Consideration of Thermal Comfort, Daylighting Comfort, and Life-Cycle Decarbonization in the Retrofit of Kindergarten Buildings in China: A Case Study

Abstract

Kindergartens play a crucial role in nurturing the physical, cognitive, and social development of children. Hence, designing kindergarten buildings requires the consideration of the unique requirements and behavior of children. Considering the rapid urbanization of China and its commitment to achieving the 3060 carbon goal, in this study, we examine the retrofitting of kindergarten buildings in China and propose a retrofit optimization method for kindergarten buildings that considers thermal comfort, daylighting, and life-cycle carbon emissions. Through this method, information on the thermal and daylighting comfort of occupants, weather data, occupant scheduling, and envelope and energy system of the kindergarten building to be retrofitted can be obtained through various approaches, such as video playback, field investigation, literature research, and consult drawings. On this basis, optimization variables are selected, and a physical model is established to guide the retrofit process. Afterward, a rapid comprehensive optimization framework based on parallel computing is adopted to obtain the comprehensive optimal design scheme for the building to be retrofitted. The proposed method is applied to a kindergarten building retrofit case in Nanjing, China, and the results show that the optimal comprehensive scheme results in a reduction in carbon emissions of 34,158.3 kg, an increase in the thermal comfort period of 2.7%, and an improvement in daylighting comfort of 79.7% over the benchmark scheme. The significance of this study extends beyond its potential for widespread application in kindergarten building retrofits. It contributes to advancing sustainable building design and environmental stewardship, creating healthier and more comfortable learning environments for children while mitigating the environmental impact of buildings. Moreover, it emphasizes the importance of considering children’s unique needs and behaviors in building design, ultimately leading to better outcomes for their overall development.

1. Introduction

1.1. Background

Human-centered and sustainable development are two essential themes in the realm of building design and are crucial concepts that must be considered in the building design process. The kindergarten, a quintessential educational building, has received much attention from society due to its primary occupants, namely, children. With the urbanization rate of China surpassing 60% [1], urban renewal and retrofitting will serve as a primary focus in the field of building design in the future, with kindergartens as a key area of interest among designers and researchers. In traditional retrofitting of kindergartens, emphasis is placed on seismic reinforcement, thermal insulation, and energy conservation, while the comfort of occupants has been considered less [2,3].

Kindergartens are occupied primarily by children and teachers, with children constituting 90% of the total occupants. As children are at the stage of growth and development, the comfort of the indoor environment greatly impacts their physical and mental health. It has been shown in previous studies that a reasonable lighting environment can enhance the learning efficiency and decrease the incidence of myopia among children [4], while a comfortable thermal environment is beneficial to brain development and the mental health of children [5,6]. While trade-offs between thermal comfort, daylighting, and energy have been considered in some optimization studies of educational buildings [7], few studies have targeted design optimization of kindergarten building environments by fully considering the physiological and behavioral differences between children and adults.

Presently, the carbon peaking and carbon neutrality policy of the Chinese government, released in 2020, imposes higher requirements on energy conservation and emission reduction for educational buildings. Traditional energy-saving retrofitting only targets the operational energy consumption of buildings and fails to consider their carbon emissions from a life-cycle perspective. Therefore, achieving life-cycle decarbonization retrofitting of kindergarten buildings while satisfying the comfort requirements of occupants is a crucial challenge. This is very important for promoting the healthy growth of children and realizing building decarbonization.

1.2. Literature Review

1.2.1. Study on the Thermal Comfort of Children

Children are the primary occupants of kindergartens, and the indoor thermal environment is closely linked to their growth and development. However, the thermal comfort of children is often neglected by adults. Existing research on indoor thermal comfort primarily focuses on office and residential spaces for adults, with relatively few studies addressing the thermal environment of educational buildings. The main indicators of thermal comfort currently include the air temperature, operative temperature, predicted mean vote (PMV), and predicted percentage dissatisfied (PPD). In several studies, thermal comfort in educational buildings has been explored. For instance, Wang et al. investigated thermal comfort in a school classroom in Germany, with the air temperature as an evaluation indicator, and they targeted the thermal comfort range between 20 °C and 26 °C [8]. Xu et al. studied thermal comfort optimization of primary and secondary school classrooms in China, utilizing the satisfaction of the temperature set point and PMV to measure thermal comfort in classrooms [9,10]. Al-Rashidi et al. calculated clothing insulation for Kuwaiti boys and girls aged 6 to 17 in winter and summer, highlighting the importance of studying the clothing insulation characteristics of different age groups in PMV calculations [11]. Jindal studied the thermal environment in a classroom of a government residential school in the composite climate zone of Ambala, India, noting that students exhibit favorable heat resistance in terms of the operative temperature, with their thermal comfort range exceeding Indian and international adult standards [12]. Zomorodian et al. reviewed the research on thermal comfort in different educational buildings and examined different thermal comfort evaluation indicators and their applicability [13]. Li et al. investigated thermal comfort in kindergartens in Chongqing and Wuhan, China, using the PMV and thermal sensation vote (TSV) indicators to quantify the thermal comfort of children. They found that the main factors affecting the thermal comfort of children were the activity level and subjective regulation of the indoor thermal environment by teachers [5].

The above clearly indicates that the most widely used evaluation indicator for the thermal comfort of children is the PMV, which is suitable for steady-state thermal comfort evaluation in air-conditioned rooms. However, only a few researchers have fully considered the adaptability of the PMV indicator to the evaluation of thermal comfort among children of different ages. Additionally, there are few studies on the thermal comfort of children in kindergartens, where children are typically 4–6 years old (corresponding to grades 1, 2, and 3). Their physiological and activity characteristics significantly differ from those of primary and secondary school students. Therefore, further study is needed to determine the annual hourly thermal comfort of children in kindergarten buildings.

In contrast to office and commercial buildings, kindergarten buildings exhibit notable regional characteristics due to the influence of region, climate, customs, and teaching arrangements. In most existing research on thermal comfort optimization of kindergartens in China, certain measures have been adopted to improve the thermal environment in kindergartens (such as strengthening natural ventilation and local space adjustment), but the thermal environment in kindergarten buildings has been optimized from a whole-building perspective in few studies.

1.2.2. Study on the Daylighting Comfort of Children

Daylight irradiation has been shown to promote bone development, prevent rickets, osteoporosis, and other diseases, and enhance the metabolism and blood circulation in children [14]. The optimization of daylighting comfort in kindergarten buildings has been widely studied. For example, Pagliolico et al. investigated the application of a photobio screen (PBS) shading system in a kindergarten in San Marcel, Italy, using various daylighting indicators to measure visual comfort. They found that the PBS shading system effectively facilitated an improvement in visual comfort in the kindergarten classroom [15]. Samiou et al. optimized daylighting in Greek kindergarten classrooms using spatial daylight autonomy (sDA), annual sun exposure (ASE), and daylight glare probability (DGP) as evaluation indicators for daylighting comfort. They optimized the size, position, and shading equipment of external windows and found that ASE and glare effects in the classroom were effectively alleviated after optimization [16]. Vásquez et al. proposed a multimethod approach to determine the preferences of children for the light environment and window landscape in kindergarten classrooms, and found that children can distinguish daylighting needs according to their activities [17]. Salleh et al. investigated the indoor environment in kindergarten buildings in Malaysia and found that glare in the classroom is the main factor affecting visual comfort [18]. Gao et al. studied the daylighting optimization of Chinese kindergartens using the daylight factor (DF) and illuminance, analyzing the effects of various local daylighting optimization measures for kindergarten classrooms, including orientation, interior decoration, classroom size, side window height, skylight position, and reflector features [19].

Although the daylighting of kindergartens has been extensively researched, kindergarten buildings often have distinct regional characteristics. Notably, the building scale, number of stories, classroom size, and number of students are closely related to regional characteristics. Chinese kindergartens are generally multistory buildings with internal corridors, and their classrooms typically serve multiple functions, such as teaching, activities, lunch breaks, and dining, resulting in larger classroom areas than those found in primary and secondary schools. The architectural features of large classrooms and inner corridors (external windows cannot be set on both the northern and southern sides) make it difficult to optimize the daylighting of kindergarten classrooms, and further research is needed.

1.2.3. Energy Conservation and Emission Reduction of Kindergarten Buildings

Carbon emissions have always been a research hotspot in the field of building design, and many researchers have explored this topic [20,21,22,23]. In 2019, the Ministry of Housing and Urban–Rural Development of China issued a standard for calculating building carbon emissions, which clearly specifies the carbon emission calculation methods for the production, construction, operation, and final demolition stages [24]. In 2020, the Chinese government issued a carbon peaking and carbon neutrality policy, which greatly promoted research in this field [25,26]. Therefore, the optimization of kindergarten design must account for the issue of carbon emissions. Hammad et al. investigated energy conservation and emission reduction of kindergartens in Jordan, achieving an annual reduction in carbon dioxide emissions of 11.7 t through increased wall insulation, solar hot water systems, photovoltaic systems, and fresh air heat recovery [27]. Zhang et al. proposed a green design method for kindergartens based on renewable energy, reducing fossil energy consumption through the integration of renewable energy at the site and surrounding environment [28]. Causone et al. studied the retrofitting of a zero-energy kindergarten in a smart district renovation project, significantly reducing HVAC energy consumption and improving the indoor environmental quality through optimization [3]. Gajic et al. proposed an energy performance indicator by designing and measuring the parameters of representative samples of old and new kindergartens located in the temperate climate zone of Banja Luka, suggesting that kindergartens should be regarded as a unique type of building in building energy audits and certifications [2].

While energy conservation and emission reduction of kindergartens have been investigated in the above studies, the focus was on operating energy consumption and carbon emissions, with few studies analyzing kindergarten carbon emissions from a life-cycle perspective. Additionally, there are significant differences between the optimization of new buildings and retrofit optimization of existing buildings. In retrofitting existing buildings, greater attention is given to the application of envelope thermal insulation materials and renewable energy, such as photovoltaic systems [29]. Thus, the embodied carbon emissions of materials must be considered in the optimization process.

In references [8,30,31], it was noted that maintaining a comfortable indoor environment often requires high energy consumption, which could also lead to an increase in building carbon emissions. Therefore, multi-objective optimization methods are often applied in design optimization studies that balance comfort and carbon emissions to obtain suitable design schemes (including the weight method and Pareto method [32]). However, existing research indicates that there are significant differences in comfort needs among different populations [33,34]. If building design optimization can be further tailored to the comfort range of a specific group of people, it is clear that a better balance between comfort and carbon emissions can be attained. Children and kindergarten buildings are typical representatives of specific population groups and building types for retrofitting, respectively, and there is currently a lack of research on the correlation between the comfort of children and energy consumption in kindergarten buildings.

1.3. Research Gaps and Main Contributions

Drawing from the aforementioned review, it is evident that additional research is imperative to explore retrofit optimization strategies for kindergarten buildings in China, with a particular emphasis on assessing the annual hourly thermal and daylighting comfort levels of 4-to-6-year-old children as the primary occupants. Furthermore, the trade-off between occupant comfort and life-cycle decarbonization of these buildings warrants further investigation. Therefore, this paper proposes a design optimization method for retrofitting kindergarten buildings, which holistically integrates occupants’ comfort evaluation and life-cycle decarbonization considerations.

The remainder of this paper is arranged as follows: in Section 2, the overall method of life-cycle decarbonization of comfortable kindergarten buildings is introduced in detail. In Section 3, the method is applied to the retrofitting of a real kindergarten project, and the results are analyzed. In Section 4, the study findings are examined, and prospects for future research are provided. A summary is presented in the last section of this paper.

2. Methodology

The fundamental process of this study, as depicted in Figure 1, can be primarily segmented into three components: theoretical research on children’s comfort, information gathering and evaluation prior to retrofit design, and the optimization and analysis of design schemes. Given that the comfort of indoor thermal and daylight environments is primarily influenced by the building envelope and energy systems, the primary focus of this study is on optimizing the envelope and energy system of kindergarten buildings.

2.1. Information Acquisition and Evaluation before Retrofit Design

2.1.1. Information about the Thermal Comfort of Occupants

Based on the literature review provided in Section 1.2.1, the PMV is the most commonly used indicator for evaluating the thermal comfort of children. It comprehensively considers the effects of ambient air temperature, radiation temperature, air humidity, air flow rate, human activity level, and clothing insulation on thermal comfort. In this study, the PMV is utilized as an indicator for evaluating the indoor thermal comfort in kindergarten buildings. Notably, in addition to environmental variables, the PMV is significantly influenced by the clothing insulation and metabolic rate of occupants [35]. It is important to recognize that the primary occupants of kindergartens are children. Therefore, when evaluating thermal comfort, it is crucial to consider the clothing insulation and active metabolic rate of children.

Clothing insulation is influenced by various factors, including the outdoor temperature, living habits, and activities. Although simple clothing insulation estimation methods have been provided in standards such as ASHRAE Standard 55-2013 [36] and ASHRAE Handbook 2009 [37], these standards are primarily applicable to adults. In Chinese kindergartens, clothing for children is often managed by nursing teachers to facilitate teaching and management. These teachers require children to dress or undress according to real-time activities and the indoor temperature. To ensure the practicality of retrofit optimization, we employ field investigations and video reviews in this study to determine the clothing worn by children during typical periods in different seasons. Equation (1) can then be used to calculate the insulation of the clothing worn by children during different periods [36].

$$I_{cl}=0.835{\sum }_i{I}_{clu,i}+0.161$$

(1)

where $I_{cl}$ is the insulation of the clothing as a whole in clo and $I_{clu,i}$ is the insulation of a single piece of clothing in clo. Numerous standards and studies have provided metabolic rate values for adults engaged in activities of different levels [38]. However, the metabolic rate of children significantly differs from that of adults, and there are considerable variations in the metabolic rate among children of different ages [5]. In this study, Equations (2)–(5) can be used to calculate the basic metabolic rate of children, which involves determining the resting metabolic rate and skin surface area of children [39]. These parameters depend on the child’s weight and height, as well as other factors such as age, region, and eating habits. It should be noted that Equations (2) and (4) only apply to children aged 3–10 years [39]. To obtain the metabolic rate of children engaged in different activities, field investigation should be conducted to obtain physical information of children in the kindergarten to be retrofitted. Afterward, various activity-related metabolic rates can be calculated based on the activity level conversion specified in ISO 8996 [40].

$$RMR=0.082W_c+0.545H_c+1.736$$

(2)

$$Q={\displaystyle \frac{RMR\times {10}^6}{3600\times 24}}$$

(3)

$$A_D=0.00659H_c+0.0126{AW}_c-0.1603$$

(4)

$$M_b={\displaystyle \frac{Q}{A_D}}$$

(5)

where $RMR$ is the resting metabolic rate in MJ/d; $W_c$ is the child’s weight in kg; $H_c$ is the child’s height in m; $Q$ is the metabolic rate per unit time in W; $A_D$ is the child’s skin surface area of the child in m²; and $M_b$ is the basic metabolic rate of the child in W/m².

2.1.2. Information on Occupant Daylighting Comfort

Based on Section 1.2.2, an excessively high illuminance can cause glare, while insufficient illuminance can result in visual impairment and other problems for children. In previous studies, values of 100 lux and 2000/3000 lux have been utilized as the lower and upper limits of illuminance, respectively, as per references [7,41,42]. However, these values are primarily applicable to adults, and the acceptable range of illuminance for the visual comfort of children differs significantly from that for adults. Before retrofitting, the lower and upper limits of illuminance should be determined based on the age of the children [19].

2.1.3. Information on Weather Data, Occupant Scheduling Data, and Material Data

Weather data, occupant scheduling data, and material data involved in the retrofit are also important contents of the information acquisition process before retrofit design. Weather data primarily comprise the annual hourly outdoor dry bulb temperature, humidity, wind speed, wind direction, and illuminance, which can be obtained through literature investigation and research. Occupant scheduling data primarily involve hourly usage patterns in different rooms, including timing control strategies for equipment in rooms, and can typically be obtained through field investigation. Material data involved in retrofitting can generally be obtained through literature reviews and research.

2.1.4. Information on the Original Building Envelope and Energy System

During pre-retrofitting information acquisition, evaluating the envelope and energy system of the original building is essential. While some original buildings may contain a building energy management system (BAS) that can be used to perform a comprehensive and detailed evaluation, the equipment performance in most original buildings is relatively outdated, and a BAS is not commonly configured. Therefore, in this study, the original drawings and field investigation data are used to evaluate the building envelope and energy system to determine the parts that require retrofitting. During envelope investigation, it is crucial to obtain information on the thermal insulation performance of the external wall, external windows, roof, and ground of the original building. Similarly, during energy system investigation, it is essential to evaluate the performance of existing lighting lamps, domestic hot water equipment, elevator equipment, and HVAC equipment.

2.2. Optimization Calculation for the Comprehensive Optimal Scheme

After information acquisition and evaluation before retrofit design, the design optimization of the scheme can be targeted.

2.2.1. Selection of Optimization Objectives and Variables

Section 1.3 indicates that we focus on the trade-off between thermal comfort, daylighting, and life-cycle decarbonization through multi-objective optimization. The optimization objectives are defined in Equations (6)–(8), where larger TC and UDI values indicate better performance, whereas smaller $C_{lc}$ values indicate better performance.

$$\left\{\begin{array}{c}TC={\displaystyle \frac{{\sum }_i\left(t{f}_i{t}_i\right)}{{\sum }_i{t}_i}}\in [0,1]\\ t{f}_i=\left\{\begin{array}{c}1 if-1\le PMV\le 1\\ 0 if PMV<-1 \vee PMV>1\end{array}\right.\end{array}\right.$$

(6)

$$\left\{\begin{array}{c}UDI={\displaystyle \frac{{\sum }_i\left(w{f}_i{t}_i\right)}{{\sum }_i{t}_i}}\in [0,1]\\ w{f}_i=\left\{\begin{array}{c}1 \mathrm{if} E_{Lower lim}\le E_{Daylight}\le E_{Upper lim}\\ 0 \mathrm{if} E_{Daylight}<E_{Lower lim}\vee E_{Daylight}>E_{Upper lim}\end{array}\right.\end{array}\right.$$

(7)

$$\left\{\begin{array}{c}{C}_{lc}=C_m+C_o+C_r\\ C_m=C_{m1}+C_{m2}\\ C_{m1}=\sum _{l=1}^n{P}_{ml}\times K_{ml}\\ C_{m2}=\sum _{j=1}^n{P}_{cj}\times K_{cj}\times A_{cj}\\ C_o=TES\times K_d\times Y\\ C_r=C_{r1}+C_{r2}\\ C_{r1}=C_m\times {cof}_r\\ C_{r2}=\sum _{l=1}^n{P}_{ml}\times K_{ml}\times {cof}_l\end{array}\right.$$

(8)

where $TC$ denotes the annual occupied time fraction of thermal comfort hours, $i$ denotes the number of occupied hours in a year, $t_i$ denotes each occupied hour in a year, $t{f}_i$ denotes a thermal comfort weighting factor, $UDI$ denotes the annual occupied time fraction of the indoor horizontal daylight illuminance observed at a given test point in a given domain, $w{f}_i$ denotes a daylighting weighting factor, $E_{Daylight}$ denotes the horizontal illuminance at a given point in lux, $E_{Upper lim}$ denotes the preset upper limit value of the horizontal illuminance at a given point in lux, and $E_{Lower lim}$ denotes the preset lower limit value of the horizontal illuminance at a given point in lux. Moreover $C_{lc}$ denotes the total carbon emissions during the building life cycle in kg; $C_m$ denotes the carbon emissions at the building manufacturing stage in kg; $C_o$ denotes the carbon emissions at the building operation stage in kg; $C_r$ denotes the carbon emissions at the building end-of-life recycling stage in kg; $C_{m1}$ denotes the embodied carbon emissions of materials in kg; $C_{m2}$ denotes the construction carbon emission in kg; $P_{ml}$ denotes the quantity of various building materials used in t; $K_{ml}$ denotes the carbon emission factors of different materials in tCO₂/t; $l$ denotes the types of materials; $P_{cj}$ denotes the quantities of a certain construction process in m³; $K_{cj}$ denotes the carbon emission factors of a certain construction process in tCO₂/m³; $A_{cj}$ denotes the area of a certain construction process in m²; $j$ denotes the construction process types; $TES$ denotes the annual power consumption in kWh; $K_d$ denotes the carbon emission factor of electricity obtained from the grid (generation using fossil fuels) in kgCO₂/kWh; $Y$ denotes the theoretical building life in years; $C_{r1}$ denotes the carbon emission during building demolition in kg; $C_{r2}$ denotes the decarbonization amount recovered from building materials in kg; ${cof}_r$ denotes the carbon emission coefficient of building demolition (0.1 in this study according to reference [43]); and ${cof}_l$ denotes the recovery factor of various building materials.

Given the optimization objectives outlined and the research focus of this paper (i.e., retrofitting kindergarten buildings), the optimization variables can be classified into four primary categories: external wall-related variables, external window-related variables, external shading-related variables, and energy system-related variables. These variables have been described in detail in our team’s previous studies [9,44], and will not be repeated here.

2.2.2. Model Establishment and Parameter Setting

In this study, we used SketchUp to create a building model, and then imported EnergyPlus through OpenStudio for parameter setting and simulation calculation purposes. The model parameters primarily include weather, scheduling, material, thermal comfort, and daylighting comfort-related parameters. Weather parameters were set using data from the EnergyPlus website. Scheduling parameters were based on hourly usage patterns from field investigations. Material parameters were set considering performance and embodied carbon emissions of building materials from literature research. Thermal comfort parameters were determined by hourly clothing insulation and active metabolic rate from video playback and field investigation. Daylighting comfort parameters include appropriate illuminance value and sensor grid density [16].

2.2.3. Fast Optimization Calculation

After setting the model parameters, fast comprehensive optimization of the scheme can be conducted. Meta-models represented by artificial neural networks (ANNs) are currently commonly used in design optimization research, and their adoption can effectively increase the efficiency of design optimization by replacing physical models. However, our team has noted in previous studies that generating sample spaces and optimizing ANN models typically require a significant amount of time [10,44]. To address this issue, we propose a scheme for a fast, comprehensive optimization framework based on parallel computing, as shown in Figure 2. The parallel computation process of this framework involves two main components: generating the sample space and optimizing the ANN model. In the calculation process, an initial sample with a size of 100 was generated by coupling EnergyPlus with Python language, and the sample was used for ANN model training and testing (divided into a training set and a test set at a ratio of 4:1). According to our team’s reference [10], the SEGA-Q algorithm was used in this study for ANN model hyperparameter optimization, and the specific process will not be repeated here. When the ANN model is optimized, the sample space generation part is continued. If the mean relative error (MRE) of the optimized ANN model meets the requirement (MRE ≤ 1%), the model is adopted in the subsequent analysis. If not, the generated samples are added to the initial sample set to form a new sample, which is then used to optimize the ANN model. After obtaining the optimal ANN model, it can be combined with optimization algorithms for multi-objective optimization calculations to generate a Pareto solution set. Our team has compared the performance of the above three algorithms in building design optimization in previous studies [9,10], and the results show that the NSGA-II provides significant advantages. However, in reference [45], the performance of the NSGA-II and NSGA-III models was further compared in building design optimization, and the results showed that the NSGA-III model is more suitable for solving multi-objective optimization problems with more than two objectives. Therefore, the NSGA-III model was used as a multi-objective optimization algorithm in this study. On this basis, this study uses the entropy weight method to screen the Pareto set, in order to select a comprehensive optimization scheme that takes into account thermal comfort, daylighting, and life-cycle decarbonization [46].

3. Case Study

3.1. Case Information

To demonstrate the effectiveness of the proposed approach for retrofitting kindergarten buildings, we conducted a case study of a kindergarten in Nanjing from the 1990s and optimized its retrofit design scheme. The geographical coordinates of Nanjing are 31°14″ to 32°37″ N and 118°22″ to 119°14″ E. It is situated in a subtropical monsoon climate zone and is classified as a hot-summer and cold-winter area in the Chinese construction thermal zoning classification. Throughout the year, there is a demand for both cooling and heating. The typical annual outdoor air dry bulb temperature change in Nanjing is shown in Figure 3.

Detailed information on the kindergarten is presented in

Table 1. Detailed information on the case kindergarten.

Parameters	Values (Building 1)
Number of floors	2
Number of classrooms per floor	5
Floor height (m)	3.5
Building area (m2)	321
Total floor area (m2)	664.8
Classroom length and width (m)	6.4
Total number of children	183
Number of teachers	32
Vacation	Winter vacation: 31 January to 26 February Summer vacation: 1 July to 31 August
Occupied period (weekdays)	8:00–16:00

, and its appearance (before retrofit) and overall layout are shown in Figure 4. Notably, the kindergarten comprises two buildings, with Building 1 serving as the primary teaching facility, featuring a total of nine standard classrooms and one multifunctional classroom. The first grade encompasses four classrooms, while the second and third grades encompass three classrooms, with approximately the same number of students in each class. Building 2 is an office building that includes offices, canteens, and teacher activity rooms. Given that the primary focus of this study is to ensure a comfortable environment for children, design scheme optimization is solely confined to the main teaching building (Building 1). In addition, there are multiple residential buildings around the kindergarten, and in practical terms, the obstruction of these surrounding buildings will have a significant impact on the daylighting of the kindergarten. However, considering that the focus of this study is on the design optimization of kindergarten buildings, the occlusion of these surrounding buildings was ignored.

3.2. Information Acquisition and Evaluation of the Case Kindergarten

Based on the content in Section 2, the first step involves conducting a survey to evaluate the thermal comfort of children in the kindergarten. To this end, we employed field investigation and video playback techniques (as illustrated in Figure 5) to gather information on the annual clothing conditions of children across three grades in the kindergarten. Subsequently, we calculated the clothing insulation for each period using Equation (2), and the results are listed in

Table 2. Average clothing insulation for each period.

Seasons	Activity Types	Values [clo]
Summer	Play activities	0.31
	Classes
	Meals
	Noon break	0.33
Winter	Play activities	1.52
	Classes
	Meals
	Noon break	1.80

. Additionally, we collected information on the height and weight of the children in these three grades. With the use of these data and Equations (3)–(6), we calculated the metabolic rate of the children under various activities, as presented in

Table 3. Average height, weight, and metabolic rate of the children in each grade.

Grades Parameters	1	2	3
Height [cm]	101.3	107.6	115.1
Weight [kg]	18.2	19.4	22.5
Basic metabolic rate [W/m2]	62.5	60.2	58.0
Metabolic rate during the various activities [W/m2]	Play activities: 187.5	Play activities: 180.6	Play activities: 174
	Classes: 90.6	Classes: 87.3	Classes: 84.1
	Meals: 154.4	Meals: 148.7	Meals: 143.3
	Noon break: 63.1	Noon break: 60.8	Noon break: 58.6

Note: The conversion relationship between the above basic metabolic rate and the metabolic rate during the various activities is based on [34].

. Notably, there is no essential difference in the activity metabolic rate among the students of different grades, and there is no need to treat this factor differently during optimization.

Subsequently, it is crucial to ascertain the required daylighting threshold for children. As per reference [19], the suitable range of illuminance for children varies between 100 and 1000 lux. Furthermore, according to reference [47], the recommended upper limit of the glare index in classrooms, laboratories, and offices varies between 20 and 22. In this study we selected a value of 22 as the upper limit value of the glare index. Then, through field investigation, the occupant scheduling data in this case can be collected, as listed in

Table 4. Occupant scheduling data for the case kindergarten (weekdays).

Grades	Average Number of People in Each Class	Play Activities	Classes	Meals	Noon Break
1	18	8:00–9:00; 10:00–11:30;	9:00–10:00; 14:00–15:00	11:30–12:30	12:30–14:00
2	24	9:00–10:00; 15:00–16:00	8:00–9:00; 10:00–11:30; 14:00–15:00
3	24

. Weather data were obtained directly from the EnergyPlus website.

As stated in Section 2.1.4, the sole focus of this paper is the retrofitting of kindergarten building envelopes and energy systems. Hence, it is crucial to evaluate the performance of the building’s envelope and energy system before retrofitting in this case. Based on the original drawings, the performance parameters of the case building envelope can be obtained, as listed in

Table 5. Performance parameters of the case building envelope (before the retrofit).

Parameters	Main Materials and Thickness	Performance Indicator Value
External wall	Main material: concrete bricks + cement mortar + ceramic tile Total thickness: 240 mm	U value = 1.636
Floor	Main material: floated coat + concrete + plasterboard Total thickness: 150 mm	U value = 2.322
Roof	Main material: concrete + cement mortar + waterproof roll + artificial turf Total thickness: 300 mm	U value = 0.510
External window (sliding window)	Outside layer: clear_3 mm; middle layer: air, 6 mm; inside layer: clear, 3 mm	U value = 3.159; SHGC = 0.762; VT = 0.812

. The table indicates that the external walls and roofs of the building lack insulation materials, and the structure of the external windows is not rational. These factors were treated as variables in the subsequent design optimization. The artificial lighting equipment of the kindergarten was updated several years ago and energy-saving eye protection lights are now used, rendering it unnecessary to make any adjustments as part of the retrofit. The HVAC system currently uses split air conditioners, which have been in use for many years and include different equipment brands (purchased gradually), making it difficult to estimate their operational energy efficiency. Therefore, to improve the energy efficiency of the building, a VRV system with fresh air will be uniformly installed as part of the retrofit scheme.

3.3. Design Optimization of the Case Kindergarten Building

In Section 3.2, an investigation was conducted of the personnel, envelope, and equipment information for retrofitting the kindergarten building. Based on the acquired data, design optimization calculations are conducted in this section. In this paper, three optimization objectives are introduced in Section 2.2.1, namely, TC, UDI, and $C_{lc}$. Considering the particular context of the kindergarten under investigation, the optimization variables of this case study are presented in

Table 6. Optimization variables of the kindergarten building retrofit.

No.	Optimization Variables	Value Range
X1	External wall insulation	EPS board: 20–50 mm (interval: 5 mm)
X2	External roof insulation	EPS board: 30–70 mm (interval: 5 mm)
X3	Glass types of the different facade windows (double-paned window; filling gas: air, 6 mm)	Outer/Inner glass: clear, 3 mm; clear, 6 mm; bronze, 3 mm; bronze, 6 mm; gray, 3 mm; gray, 6 mm; green, 3 mm; green, 6 mm
X4	WWR of the different facades	South: 0.2–0.8 (interval: 0.1)
		North: 0.2–0.8 (interval: 0.1)
		East: 0.2–0.8 (interval: 0.1)
		West: 0.2–0.8 (interval: 0.1)
X5	Fixed sunshades for the windows in the facades	South (overhanging depth): 0–2.0 m (interval: 0.1 m)
X6	Cooling set point in the classrooms during each activity period	Play activities: 24 °C–30 °C (interval: 0.5 °C)
		Classes: 24 °C–30 °C (interval: 0.5 °C)
		Meals: 24 °C–30 °C (interval: 0.5 °C)
		Noon break: 24 °C–30 °C (interval: 0.5 °C)
X7	Heating set point in the classrooms during each activity period	Play activities: 16 °C–22 °C (interval: 0.5 °C)
		Classes: 16 °C–22 °C (interval: 0.5 °C)
		Meals: 16 °C–22 °C (interval: 0.5 °C)
		Noon break: 16 °C–22 °C (interval: 0.5 °C)
X8	Proportion of the roof photovoltaic area	0–1 (interval: 0.001)

.

As per Section 2.2.2, this study established a physical model using SketchUp, as depicted in Figure 6. Subsequently, the physical model is imported into EnergyPlus for parameter settings. The weather parameters, scheduling-related parameters, and thermal-comfort-related parameters are introduced in Section 3.2, while the material-related parameters (carbon-emission-related parameters) involved in this study are shown in

Table 7. Material-related parameters (carbon-emission-related parameters).

Parameters	Values
$K_{mEPS}$	5220 kg CO2/t
$K_{mconcrete}$	135 kg CO2/t
$K_{mglassclear}$	500 kg CO2/t
$K_{mglassbronze}$	600 kg CO2/t
$K_{mglassgrey}$	600 kg CO2/t
$K_{mglassgreen}$	600 kg CO2/t
$K_{mphotovoltaic}$	158 kg CO2/m2
$K_d$	0.59 kgCO2/kWh
$Y$	50
${cof}_r$	0.1
${cof}_{EPS}$
${cof}_{concrete}$
${cof}_{glass}$	0.3
${cof}_{photovoltaic}$	0.3

Note: This study assumes that the service life of wall-related materials is 50 years, while the service life of windows and photovoltaic-related materials is only 25 years.

. Moreover, to ensure the rationality of parameter settings related to daylighting comfort, this study used a classroom as an example to verify the grid independence of the illuminance sensor.

Upon completing the parameter settings, the Python programming language can be utilized to construct the rapid comprehensive optimization framework, as discussed in Section 2.2.3. The optimal ANN model can be generated first, as presented in

Table 8. Hyperparameter settings of the ANN model.

Hyperparameters	Optimal Values
Sample size	1700
Solver	lbfgs
Activation function	tanh
Hidden layer sizes	(18, 12, 6)
Hidden layer numbers	3
Initial learning rate	0.1
Maximum number of iterations	20,000
MRE (%)	0.98
R2	0.991

. When the sample size reaches 1700, the mean relative error of the optimized ANN model satisfies the requirement (≤1%). Compared to references [10,44], this method significantly reduces computational costs and does not necessitate generating a vast sample set before conducting metamodel training.

On this basis, the optimal ANN model is combined with the NSGA-III to perform a three-objective optimization calculation (with a population of 50), and the outcomes are demonstrated in Figure 7. It can be seen that there are a total of 50 Pareto solutions in the Pareto solution set, among which the range of $C_{lc}$ is −77,336.1–785,327.2 kg; the range of $TC$ is 38.5–66.4%; and the range of $UDI$ is 1.2–93.3%. As all the aforementioned solutions are nondominated, a multi-attribute decision-making method is necessary to further select the optimal design scheme. Based on the relevant content in Section 2.2.3, the comprehensive optimal scheme attained in this research is presented in

Table 9. Comprehensive optimal design scheme for kindergarten building retrofit.

No.	Optimization Variables	Comprehensive Optimal Scheme	Benchmark Scheme
X1	External insulation of wall	EPS board: 30 mm	EPS board: 30 mm
		(Total U value of the wall = 0.670)	(Total U value of the wall = 0.670)
X2	External insulation of roof	EPS board: 30 mm	EPS board: 30 mm
		(Total U value of the roof = 0.362)	(Total U value of the roof = 0.362)
X3	Glass types of different facade windows (double-paned window; filling gas: Air_6 mm)	Outer glass: bronze, 6 mm;	Outer glass: clear, 3 mm;
		Inner glass: gray, 6 mm	Inner glass: clear, 3 mm
		(U value = 3.111; SHGC = 0.442; VT = 0.526)	(U value = 3.159; SHGC = 0.762; VT = 0.812)
X4	WWR of different facades	South: 0.4	South: 0.6
		North: 0.4	North: 0.6
		East: 0.2	East: 0.4
		West: 0.2	West: 0.4
X5	Fixed sunshades for windows on south facade	Overhanging depth: 2.0 m	Overhanging depth: 2.0 m
X6	Cooling set point in classrooms of each activity period	Play activities: 25.5 °C	26 °C
		Classes: 30 °C
		Meals: 25.5 °C
		Noon break: 30 °C
X7	Heating set point in classrooms of each activity period	Play activities: 16.5 °C	18 °C
		Classes: 22 °C
		Meals: 16.5 °C
		Noon break: 17.5 °C
X8	Proportion of the roof photovoltaic area	1	1
	$TC$	41.2%	38.5%
	$UDI$	92.6%	12.9%
	$TES$	−4205.89 kWh	−2155.73 kWh
	$C_o$	−2481.48 kg	−1271.88 kg
	$C_{lc}$	−9647.4 kg	24,510.9 kg

and Figure 8. Additionally, considering that most of the retrofit schemes for buildings in actual engineering are designed according to codes, this study also generated the benchmark scheme based on the code [48] (as shown in Table 9) and compared it with the comprehensive optimal scheme.

Upon comparing the Pareto scheme set obtained from three-objective optimization and the benchmark scheme, it is evident that three-objective optimization can achieve a maximum carbon emission reduction of 101,847 kg (i.e., 77,336.1 kg of carbon emissions can be saved for the region on the basis of achieving zero emissions during the life cycle). Moreover, it can increase the thermal comfort period by up to 27.9% and the daylighting comfort period by up to 80.4%. Upon comparing the comprehensive optimal scheme and the benchmark scheme, it is evident that the annual operating carbon emissions in both schemes are negative. Notably, the annual production energy is higher than the energy consumption. However, the comprehensive optimal scheme can also achieve savings of 9647.4 kg of carbon emissions based on achieving zero emissions during the life cycle. Additionally, it can facilitate increases in the thermal comfort period of 2.7% and the daylighting comfort period of 79.7%.

4. Results and Discussion

4.1. Analysis of the Research Results

The above case study indicates that the approach adopted in this study for setting the cooling/heating temperature set points during different periods is rational and better aligned with the thermal comfort requirements of children in the kindergarten building. While increasing the thermal comfort period, it can also yield a reduction in the energy consumption (i.e., reducing operating carbon emissions). Additionally, the current design codes fail to consider the differences in daylighting comfort between children and adults. This leads to the illuminance values on the working surface of the classroom often exceeding the comfort range of children, resulting in a low $UDI$ value in the benchmark scheme. However, by optimizing the window glass type, WWR, and shading, it is possible to effectively address this issue.

In this paper, we examine life-cycle decarbonization retrofitting of comfortable kindergarten buildings in China (considering children in particular). Through multi-objective optimization combined with multi-attribute decision-making, a balance is achieved between thermal comfort, daylighting comfort, and life-cycle decarbonization. In traditional design concepts, maintaining a comfortable indoor environment often requires high energy consumption. However, we found that, while fully considering the metabolic rate and clothing insulation of children during different periods, the comfort of children and building decarbonization are not completely opposite goals, and can be balanced within a certain range (as demonstrated by the comparison between the comprehensive optimal scheme and the benchmark scheme in Table 9).

In addition, the parallel computing framework proposed in this study can significantly reduce the sample set required for meta-model training, thereby significantly improving the efficiency of generating high-precision meta-models.

4.2. Comparison with Similar Studies

The optimization of educational buildings has always been a hot topic in the field of building research. In reference [49], the impact of an egg-crate shading device on the indoor environmental quality of classrooms in southern Spain was studied. The results indicated that the implementation of the egg-crate shading device resulted in improvements in the operative temperature and illuminance in the classroom. In reference [50], the adaptive thermal comfort of students aged 10–11 years in summer classrooms in the Seville region was studied. The researchers compared the PMV and thermal sensation vote (TSV) results, and the results showed that the PMV model is not suitable for adaptive thermal comfort evaluation of children aged 10–11 in the Seville region. In reference [7], three-objective optimization of primary school classrooms in Tehran was studied, aiming to simultaneously improve the energy, thermal comfort, and daylighting performance of primary school classrooms. In reference [51], design optimization of natural ventilation in educational buildings was studied. In the research process, a multi-objective optimization algorithm was used to achieve the optimal balance between the indoor air quality, thermal comfort, energy consumption, and life-cycle cost.

Compared with the above studies, the innovation of this study encompasses: (1) proposing a design optimization method that comprehensively considers children’s thermal comfort, daylighting comfort, and life-cycle decarbonization; (2) proposing a rapid design optimization framework based on parallel computing; and (3) exploring the correlation between the thermal comfort of children in the classroom and building decarbonization on the basis of considering their metabolic rate and clothing insulation. Overall, this study has important reference value for thermal comfort, daylighting, and decarbonization optimization of kindergarten buildings in various regions

4.3. Analysis of Research Limitations and Prospects for Future Directions

This study is not without limitations. Firstly, in the simulation process, we utilized annual data generated from meteorological parameters in 2002 to account for the inherent uncertainty in these parameters. However, these simulated data deviate from actual weather data, which may introduce a certain degree of deviation in the practical application of the optimization results. Secondly, as the building in this case study is still undergoing renovation, actual test data from the renovated building were not obtainable for this research. This lack of data precluded further verification of the reliability of the simulation results. Future research will entail following up on the progress of this project to obtain such data. Thirdly, the focus of this study was on the overall optimization effect of buildings. Therefore, in the simulation calculation process for thermal comfort, it was assumed that the distributions of wind speed, temperature, and humidity within the classroom were uniform. In reality, however, there are variations in the distribution of thermal comfort within classrooms.

Furthermore, it is important to note that children are in a stage of rapid growth and development, with significant physiological differences among different age groups, and correspondingly, distinct needs regarding thermal and light environments. Unfortunately, existing design codes lack relevant content on daylighting comfort for children and do not account for the differences in the range of daylighting comfort between children and adults. Given that moderate daylighting is beneficial for stimulating the development of children’s visual nerves, further research is necessary to determine the appropriate range of illuminance for children of different ages. Additionally, climate change will inevitably impact the indoor environmental comfort and operational energy consumption of buildings. Therefore, subsequent research could further investigate the impact mechanism of climate uncertainty on various types of educational buildings and quantify the degree of uncertainty. This would facilitate the development of optimization methods for educational building design that consider climate uncertainty.

5. Conclusions

In this paper, we propose an optimization methodology for retrofitting kindergarten buildings, considering thermal comfort, daylighting, and life-cycle carbon emissions. After conducting a comprehensive investigation of the parameters pertaining to children’s comfort, we calculated relevant indicators of thermal and daylighting comfort, ensuring that the retrofitted kindergarten building promotes optimal child growth. In the course of our research, necessary retrofit parameters were obtained through a combination of video playback, field investigation, literature review, and examination of drawings. Based on this rigorous approach, optimization variables were selected, and a physical model was established. Ultimately, a rapid comprehensive optimization framework was employed for optimization calculations to derive the optimal scheme for kindergarten buildings. By utilizing a typical kindergarten building in Nanjing as a case study, we validated the effectiveness of the proposed method. The results revealed that: (1) in comparison to the benchmark scheme, multi-objective optimization could reduce carbon emissions by up to 101,847 kg, increase the thermal comfort period by up to 27.9%, and enhance the daylighting comfort period by up to 80.4%. (2) When compared to the benchmark scheme, the comprehensive optimal scheme resulted in a reduction in carbon emissions of 34,158.3 kg, an extension of the thermal comfort period by 2.7%, and an improvement in daylighting comfort by 79.7%. (3) The determined cooling/heating temperature set points are highly suitable for kindergarten buildings, as they fulfill children’s thermal comfort needs while also mitigating carbon emissions.

Compared with other similar studies, the main innovation of this research lies in: (1) proposing a design optimization method that comprehensively considers children’s thermal comfort, daylighting comfort, and life-cycle decarbonization; (2) proposing a rapid design optimization framework based on parallel computing; and (3) exploring the correlation between the thermal comfort of children in the classroom and building decarbonization on the basis of considering their metabolic rate and clothing insulation.