing Room, Department of Architecture, Northeastern University, Shenyang, 1923	Undergraduates—Large Space Classroom
8	Drawing Room, Department of Architecture, Harbin Institute of Technology, Harbin, 1929	Undergraduates—General Small Classroom
9	Drawing Room, Department of Architecture, Chongqing University, Chongqing, 1920	Undergraduates—General Small Classroom Graduate Students—Tutor’s Studio

Notes: The sample space prototype selected in this study is based on the 7th space.

).

With the gradual promotion of open teaching and the exchange of professional knowledge between grades or classes in China, the two-way open large-space classroom has gradually become the mainstream form of architecture professional drawing classroom spaces, and the architecture professional drawing space of a university in Shenyang conforms to the characteristics and development trends of this type of space. Therefore, the sample space selected in this research is an open spacious space with opposing windows for architectural drawing, and the sample space prototype and parameter extraction were based on the drawing space of the architecture major of the university.

2.4.2. Sample Selection

The university in Shenyang utilized in this study is a national key university, and its architecture drawing classroom conforms to the spatial characteristics of an opposing window and open and large space, and the spatial characteristics and development trend of the sample space prototype. The sample space prototype selected in this research was based on the current state of space (Figure 3), and the sample space prototype was refined and parameter setting carried out by configuring the sample space size setting and the professional use mode.

First of all, from the perspective of the sample space prototype size setting, according to the provisions of the fourth volume of China’s latest published Architectural Design Data Collection (third edition), the physical space used for architectural professional teaching should be not less than 5.70 m²/person ratio requirements [35]. According to survey statistics, China’s colleges and universities recruit 90–100 undergraduate students in architecture every year (3 classes), and the space size of each grade of architecture professional teaching space should be between 513–570 square meters. The prototype refining space takes “three bays and two depths” as a teaching unit, with a space bay of 10 m, a depth of 9 m, an inner corridor width of 2.7 m, and a total area of 540 m², which meets the requirements of enrollment scale and teaching space configuration indicators. Therefore, the space can be used as the basic set size of our sample prototype space, namely: the prototype foundation consists of three bays and two depths; the bay and depth are 10 m and 9 m, respectively; the floor height is 4.20 m; and the rectangular structural column in the space is 0.6 m × 0.6 m.

Secondly, from the perspective of the use of the sample space prototype, combined with the basic dimensions of the above-mentioned three-bay, two-depth, open professional drawing space, according to its space use behavior, the teaching and use division of the space was implemented, namely the functional division of the middle public transportation area and the professional teaching and learning areas on both sides. The teaching and learning areas are separated on both sides along the direction of the space bay, and the external wall interfaces corresponding to their respective areas have the requirement to open the windows for natural light.

Therefore, the sample prototype space takes the form of an opposing exterior window. In the space studio, the drawing area, discussion area, and aisle area are refined as a unit to form a continuous “U” shaped area of “3 through 4 zones”, and two optimal window opening schemes: a whole window and three vertical windows corresponding to the U-shaped area of a “3 way 4 zone” (Figure 4).

3. Simulation Setup and Data Processing

3.1. Simulation Tools

Rhinoceros and Grasshopper are mainstream pieces of software used for parametric modeling programming and data design. Based on Rhinoceros and Grasshopper as the modeling and visualization platforms, Ladybug and Honeybee were used as the performance simulation plug-ins for daylighting simulation calculation, verification, and screening of models that met the standards [22]. Based on the above, DIALux evo analysis and simulation software was used to carry out experimental simulation output of space lighting energy consumption, lighting cost, and carbon emissions [36]. This series of simulation platforms, plug-ins, and analysis software are widely used in the research and analysis of building energy consumption, cost generation, and carbon emissions in light environments. A number of related studies have confirmed that this series of simulation tools has good accuracy, credibility, and scientific nature [37,38,39,40]. DIALux evo is a lighting analysis and prediction tool that can couple natural light utilization and artificial lighting operations. It has a higher adaptability and reliability when analyzing large spaces. Under the premise of ensuring the construction of a suitable lighting environment, the software can set targets based on the comprehensive lighting environment, and carry out quantitative simulation output of lighting loss, lighting costs, and carbon emissions. It can eliminate the unstable and uncertain risks caused by the secondary analysis and correlation processing of previous models and data. DIALux evo uses the sky without direct sunlight described in the CIE standard [41]. Ladybug is a performance plug-in for microclimate analysis based on Grasshopper for data analysis and visualization simulation output. The weather files (EPW) were imported from EnergyPlus [42]. IWEC data were used in this research. Honeybee connects building performance analysis software and parametric platforms such as Radiance, Daysim, and EnergyPlus to complete the automatic input of set parameter simulation analysis, and the analysis data conclusions are read out and visualized in Rhinoceros and Grasshopper. SPSS is one of the most widely used pieces of software for data entry, collation, and analysis in scientific research.

3.2. Simulation Modeling

The spatial perspective view (Figure 5) and the floor plan (Figure 6) of the space model are illustrated as follows. The space model contains modeling parameters in three dimensions.

First, we set the basic parameters of the sample space. The sample is a three-bay and two-depth open classroom for professional drawing. The space has a structural frame within which equidistant rectangular structural columns are placed along the bay orientation. The relevant parameters are as follows: length of 30.00 m (W_pc), width of 18.00 m (D_pc), and height of 4.20 m (H_pc) [35]; thereinto, the size of the rectangular structural columns is 0.60 m * 0.60 m. The distances between the depth and bay orientation of the structural column are 9.00 m and 10.00 m, respectively.

The arrangement of facilities inside the space refers to the survey sample. The inside of the sample space is divided into three areas: A, B, and C. Areas A and B are the plot area and discussion area, with depths of 9 m(b) and 6.3 m(d), respectively. Each bay contains three aisle areas in the depth direction. Area C is the public aisle area with a depth of 2.70 m(c). The drawing area is categorized into a one-sided table zone and a double-sided table zone. The table for drawing is 0.75 m in height and 1.1 m in width.

The window design of the sample space model determines the number and parameter settings of the opening windows. Exterior window open setting is deployed outside of the facades in the depth direction of the sample space. The forms of the exterior windows in each bay are horizontal full windows and vertical multi-belt windows. The horizon opening border of the exterior window is an absolute value of 0.3 m(f) in both the left and right directions; the height of the sill is larger than 0.75 m(H_ws); the height of the upper edge of the window is less than 3.3 m(e ≥ 0.9 m); the wall width between the windows is represented by W_wbw; and the window height is represented by H_w. The window width is evenly distributed according to the wall width between the windows. Meanwhile, walls between windows are designed so as to achieve direct correspondence between the open window area and the indoor drawing area, and therefore we made full use of the natural lighting in the drawing area. The meanings of symbols and signs are represented in

Table 2. The meanings of symbols and signs.

Symbols and Signs	Significance
Hw	window height
Hws	windowsill height
Wwbw	width of wall between windows
Ww	width of window
WWR	window-to-wall ratio
WFR	window-to-floor ratio
DA	daylight autonomy
sDA	spatial daylight autonomy
DF	daylight factor
C	carbon emissions with daylight-controlled artificial lighting

.

3.3. Lighting Layout

The lighting layout scheme is a targeted scheme for regression prediction in this research. This is the fixed scheme after space layout has been optimized. The current sample adopts the lighting layout scheme, in which lighting is designed to be perpendicular to the working interface. The lighting layout scheme after being optimized is the simulation scheme used in the sample, in which the lighting is perpendicular to and corresponds with the analysis interface (Figure 7). The specific layout scheme is illustrated as follows. According to JGJ 310-2013 [43], the lighting selected is wide-distribution LED lighting. According to GB 50099-2011 [44], the suspension height of the lighting from the desk should not be less than 1.70 m. Taking the generally designed shape of space floor height, floor thickness, and the height of beams into consideration, the height of the lighting above the desk is designed to be 2.5 m, which can simultaneously meet the needs of illuminance of analysis interface and glare avoidance. After being calculated, the room cavity ratio of the sample is 1.11. The optimized space lighting layout adopts the setting that the lighting is parallel to the working interface so as to build an optimal lighting arrangement scheme. The height-to-distance ratio of the lighting is controlled between 1.16–1.52, satisfying the requirement of the lighting arrangement in GB/T 31831-2015 [45].

In the meantime, the professional drawing southward classroom was the model chosen to carry out the comparative simulation experiment in before and after optimization. We set Ww to be 9.6 m, Hw to be the maximum, which is 2.55 m, and Hws to be 0.75 m. The pre-optimized and optimized lighting layout schemes were used for simulation analysis, and the daytime carbon emissions obtained were 1979 kg/a and 1855 kg/a, respectively. Therefore, the optimized lighting layout scheme had a relatively remarkable energy-saving potential compared to the former scheme.

3.4. Interface Parameters

According to the relative rules in GB 50033-2013 [28] and GB 50189-2015 [27], the internal interface parameters of the sample model adopted take on mean values specified by the rules, and the exterior windows take the minimum value. The specific settings are as follows: The reflectance of the ceiling was set to 0.75, the wall reflectance was set to 0.55, and the ground reflectance was set to 0.30; the interface reflectance of both the drawing area and discussion area was set to 0.40, and the exterior window transmittance was also 0.40 (

Table 3. Space interface setting parameters.

Surface Parameters	Specification Requirements	Model Setup
Roof Reflectance Ratio	0.60–0.90	0.75
Wall Reflectance Ratio	0.30–0.80	0.55
Floor Reflectance Ratio	0.10–0.50	0.30
Worktop Reflectance Ratio	0.20–0.60	0.40
External Window Transmission Ratio	≥0.40	0.40

).

3.5. Run Settings

In accordance with the survey on the operation of teaching space in universities in Shenyang, in terms of weekly and monthly operation, weekdays are the daily teaching-using period, therefore the average accumulative teaching days in a month amount to 22 days; in terms of daily operation, college teaching is divided into three periods of time in a day, which are morning (8:00~12:00), afternoon (14:00~18:00), and evening (19:00~22:00). Based on the operation cycle and periods above, the sample space adopts a dynamical and continuous coupling between natural lighting and artificial lighting in the daytime, while in the nighttime artificial lighting operates more individually. It forms a general daily pattern and gives a correlation of carbon emission impact of the sample space. This research explores the carbon impact of exterior window geometry under the combined effect of natural lighting and artificial lighting and will carry out the statistical analysis of yearly daytime carbon emissions during working hours of the sample space subsequently.

3.6. Value of the Independent Variable

According to the relative rules in GB 50033-2013 [28] and GB 50189-2015 [27], the window-to-floor ratio of the sample should be bigger than 1/5, and the window-to-wall ratio should be less than 0.6. Based on this premise, the other three dependent variables referring to exterior window geometry are the height of the exterior window (H_w), the height of the windowsill (H_ws), and the width of the wall between exterior windows (W_wbw).

When the window is a one-piece shape and the WFR is 1/5, H_w will reach its minimum. Therefore, H_{w min} = 1.95 m. Meanwhile, the upper limit of the window is equal to the height of the bottom of beam. The former takes the maximum value, and we can determine that H_{w max} = 4.2 m (floor height) − 0.15 m (board thickness) − 0.75 m (beam height) − 0.75 m (windowsill height) = 2.55 m. Consequently, the range of exterior window height (H_w) is between 1.95 m and 2.55 m.

According to GB50352-2019 [46], when lighting from the side, the portion of fenestration in civil buildings that is located less than 0.75 m above the finished floor level should not be included in the effective daylighting area, so H_ws min = 0.75 m. Simultaneously, when the sample space has a full window under the beam and the W_w takes the maximum, H_ws reaches its maximum, so H_{ws max} = 4.2 m (floor height) − 0.15 m (board thickness) − 0.75 m (beam height) − 1.95 m (H_{w max}) = 1.35 m. Consequently, the range of H_ws is between 0.75 m and 1.35 m.

When W_w is set at 9.6 m, with the exterior window being full-sized and the width taking the maximum value, W_wbw will reach the minimum, meaning W_{wbw min} = 0.00 m. According to the rules in GB50099-2011 [44], W_wbw shall not be larger than 1.20 m. Hence, the range of W_wbw is between 0.00 m–1.2 m.

3.7. Orthogonal Experiments

Targeting the variety and complexity of design factors and factor level of the exterior window openings in the sample space of this research, we introduced orthogonal experimental design methods. Aligning with the representative factors and horizontal selection characteristics of even distribution, the orthogonal forms of three factors of exterior window opening design and seven factor levels are determined [47]. We combined the values of various factors and levels and used SPSS to generate orthogonal experiments, building 49 pre-analysis simulation models. In the meantime, area A of the sample space was set as southward, which ensured that the sample space embodied the four ambient properties of east, west, south, and north. Ultimately, we attained 196 simulation models of sample studies (

Table 4. Design factors and levels of window geometry in professional drawing classroom.

Level	Window Height	Windowsill Height	Width of Wall between Windows
1	1.95 m	0.75 m	0.00 m
2	2.05 m	0.85 m	0.20 m
3	2.15 m	0.95 m	0.40 m
4	2.25 m	1.05 m	0.60 m
5	2.35 m	1.15 m	0.80 m
6	2.45 m	1.25 m	1.00 m
7	2.55 m	1.35 m	1.2 m

).

3.8. Model Screening

According to research sampling and experimental analysis, when the drawing area and discussion area in the sample space meet the goal of light environmental quality, the aisle area related to them will also meet its goal. Hence, according to the rules in GB/T5699-2017 and GB50035-2013, we used the horizontal work surface of the drawing area and discussion area (0.75 m) as a reference surface. A rectangular analysis grid of 0.5 m * 0.5 m was evenly divided, and measurement points were arranged equidistantly at the center of each grid.

It can be seen from the simulation analysis of daylighting performance using Ladybug and Honeybee that the 196 simulation models drawn from orthogonal experiments all meet the minimum daylight factor of working surface (DF ≥ 3.0) and space daylight autonomy sDA (sDA_500lx,50% ≥ 75%), which can be subsequently used for the calculation of carbon emissions for lighting in different time periods.

3.9. Statistical Analysis

This research adopts relative analysis, boxplot analysis, and multiple linear regression analysis, examining the obvious carbon correlation effect on the drawing classrooms of architecture majors in universities when exterior window geometry is under a certain artificial lighting layout scheme and lighting operation period. We built 196 simulation models based on 49 pre-analysis simulation models according to the four ambient characteristics of east, west, south, and north, which all met the minimum daylight factor of working surface (DF ≥ 3.0) and spatial daylight autonomy sDA (sDA_500lx,50% ≥ 75%). We simulated carbon emissions under artificial lighting multiple times and collected the data. We also conducted relative analysis and boxplot analysis in terms of independent variables (window geometry) and dependent variables (carbon emissions with daylight-controlled artificial lighting, hereafter abbreviated as C) to discuss the relative extents and trends between each parameter and C. Simultaneously, the Pearson correlation coefficient in SPSS was used to assess relativity and connection between each variable. Finally, we used multiple linear regression analysis to build a model for the geometric parameters of the exterior window of the professional drawing space and the carbon emission prediction model of artificial lighting regulated by daylight derived from the regression equation.

4. Results

4.1. The Impact of Exterior Window Geometry on Carbon Emissions in College Professional Drawing Classroom Spaces

This research focuses on college professional drawing classrooms and aims to meet three indicators: the minimum illuminance of a working surface (lx), the minimum daylight factor (%), and daylight autonomy (%). After we carried out linear correlation analysis of carbon emission impacts of the geometry and artificial lighting of the opposing exterior windows facing east, west, south, and north, running coupling from time to time, and adopting Pearson’s correlation coefficient, we derived the analysis result of relativity (

Table 5. Correlation analysis results of exterior window geometry with carbon emissions C.

Orientation	Index	Hw	Hws	Wwbw	Ww	WFR	WWR
South	CO2 (Daytime)	−0.529 **	−0.563 **	0.581 **	−0.581 **	−0.794 **	−0.794 **
North	CO2 (Daytime)	−0.599 **	−0.463 **	0.591 **	−0.591 **	−0.848 **	−0.848 **
West	CO2 (Daytime)	−0.578 **	−0.527 **	0.579 **	−0.579 **	−0.825 **	−0.825 **
East	CO2 (Daytime)	−0.547 **	−0.524 **	0.585 **	−0.585 **	−0.809 **	−0.809 **

Notes: ① When the significance value Sig < 0.05 and the significance value Sig < 0.01 are labeled as two stars (**), respectively. ② The larger the absolute value of the correlation coefficient (r), the closer the relationship between the two.

).

The experimental data show that the correlation coefficients between W_w and W_wbw, as well as WFR and WWR, are identical. This conforms to the linear relationship between the width of the exterior window and the width of wall between windows in a defined space and indicates a constant relationship between WFR and WWR.

W_wbw is positively correlated with the daytime carbon emissions (C) of professional drawing classrooms with different orientations, with a significance level below 0.01. Conversely, WFR, WWR, W_w, H_w, and H_ws are negatively correlated with C, exerting an inhibitory effect, with a significance level below 0.01. The |r| values range between 0.463 and 0.848. Furthermore, the correlation strength decreases in the following order: WFR and WWR, H_w, W_w and W_wbw, H_ws.

4.2. Influence of Different Exterior Window Geometries on Carbon Emissions in College Professional Drawing Classroom Space

According to Table 4, the greatest degree of correlation is between northward W_w and W_wbw (|r| = 0.591) and daytime C, with eastward (|r| = 0.585) and southward (|r| = 0.581) taking the second place, while the westward (|r| = 0.579) has the least correlation. Northward H_w (|r| = 0.599) has the biggest correlation with daytime C, followed by westward (|r| = 0.578) and eastward (|r| = 0.547), while southward (|r| = 0.529) has the least. Southward H_ws (|r| = 0.563) has the biggest correlation with daytime C, followed by westward (|r| = 0.527) and eastward (|r| = 0.524), while northward (|r| = 0.463) has the least. Northward WWR and WFR (|r| = 0.848) have the biggest correlation with daytime C, followed by westward (|r| = 0.825) and eastward (|r| = 0.809), while southward (|r| = 0.794) has the least.

We explored and analyzed the associated impact of C and exterior window geometry in different orientations using boxplots. In the following boxplot, the horizontal line in the middle of the box represents the median value of the data, and the square in the middle of the box represents the mean value of the data. Both represent the average level of the data. The inside of the box contains 50% data, whose upper and lower limits symbolize the upper quartile and lower quartile of the data, respectively. The distance between the top and bottom of the box shows the fluctuation of the data to a certain extent. There are lines outside of the box both in the upper and lower positions, which represent the maximum and minimum of the data.

C, under the effect of exterior window geometry, negatively correlated with H_w, H_ws, and W_w. With H_w, H_ws, and W_w increasing, carbon decreases. However, W_wbw has a positive correlation with C, which means when W_wbw increases, C will increase as well.

When W_w = 1.95 m, H_ws = 0.85 m, W_w = 2.33 m, and W_wbw = 1.2 m, C reaches its maximum in the southward classroom, which is 2107 kg/a. When H_w = 2.45 m, H_ws = 1.35 m, W_w = 3.13 m, and W_wbw = 0 m, C decreases to its minimum in the westward classroom, which is 1666 kg/a.

When H_w, H_ws, and W_w increase, the median value of C in classrooms with different orientations shows a constant trend of decline. The average value of C also declines, except when H_w changes from 2.45 m to 2.55 m and H_ws changes from 1.25 m to 1.35 m (Figure 8, Figure 9 and Figure 10).

When W_wbw increases, the median value of C in southward, westward, and eastward classrooms constantly increases, except when W_wbw changes from 0.2 m to 0.4 m. The mean value of C in all classrooms increases.

There are fluctuations in the changing tendency of median and mean values of C in different classrooms. Taken together, the median and mean values of C in eastward and westward classrooms have a similar changing tendency and the latter is slightly lower, while values in southward and northward classrooms vary greatly, but are generally bigger than the eastward and westward classrooms.

In terms of the WFR and WWR of professional drawing classrooms, they have a negative correlation with C. When WFR or WWR increase, C in classrooms with different orientations shows a tendency to decline.

When WFR is 0.169 and WWR is 0.362, C in the southward classroom reaches its maximum, which is 2107 kg/a. Correspondingly, when WFR is 0.244 and WWR is 0.522, C in the westward classroom shows the minimum value, which is 1666 kg/a.

When WFR and WWR both increase, the median and mean value of carbon in classrooms with different orientations decrease stably, and the tendency is westward < eastward < southward < northward.

Compared with H_w, H_ws, W_w, and W_wbw, the difference in the median and mean values of C among classrooms resulting from the change of WFR and WWR is relatively fixed, which is westward < eastward < southward < northward.

By means of relativity analysis and boxplot analysis, we chose geometric parameters that have prominent relativity as the independent variable and carbon emissions during the daytime working period as the dependent variable to establish a prediction model of different orientations of professional drawing classroom space in the comprehensive light environment quality and carbon impact.

This research adopts multiple linear regression and weighted least squares (WLS). We seek to provide direct guidance for windows in professional drawing classrooms during both the early designing and actual operation periods. We choose Hw, Hws, and Ww as independent variables to carry out multiple linear regression and fit linear regression models.

The regression equations of carbon emissions (C) during yearly daytime working hours in professional drawing classrooms from four orientations are as follows.

C (Southward) = −282.937·H_w − 290.334·H_ws − 213.090·W_w + 3468.990

C (Northward) = −256.752·H_w − 184.515·H_ws − 162.999·W_w + 3167.977

C (Westward) = −320.336·H_w − 276.692·H_ws − 217.189·W_w + 3533.945

C (Eastward) = −278.283·H_w − 265.719·H_ws − 219.304·W_w + 3438.248

The adjusted R² values for the multiple linear regression models of the annual daytime C of professional drawing classrooms in the south, north, west, and east orientations are 0.940, 0.926, 0.962, and 0.913, respectively. The prediction model showing the regression equation can illustrate the change of carbon emissions (C) during yearly daytime working hours as 94.0% in southward, 92.6% in northward, 96.2% in eastward, and 91.3% in westward professional drawing classrooms. Adjusted R² shows that this model has a goodness of fit.

A relativity analysis of the impact of exterior window geometry parameters and artificial lighting operation on carbon in professional drawing classrooms of different orientations in universities was carried out, and the statistical results are as follows (

Table 6. Results of the combined analysis of multiple linear regression models for C in professional drawing classrooms with different orientations.

Orientation	Adjusted R2	Sig	DW	Independent Variable	Standardized Coefficients	t	Sig	VIF
South	0.940	0.000	2.068	Constant	——	58.747	0.000	——
				Hw	−0.562	−15.924	0.000	1.000
				Hws	−0.576	−16.341	0.000	1.000
				Ww	−0.541	−15.341	0.000	1.000
North	0.926	0.000	1.519	Constant	——	63.830	0.000	——
				Hw	−0.598	−15.214	0.000	1.000
				Hws	−0.487	−12.377	0.000	1.000
				Ww	−0.574	−14.578	0.000	1.000
West	0.962	0.000	1.822	Constant	——	71.841	0.000	——
				Hw	−0.629	−22.467	0.000	1.000
				Hws	−0.544	−19.406	0.000	1.000
				Ww	−0.510	−18.218	0.000	1.000
East	0.913	0.000	2.032	Constant	——	49.529	0.000	——
				Hw	−0.552	−12.982	0.000	1.000
				Hws	−0.528	−12.396	0.000	1.000
				Ww	−0.579	−13.604	0.000	1.000

).

In the F-test and t-test, when each significance value is less than 0.05, the former means that the regression equation has statistical significance, while the latter means that the regression coefficient of each independent variable is prominent, and the independent variables and dependent variables have remarkable relativity. In the collinearity analysis of independent variables, the value of VIF is 1.0, which is less than 0.5, meaning that there is no multicollinearity in the regression equation.

It can be seen from the standardized regression coefficient that for the southward professional drawing classroom, the impacts of parameters H_ws, H_w, and W_w diminish in sequence; for the northward classroom, the sequence is H_w, W_w, and H_ws; for the westward classroom, the sequence becomes H_w, H_ws, and W_w; and for the eastward classroom, the order is W_w, H_w, and H_ws.

4.3. The Impact of Different Operation Periods and Different Orientations on Carbon Emissions of Professional Drawing Classroom Spaces in Universities

Different operation periods during day and night have huge impacts on C. During daytime, C emissions in classrooms with different orientations differ from each other, while during nighttime they are the same (C = 851 kg/a).

With the same running time, C in the forenoon is apparently lower than C in the afternoon. In the forenoon period, the median values of C in classrooms with different orientations are approximately the same; however, in the afternoon period, its levels in classrooms with different orientations is in the order of westward < eastward < southward < northward (Figure 10).

5. Discussion

5.1. Impact of Window Geometry

By conducting correlation analysis, box-plot analysis, and multiple linear regression analysis, a simulation model was built to predict the impact of window geometry on carbon emission, using a case study of professional drawing classrooms with windows in different orientations.

The geometry of external windows, such as shape, area, and location, had an impact on the natural lighting utilization. It also affected the configuration of artificial lighting. The results of the simulation demonstrated a significant and inverse relationship between these two factors, indicating that utilization of natural lighting can reduce the demand for artificial lighting. Moreover, the operation of artificial lighting directly contributes to carbon emissions through electricity consumption. In southward, northward, westward, and eastward spaces, the Pearson correlation coefficients for the impact of Hw, Hws, Ww, and WFR on carbon emissions were −0.529, −0.563, −0.581, and −0.794; −0.599, −0.463, −0.591, and −0.848; −0.578, −0.527, −0.579, and −0.825; and −0.547, −0.524, −0.585, and −0.809, respectively. Therefore, the hypothesis regarding the association between the geometry of windows and the impact on carbon emissions is feasible.

The geometric parameter system that consists of H_w, H_ws, W_wbw, W_w, WWR, and WFR had a direct influence on carbon emissions. H_ws and H_w, as well as W_wbw and Ww, respectively, defined the vertical and horizontal position of the windows. Additionally, WWR and WFR determined the area of opening for double-sided windows.

Different window geometries impacted the total carbon emissions in various ways. When the building area is fixed, WFR and WWR appeared to have the strongest connections to carbon emission. WFR, WWR, W_w, H_w, and H_ws had a suppressing effect on daytime carbon emissions in professional drawing classrooms. Additionally, W_wbw showed a positive correlation with carbon emissions. Among these factors, H_w, Ww, and H_ws determined the geometric shape, area, and position of the external windows. Based on the computer simulation results, it was found that windows with larger opening areas and higher sills could best reduce carbon emissions.

Under the condition of double-sided windows, the overall driver of higher carbon emissions in professional drawing spaces was as follows: westward > eastward > southward > northward. These different window orientations showed a positive correlation with the variable DA. At the same time, there were variations in the influence of independent variables on carbon emissions in the multivariate linear regression equations for different orientations. Specifically, the influence of H_ws, H_w, and Ww gradually decreased in south-facing professional drawing classrooms, while the influence of H_w, Ww, and H_ws gradually decreased in north-facing classrooms. For west-facing classrooms, the influence of H_w, H_ws, and W_w gradually decreased, and for east-facing classrooms, the influence of W_w, H_w, and H_ws gradually decreased.

During the daytime operation period, the geometry of external windows directly affects the natural lighting use of the space. Natural lighting and artificial lighting have a dynamic coupling relationship. Therefore, there exists significant association between the window geometry and carbon emission. Taking the selected professional drawing classrooms in a university in Shenyang as an example, the natural lighting in the morning period (8:00–12:00) was significantly better than in the afternoon period (14:00–18:00). During the morning period, the dependence on artificial lighting is lower compared to the afternoon, resulting in lower energy consumption and lower carbon emission. During the nighttime operation period (19:00–22:00), natural lighting is not available, and therefore there is no data association between window geometry and carbon emissions.

5.2. Limitations

This study has certain limitations. The selected location for this research is in Shenyang, belonging to the severely cold climate zone (C zone), making the results region-specific. Secondly, the study does not consider the vertical layers of the sample prototype space. The window opening quantity only considers two conditions: full-window and three-band windows. The indoor artificial lighting scheme is an optimized fixed lighting plan. Glare issues within the sample prototype space are not included in this study. Due to limitations in software development, DIALux evo allows setting only one sensor for single-space simulation analysis.

Future studies need to be conducted on the impact of carbon emissions using different building types as case studies. Future studies also need to focus on various different artificial lighting schemes and diverse climate zones. The predictive model also needs to input more detailed parameters such as floor-to-ceiling windows, multiple strip windows, and different lighting fixtures. By integrating specific operation plans for different building types, we aim to derive segmented equations for the daily carbon emissions that are applicable to various time periods.

6. Conclusions

Energy-efficient lighting is a key factor in achieving a low-carbon campus. The electricity consumption of illumination in professional drafting classrooms is influenced by both natural daylight utilization and artificial lighting schemes. The geometry of exterior windows significantly impacts natural daylight utilization, thereby influencing its associated carbon emissions. Taking Shenyang, China as an example, simulations, correlation analysis, and multiple linear regression were performed on six design parameters to investigate the effect of various parameters of external window design on the carbon emissions of professional drafting classrooms. The results indicate significant correlations between WFR, WWR, H_ws, W_w, W_wbw, H_w, and C, with WFR and WWR showing the highest correlation with C. In southward, northward, westward, and eastward spaces, the standardized regression coefficients for H_w, H_ws, and W_w in influencing carbon emissions were −0.562, −0.576, and −0.541; −0.598, −0.487, and −0.574; −0.629, −0.544, and −0.510; and −0.552, −0.528, and −0.579, respectively.

Using H_w, H_ws, and W_w as independent variables, and considering the prerequisite of achieving the sDA standard in professional drafting classrooms, predictive models for natural lighting, artificial lighting, and building carbon emissions were constructed for different orientations of the classrooms. Under the constraints of orthogonal experimental factor levels for window geometry, and while meeting the requirements for WFR and WWR, the west-facing specialized drawing classroom with H_w of 2.55 m, H_ws of 0.75 m, and W_w of 9.6 m achieved the minimum carbon emissions (C) while satisfying the requirements for WFR and WWR. This research proposes quantitative design references for low-carbon design of professional drafting classrooms in severe cold regions. Additionally, it aims to offer quantitative evaluation criteria for the carbon operation status and low-carbon campus construction of university buildings.