*Article* **Mediating Factors Explaining the Associations between Solid Fuel Use and Self-Rated Health among Chinese Adults 65 Years and Older: A Structural Equation Modeling Approach**

**Qiutong Yu 1,2, Yuqing Cheng 1,2, Wei Li 1,2 and Genyong Zuo 1,2,\***


**Abstract:** Exposure to indoor air pollution from cooking with solid fuel has been linked with the health of elderly people, although the pathway to their association is unclear. This study aimed to investigate the mediating effects between solid fuel use and self-rated health by using structural equation modeling (SEM) with the baseline data from Chinese Longitudinal Healthy Longevity Survey (CLHLS). We conducted a cross-sectional survey among 7831 elderly people aged >65 years from the CLHLS. SEM was used to analyze the pathways underlying solid fuel use and self-rated health. We estimated indirect effects of sleep quality (β = −0.027, SE = 0.006), cognitive abilities (β = −0.006, SE = 0.002), depressive symptoms (β = −0.066, SE = 0.007), systolic blood pressure (β = 0.000, SE = 0.000), and BMI (β = −0.000, SE = 0.000) on the association between solid fuel and the self-rated health using path analysis. Depressive symptoms emerged as the strongest mediator in the relationship between solid fuel use and self-rated health in the elderly. Interventions targeting sleep quality, cognitive abilities, depressive symptoms, systolic blood pressure, and BMI could greatly reduce the negative effects of solid fuel use on the health of the elderly population.

**Keywords:** household air pollution; solid fuel; self-rated health; structural equation modeling; elderly

## **1. Introduction**

The aging population in China has grown rapidly in the past few decades. According to the report of the seventh census in 2020, the population of those aged >60 years was 264.02 million, accounting for 18.70% of the total population, while the population of those aged >65 years was 190.64 million, accounting for 13.50% [1]. The Chinese population is approaching the depth of the aging stage. With the deepening of social aging in China, more attention should be paid to both physical and mental health problems in elderly people.

Approximately 490 million people in China are exposed to indoor air pollution from cooking with solid fuel, such as coal, charcoal, and wood [2]. The particulate matters produced by burning solid fuel, such as PM2.5, PM10, carbon monoxide, nitrogen dioxide, sulfur dioxide, or other volatile organic compounds [3], have a negative impact on the physical or mental health of elderly people [4–6]. Therefore, it is necessary to explore the effect of indoor air pollution caused by using solid fuel on the health of elderly people.

Previous studies have shown that the use of solid fuel seriously affected both the mental and the physical health of the elderly [5,7–11]. Many studies on middle-aged and elderly people have concluded that solid fuel use was significantly correlated with the health of elderly people, in terms of poor sleep quality [7], low cognitive function [8], high incidence of arthritis [12], high incidence of depression [5], and high incidence of hypertension [9,10]. In pathogenesis research, there is an increasing link between indoor air

**Citation:** Yu, Q.; Cheng, Y.; Li, W.; Zuo, G. Mediating Factors Explaining the Associations between Solid Fuel Use and Self-Rated Health among Chinese Adults 65 Years and Older: A Structural Equation Modeling Approach. *Int. J. Environ. Res. Public Health* **2022**, *19*, 6904. https:// doi.org/10.3390/ijerph19116904

Academic Editors: Ashok Kumar, M Amirul I Khan, Alejandro Moreno Rangel and Michał Piasecki

Received: 1 May 2022 Accepted: 2 June 2022 Published: 5 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

pollution and physical diseases from solid fuel [13]. The underlying mechanism is that solid fuel in the combustion chamber produces toxic volatile organic compounds (VOCs), which can easily turn into vapors and are involved in metabolic processes that lead to low cognitive function or an increased blood pressure [8,14,15]. However, all the abovementioned effects are single indicators of health measurement. Self-rated health is a comprehensive measurement indicator of health that can reflect the respondents' physiological state, their knowledge of this state, and their health expectations [16,17]. However, the association of solid fuel use with self-rated health in elderly people is unclear. Moreover, evidence on the effect of solid fuel use on the self-rated health of elderly people directly and indirectly through multiple mediators and the distinctive pathways, particularly in China, is lacking.

The available evidence suggests that people with long-term exposure to indoor air pollution from cooking with solid fuel are more likely to have poor sleep quality [7], high risk of depressive symptoms [5], low cognitive function [8], high blood pressure [9,10], and low body mass index (BMI) [11], which might result in poor self-rated health as mediators in the elderly. Therefore, structural equation modeling (SEM) was used to evaluate the total, direct, and indirect effects of exposure to indoor air pollution from cooking with solid fuel on the self-rated health in a mediation analysis and to assess the indirect effect within these distinctive paths.

This study aimed to investigate the mediating effects between solid fuel use and self-rated health using SEM with the baseline data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS). Changing the fuel for cooking is a long-term project and may be costly, which may impose a huge financial burden on developing countries such as China. However, we can change the mediating factors to minimize the danger of solid fuel use to the health of the elderly, which could be of great help to achieve healthy aging in countries that are evolving into an aging society.

#### **2. Materials and Methods**

#### *2.1. Setting and Participants*

We used secondary data derived from the 2018 CLHLS, which has been a cohort project since 1998, to conduct a longitudinal population-based study of people aged >65 years in China. We used the 2018 CLHLS because it contained the latest data which best fit the current situation of China's aging society. Using the multistage stratified proportional probability sampling design, approximately 16,000 elderly people in urban and rural communities were randomly selected from 500 sample areas in 23 provinces, and 15,874 people were interviewed successfully. The biomedical ethics committee of Peking University approved the study, and all study participants signed an informed consent form. After excluding 95 participants who were younger than 65 years, 45 participants who "never cook", 51 participants who had technical problems, 2409 participants who refused to answer, 1103 participants who answered "not applicable", and 4340 participants who were unable to provide the data, 7831 individuals were finally included in the study (Figure 1).

#### *2.2. Outcome Variables*

Respondents were asked, "how do you feel about your health?", and, according to the five-point Likert scale, their health status was rated as "very good", "good", "fair", "poor", or "very poor". Self-rated health status has been identified as a reliable predictor of health and has been widely used in previous health studies conducted in China [18].

#### *2.3. Exposure Variables*

Respondents were asked, "what is the main source of cooking fuel in your family?", and those who answered "other" were excluded. We defined coal, charcoal, and wood as solid fuel, whereas solar energy, natural gas, induction cooker, and other electrical appliances were defined as clean fuel.

**Figure 1.** Study flowchart of participant selection (aged ≥65 years) from the Chinese Longitudinal Healthy Longevity Survey 2018 survey data. Abbreviation: CLHLS: Chinese Longitudinal Healthy Longevity Survey.

#### *2.4. Mediators*

Previous studies [5,7–11] have found that solid fuel use affects sleep quality, cognition abilities, depressive symptoms, blood pressure, and BMI. Respondents were asked "how is your sleep quality?" according to a five-point Likert scale, and sleep quality was rated as "very good", "good", "moderate", "poor", or "very poor". Blood pressure was the average of two measurements on the right arm of participants using a mercury sphygmomanometer after the participants had rested for 5 min. Body mass index was calculated as the weight in kilograms divided by the height in meters squared (kg/m2).

Cognitive function was measured using the Chinese version of MMSE (the Mini Mental State Examination) [19]. The MMSE has been validated in previous studies for the Chinese elderly [20,21]. The correct answers were encoded as 1, while incorrect answers or "inapplicable" answers were encoded as 0. Then, we summed the cognitive scores of each participant. Higher cognitive scores indicated better cognitive function.

Depression was measured using the questions from the 10-item Center for Epidemiologic Studies Short Depression Scale (CES-D scale) [22], which has been translated and validated in previous studies for assessing cognitive levels in the Chinese elderly [23]. There are five levels of answers to all questions; we reverse-coded the positively oriented questions and recoded all responses as follows: "always" as 5, "often" as 4, "sometimes" as 3, "seldom" as 2, and "never" as 1. The depression scores of the participants were summed. Higher depression scores indicated greater depression severity.

#### *2.5. Covariates*

Notably, many studies have estimated the modification effects by sociodemographic factors and lifestyle behaviors when exploring the effects of household air pollution on the mediators in this paper [24,25]. Demographic characteristics including age (years), sex (female/male), and marital status (not married and married) were analyzed. The "not married" status included widowed, divorced, separated, or never married. Community location (urban/rural) and socioeconomic status, including education and family income, were also analyzed. Education was recoded into three levels (0 years/1–6 years/<6 years). Family income was recoded into three levels (>10,000 CNY, 10,000–50,000 CNY, and <50,000 CNY). Lifestyle behaviors including smoking status (not current/current), alcohol use (not current/current), and regular exercise (not current/current) were also evaluated [26].

#### *2.6. Statistical Analysis*

The chi-square test was used for dichotomous variables, including sex, marital status, community location, education years, household income, smoking status, alcohol use, physical activity, sleep quality, and self-rated health, and the *t*-test was used for continuous variables, including cognitive abilities, depressive symptoms, systolic blood pressure, and BMI between those who cook with solid fuel and those who cook with clean fuel.

Structural equation modeling (SEM) is a multivariate analytic technique. It is used to simultaneously assess multiple relationships among variables. SEM was used to conduct a formal mediation test and disaggregate the relationship between solid fuel use and self-rated health through causally defined indirect and direct pathways. SEM contains a series of multiple regression models, linear regression models for continuous outcomes, and logistic regression models for binary outcomes. The proportion of the total effect of solid fuel on self-rated health attributable to the mediators was calculated by dividing the ratio of the indirect effect through the mediated pathway by the ratio of the total effect. Estimation for SEM was performed using maximum likelihood. Three common measures were used to evaluate the fit indices of SEM: the root-mean-square error of approximation (RMSEA), comparative fit index (CFI), and Tucker–Lewis index (TLI). TLI and CFI values of 0.95 indicate a reasonably good fit [27]. An RMSEA value of 0.05–0.08 represents a moderate fit, while a value of 0.08–0.10 represents an acceptable fit [28]. All data were analyzed using Stata15.0.

#### **3. Results**

#### *3.1. Basic Characteristics of the Participants*

The characteristics of the study participants are shown in Table 1. Of the 7831 individuals, 2317 (29.59%) used solid fuel for cooking, and 5514 (70.41%) used clean fuel for cooking. Significant differences in sex, marital status, community location, education years, income, smoking status, alcohol use, physical activity, sleep quality, cognitive abilities, depressive symptoms, hypertension, BMI, and self-rated health were observed between individuals using solid fuel and those using clean fuel for cooking (*p* < 0.05). Participants who used clean fuel had higher socioeconomic indicators, in terms of both years of education and family income, than those using indoor solid fuel. Moreover, the proportions of individuals with current smoking (19.25% vs. 16.00%) and drinking (17.22% vs. 15.43%) habits were higher among individuals using solid fuel for cooking than among those using clean fuel. However, the proportion of individuals who engaged in physical activities was significantly higher among those using clean fuel for cooking (42.00% vs. 24.08%) than among those using solid fuel. Moreover, individuals using clean fuel for cooking reported better sleep quality, cognitive abilities, and self-rated health. Among those using solid fuel for cooking, the average systolic blood pressure was 141.07 mmHg.


**Table 1.** Characteristics of selected variables among the participants.

<sup>1</sup> SBP, systolic blood pressure; BMI, body mass index; SD, standard deviation.

#### *3.2. Structural Equation Model*

Structural equation modeling is a multivariate analytic technique used to simultaneously assess multiple relationships among variables. As shown in Figure 2, we performed SEM with a good fit to the data (RMSEA = 0.045, CFI = 0.970, TLI = 0.829), showing that the model fit quite well after adjusting for age, sex, marital status, community location, education years, household income, smoking status, alcohol use, and physical activity. To determine the extent of the impact of solid fuel, sleep quality, cognitive abilities, depressive symptoms, systolic blood pressure, and BMI on the self-rated health of the elderly people in the path model, a standardized path coefficient of the SEM was estimated (Table 2). The total indirect effect of solid fuel use for self-rated health was −0.146 (*p* < 0.001). A

significant direct effect of the use of solid fuel on self-rated health (β = −0.041, S.E. = 0.020), with indirect effects accounting for most of the total effects, was identified.

**Figure 2.** Pathways between solid fuel, mediators, and self-rated health, according to the Chinese Longitudinal Healthy Longevity Survey, 2018. Structural equation modeling was performed among Chinese older adults over 65 years old. The model was adjusted for age, sex, marital status, community location, education years, household income, smoking status, alcohol use, and physical exercise. Dashed lines denote insignificant pathways between solid fuel, mediators, and self-rated health, while solid lines denote significant pathways between solid fuel, mediators, and self-rated health.

**Table 2.** Direct, indirect, and total effects of solid fuel use on self-rated health.


<sup>1</sup> β, coefficient; SE, standard error.

Table 2 shows that sleep quality (β = 0.250), cognitive ability (β = −0.010), and depression (β = −0.049) had a direct influence on self-rated health (*p* < 0.005). The result from SEM indicated that using solid fuel exhibited a direct effect on sleep quality (β = −0.056), cognitive ability (β = −0.319), depression (β=0.910), and BMI (β = −0.733). However, we did not find that systolic blood pressure (*p* = 0.876) and BMI (*p* = 0.876) were significantly associated

with self-rated health, nor did we find that solid fuel was significantly associated with systolic blood pressure. Moreover, SEM is used either to assess the total effect (i.e., direct and indirect effects) of a treatment or exposure on an outcome in the mediation analysis or to assess a specific indirect effect with those complex paths. First, a significant negative indirect effect of solid fuel use on self-rated health via sleep quality was observed (β = −0.013, SE = 0.006). Second, cognitive abilities were also found to be a mediator between solid fuel use and self-rated health (β = −0.003, SE = 0.001). Third, a significant negative indirect effect of solid fuel on self-rated health via depressive symptoms (β = −0.045, SE = 0.007) was also observed. Additionally, the indirect effects of systolic blood pressure on self-rated health, as well as those of body mass index on self-rated health, were not detected in the model involving solid fuel exposure and self-rated health (all *p* > 0.05).

#### **4. Discussion**

#### *4.1. Main Findings*

In this cross-sectional study, we focused on assessing the potential mediating factors for the relationships between solid fuel use and self-rated health. Exposure to solid fuel was found to have a direct contribution to the decreased score of self-rated health. Moreover, we observed that exposure to solid fuel was significantly associated with a decreased score of self-rated health, and this linkage was mediated by sleep quality, cognitive ability, and depression symptoms. These effects remained significant even after controlling for confounders such as sociodemographic factors and lifestyle behaviors.

#### *4.2. Available Evidence on the Association of Solid Fuel with Self-Rated Health*

We found that the direct effect of solid fuel use on self-rated health was −0.044 (*p* < 0.001). Previous studies have demonstrated that solid fuels can affect physical health measures such as blood pressure, BMI, cognition, and mental health measures such as sleep quality and depression status. However, the dependent variables used in these studies were all single health indicators evaluating a specific aspect of the health of an individual. Nevertheless, by constructing a structural equation model, we found that solid fuel can affect the above single health indicators, thereby affecting the comprehensive indicators of personal health, because self-rated health can reflect people's physical and mental health. Therefore, screening and management of these disorders in older adults with heavy use of solid fuels is necessary.

#### *4.3. Depression Was the Strongest Mediator of the Relationship between Solid Fuel Use and Self-Rated Health*

The results showed that individuals using solid fuel had greater depression severity (β = 0.893), and depression symptoms emerged as the strongest mediator of the relationship between solid fuel and self-rated health rather than sleep quality and cognitive ability. However, we only found three similar studies exploring the association between solid fuel and depression whose findings were consistent with our results that individuals using solid fuel were at a higher risk of depression [5,29,30]. Moreover, regarding household air pollution, there is an increasing link between mental diseases and household air pollutants. This study contributes to the limited literature on the association between solid fuel and poor sleep quality in the elderly. Our results indicated that using solid fuel was significantly associated with poor sleep quality, even after accounting for a wide range of covariates, including sociodemographic factors, lifestyle behaviors, and presence of chronic diseases. Our findings were in line with previous findings [31,32]. However, limited evidence is available on the association between mental diseases and household air pollutants in pathogenesis research. One possible reason is that solid fuel combustion produces much higher levels of various gaseous pollutants than clean fuel and may increase the risk of developing mental disorders such as depression and poor sleep quality through pathways such as cerebrovascular damage, oxidative stress, neuroinflammation, or neurodegeneration [33,34].

#### *4.4. Cognition as the Protective Factor Linking Systolic Blood Pressure to Self-Rated Health*

An indirect path linking solid fuel and self-rated health was identified through cognitive abilities, although the other mediators including systolic blood pressure and BMI were not found to be associated with self-rated health. Our findings suggested that people who use solid fuel for cooking had lower cognitive abilities (β = −0.319, SE = 0.111). Several studies in China contributed to the literature on the association between solid fuel and poor cognitive ability in the elderly. Recent evidence from a study in China found that solid fuel use was significantly associated with mental health and cognitive ability in middle-aged and older adults [8]. There were also studies that further showed the potentially harmful effects of household air pollution exposure on other aspect of memory. A follow-up study showed that solid fuel use was associated with a greater decline in cognitive score, mostly in the episodic memory and visuo-construction dimensions [35]. Another prospective analysis found a significant adverse impact on cognitive abilities, especially short-term memory and mathematical reasoning. These results demonstrated that using solid fuel poses a health threat to elderly people [36].

#### *4.5. Strengths and Limitations*

This study had several strengths. Firstly, this study specified the pathways of the relationship between solid fuel use and self-rated health of the elderly. Secondly, self-rated health was not only considered a comprehensive measurement indicator of health, but also used as a predictor of morbidity and mortality [37]. This study's results can help achieve primary and secondary prevention and improve the health of the elderly by stopping the use of solid fuel. Lastly, our study included 500 sample areas in 23 provinces in China, which gives our findings strong external validity for the Chinese society. Moreover, the findings were robust across the study regions, demographic characteristics, and lifestyle behaviors.

However, a major drawback should be noted. The assessment of self-rated health in our study was based on only one self-reported question, and this may not provide exactly accurate information compared with a structured interview tool. Other limitations should also be mentioned. Firstly, constrained by the CLHLS data, we were unable to derive information on whether the individuals were responsible for cooking, types of cooking in childhood, and time spent cooking with indoor solid fuel; additionally, indoor air pollution exposure may vary by family and personal characteristics [38]. Secondly, the results of this cross-sectional study may not explain the underlying mechanisms of the relationship between solid fuel use and self-rated health. The underlying mechanisms may need to be investigated in large prospective cohort studies. Thirdly, the relationship of mediators including sleep quality, cognitive ability, depression, systolic blood pressure, and BMI with solid fuel use has been confirmed in previous studies. Other diseases or pathophysiological indicators that may be caused by the use of solid fuel, which have not been proven, can be explored in future studies.

#### **5. Conclusions**

Sleep quality, cognitive abilities, and depressive symptoms partially contributed to the association between solid fuel use and self-rated health. However, systolic blood pressure and BMI were not found to be directly associated with self-rated health. Among these mediators, depression was the strongest mediator of the relationship between solid fuel use and self-rated health. Our results demonstrated that using solid fuel poses a health threat for elderly people. Replacing solid fuel with clean fuel may be an important way to improve self-rated health of elderly people. Regarding this, priority should be giving to those with significant solid fuel exposure.

**Author Contributions:** Q.Y., writing—original draft; Y.C., data curation; W.L., formal analysis; G.Z., writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (No. 71774102).

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethical Review Committee of Peking University (IRB00001052–13074).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent was obtained from the patients to publish this paper.

**Data Availability Statement:** Data are available on the open research data service platform of Peking University. Data for this study were sourced from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) and are available at https://opendata.pku.edu.cn (accessed on 1 January 2022).

**Acknowledgments:** The authors thank the Chinese Longitudinal Healthy Longevity Survey team for providing data.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Investigation of Airflow Distribution and Contamination Control with Different Schemes in an Operating Room**

**Fujen Wang 1,\*, Indra Permana 2, Dibakar Rakshit <sup>3</sup> and Bowo Yuli Prasetyo <sup>4</sup>**


**Abstract:** Controlling contamination via proper airflow distribution in an operating room becomes vital to ensure the reliable surgery process. The heating, ventilation, and air conditioning (HVAC) systems significantly influence the operating room environment, including temperature, relative humidity, pressurization, particle counts, filtration, and ventilation rate. A full-scale operating room has been investigated extensively through field measurements and numerical analyses. Computational fluid dynamics (CFD) simulation was conducted and verified with the field measurement data. The simulation was analyzed with three different operating room schemes, including at-rest conditions (case 1), normal operational conditions with personnel (case 2), and actual conditions with personnel inside and some medical equipment blocking the return air (case 3). The concentration decay method was used to evaluate this study. The results revealed that the contamination concentration in case 1 could be diluted quickly with the average value of 404 ppm, whereas the concentration in case 2 slightly increased while performing a surgery with the average value of 420 ppm. The return air grilles in case 3, blocked by obstacles from some medical equipment, resulted in the average concentration value of 474 ppm. Other than that, the contaminant dilution could be obstructed dramatically, which revealed that proper and smooth airflow distribution is essential for contamination control. The ventilation efficiency of case 2 and case 3 dropped around 6% and 17.91% compared to case 1 in the unoccupied and ideal condition. Ventilation efficiency also decreased along with decreasing the air change rate per hour (ACH), while with increasing ACH, the ventilation efficiency in case 3 actually increased, approaching case 2 in the ideal condition.

**Keywords:** operating room; airflow distribution; contamination control; field measurement; computational fluid dynamics

### **1. Introduction**

The critical area of any hospital is the operating room. Anything in the operating room can endanger a patient's life, such as a variety of bacteria and viruses [1]. Those contaminants can be transmitted through the air that can contaminate medical tools. This could be dangerous to a patient when the operating room staff members perform the procedure in the operating room [2]. HVAC systems provide comfort and sufficient quality air for patients and staff in the operating room. A comfortable and healthy environment is generally determined by temperature, humidity, and air velocity [3]. Therefore, in maintaining a clean and healthy environment for patients and healthcare workers, thermal comfort and indoor air quality requires a valid regulation. American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) standard 170 [4] mentions

**Citation:** Wang, F.; Permana, I.; Rakshit, D.; Prasetyo, B.Y. Investigation of Airflow Distribution and Contamination Control with Different Schemes in an Operating Room. *Atmosphere* **2021**, *12*, 1639. https://doi.org/10.3390/ atmos12121639

Academic Editors: Ashok Kumar, M. Amirul I. Khan, Alejandro Moreno Rangel and Michał Piasecki

Received: 23 November 2021 Accepted: 5 December 2021 Published: 8 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

directive notices in an operating room, including the number of air changes per hour, airflow distribution, room pressurization, and filtration. ASHRAE standard 62 [5] asserts that indoor pollutants can influence occupants' activities; carbon dioxide is an example of an indoor pollutant. The total carbon dioxide exposure in the room must be less than 1000 ppm: this will give health workers a comfortable and healthy condition to stay focused on the procedure. Shortness of breath, headaches, confusion, and other symptoms can occur in the room because of the high exposure to carbon dioxide.

Additionally, researchers have conducted many studies on changing the velocity of supply air to control contamination spread in the operating room. The relatively low speed in the operating room will affect the concentration of microbial carrier particles (MCP) and the room's deposition rate. This suggests that there may be a risk of microbiological contamination from exposed surfaces to areas of low velocities, such as under lights during surgical procedures [6]. Uniform vertical laminar airflow is established, and high cleanliness is achieved in the center of the room when the surgical lamp is arranged in two axes [7]. In addition, the movement of the particles in the operating room can be affected by the position of the surgical lightings, and it was also shown that higher supply velocity (≥0.38 m/s) might affect the flow disturbance [8]. Current evidence has shown a positive relationship between the airborne concentration of bacteria-carrying particles (BCPs) in the operating room and the rate of infections. The accumulation of airborne BCPs under operating lighting poses a high risk of infection for patient safety [9]. In another study, four different supply air velocities (0.16, 0.24, 0.29, and 0.33 m/s), were investigated by Liu et al. [10]. A higher cleanliness level in the operating room can be ensured by supplying air velocities larger than 0.24 m/s. Meanwhile, when the supply air velocity increases to larger than 0.33 m/s, this will also increase bioaerosol deposition.

The air supply velocity must be optimally designed to match the energy consumption for energy saving. The HVAC system in the operating room is operated for 24 h throughout the year with intensive energy consumption [11]. The HVAC system in the operating room is operated under full load even when the room is unoccupied. Research has studied the operating room ventilation systems' best practice for energy efficiency, health, and safety [12]. Proper design, operation, and controls can reduce these costs by as much as 65% while ensuring a healthy and safe environment for the surgical team and the patient. Concerning energy saving, a preliminary study of numerical analysis has been carried out to evaluate the air velocity distribution and concentration contours while carrying out a ACH approach in an unoccupied operating room [13].

A variety of ventilation schemes have been developed for operating room use. Each has pros and cons and may be better suited than another for operations under certain conditions. The proper functioning of OR ventilation is also affected by external and internal disruptions. By applying CFD, the present study investigates the airflow and contaminant distribution in operating rooms under different conditions [14]. CFD simulation uses field measurements that are carried out as the pedestal parameters as boundary conditions [15]. In this study, CFD simulations were performed to discover the potential of HVAC systems to control air contamination, a comfortable environment for occupants, and the possibilities of energy-efficient approaches in the operating room. This simulation is based and verified on field data collection. Other than that, CFD simulation methods were also conducted in other research fields to predict and evaluate some systems in low cost and efficient ways. Zhiyi et al. [16] investigated ventilation performance in typical apartment buildings predicted by CFD in a multi-zone airflow model. The improvement in indoor air quality (IAQ) was also conducted by the measurement and CFD simulations that are shown as valid tools for IAQ indication [17]. In addition, CFD modeling of contaminant migration in a household gas furnace was investigated by Szczepanik-Scislo [18]. The results revealed that the location of the furnace could influence contaminant accumulation and migration. Such simulations can be an essential tool when designing a ventilation system concerning a furnace to improve the removal of dangerous substances.

With the intention of achieving a good environmental condition, the concentration decay method can be used to assess indoor ventilation efficiency. Tracer gas or particle experiments could be used in CFD methods. However, the pathogens as particles could be simplified without considering their biological characteristics, as most researchers have investigated [19]. Specific piecewise-linear techniques were applied to the concentration decay method to determine the ACH values for smaller time intervals. This is necessary because the plotted semi-logarithmic decay curve itself is not linear as the ventilation rate changes with time due to the changing buoyancy force [20]. Chung [21] used carbon dioxide (CO2) as a pollutant to evaluate the efficiency of indoor ventilation. The tracer gas was injected into the room and mixed into the air; a decrease in tracer gas concentration was logged over a given period [22]. This study investigates the contamination control in different schemes to discover performance improvements carried out through comprehensive field measurement tests as well as numerical simulation analysis.

#### **2. System Description**

The operating room is generally categorized as a positively pressurized bio-cleanroom that ensures a critical environment for infection control concerns to comply with applicable standards and regulations. The dimension of the investigated operating room was at a length of 6.3 m, a width of 6.0 m, a height of 3.0 m, and a total area of 37.8 m2. The function of this operating room was heart surgery. High-efficiency particulate air (HEPA) filtered the supply air with a total of 15 units located in the center of the ceiling. The operating room was classified into ISO 7 [23] with a maximum of particles per cubic meter at a size 0.5 μm is 352,000 or equal to the Federal Standard 209E [24] with a cleanliness level of 10,000 particles per cubic feet. The design specification of indoor environmental parameters in the investigated operating room included temperature of 22 ± 2 ◦C, relative humidity at 30–60%, and pressurization at 5 Pa. Figure 1 displays the HVAC system in the operating room, (a) filter in air handling unit (AHU) system to filter dirty air coming from the mixing air cabin, (b) cooling coil to cool down the temperature of the air with chilled water on the coil that comes from the process of cooling the water by the chiller, (c) heating coil for the process of heating and humidifying the air to fit with the design of the operating room, (d) fan to supply the air into the room, (e) HEPA filter to filter the air particles with the efficiency of HEPA filters over 99.97% (above 0.5 μm) so that the air entering the room becomes clean, (f) return air; the air coming from the room draughts to the AHU system and supplies the operating room.

**Figure 1.** The investigated operating room: (**a**) HVAC system; (**b**) snapshot.

#### **3. Methodology**

The proposed methodology of this research is illustrated in Figure 2. It generally consists of three steps: field measurement test, CFD simulation, and performance improvement strategy.

**Figure 2.** The proposed methodology framework.

#### *3.1. Field Measurement Tests*

This operating room requires a clean environment to prevent contamination in the room because the surgical process is sensitive to environmental parameters, including temperature, relative humidity, particles, and pressurization. The field measurement tests were conducted to examine the indoor environment parameters during an unoccupied period (at-rest). Parameters taken were airflow rate, pressurization, particle counts, temperature, and relative humidity. The apparatus tests for field measurement were as follows (1) Airflow rate: a TSI model PH-731 was employed to measure the airflow rate of each HEPA in cubic meter per hour. (2) Particle counter was used to count particles in the air of an operating room. Finding out the number of particles in the room was carried out using Met One 3413, and it was measured at the height of 1.2 m above the floor within a one minute recording. (3) Temperature and relative humidity in the room are very influential on objects in a room. TSI model 9565P was used to find out the temperature and humidity value in this operating room, and it was measured at the height of 1.2 m above the floor with three times measurement. The detailed specification of the apparatus field measurement tests is shown in Table 1. Furthermore, the measurement data were validated with the results of numerical simulation. The results of this measurement data were used as the basic parameters for ensuring that the operating room is in accordance with the desired design.

**Table 1.** Apparatus for field measurement tests.


#### *3.2. CFD Simulation and Improvement Strategy*

CFD was conducted to investigate the airflow distribution and concentration of airborne particles. The operating room simulation was performed using the ANSYS Fluent software version 2020 R2 [25]. Three different operating room schemes were conducted in this study to determine the performance of the HVAC system and reduce the concentration of the contaminant. The geometry was created based on the actual size and situation in the operating room, as shown in Figure 3. The study aims to prevent bacteria or even

fungi from entering the patient's body who is undergoing the surgical process. In addition, the analyses of environmental conditions are also considered. Various cases based on the different conditions are described below.


**Figure 3.** The geometry of the investigated operating room in three different schemes: (**a**) case 1: at-rest condition; (**b**) case 2: operational condition; (**c**) case 3: actual condition.

#### *3.3. Airflow Modelling and Boundary Conditions*

ANSYS Fluent provides several equations to solve the problems, including laminar and turbulent fluid flow problems, incompressible and compressible fluid, and other problems. In order to solve the flow and temperature fields of the problem, the equation for mass conservation, momentum, and energy is written as follows

$$\frac{\partial \rho}{\partial t} + \nabla \left(\rho \stackrel{\rightarrow}{\vec{v}}\right) = S\_m \tag{1}$$

$$
\nabla = \frac{\partial}{\partial \mathbf{x}} \stackrel{\rightarrow}{l} + \frac{\partial}{\partial \mathbf{x}} \stackrel{\rightarrow}{j} \frac{\partial}{\partial \mathbf{x}} \stackrel{\rightarrow}{k} \tag{2}
$$

Equation (1) is the general form of mass conservation equation for incompressible and compressible flows. Where the *t*, *ρ*, → *v* , ∇ are time, density, velocity, gradient operator, respectively, and *Sm* is the mass added to the continuous phase from the dispersed second phase and any user-defined sources. ANSYS Fluent defined ∇ according to the cartesian coordinate in Equation (2).

Airflow turbulence simulation uses two simulation methods carried out in this study, transient and steady-state condition with the renormalization group (RNG) *k*-ε as the turbulence model. Transient conditions can be used to monitor the reduction in the concentration in the operating room with a simulation time of about 500 s with a time step of 50 s. The steady-state condition is used to validate the field measurement data for temperature and velocity. The iterative coupling calculation for this stage is solved by the SIMPLE (Semi-Implicit Method for Pressure Linked Equation) method. The numerical simulation was calculated until it reached the residual bellow 10−<sup>3</sup> for the velocity and continuity, while energy residuals reached below 10−<sup>6</sup> to produce more precise results. The general form of the RNG *k*-ε model governing equation is as follows

$$\frac{\partial}{\partial t} \left( \rho \phi \right) + \nabla \left( \rho \phi \stackrel{\rightarrow}{\mathbf{V}} \right) = \nabla \left( \Gamma\_{\phi} \nabla \phi \right) + S\_{\phi} \tag{3}$$

where *<sup>ρ</sup>* is the density of air, <sup>→</sup> V is the air velocity vector, *φ* represents each of the three components, Γ*<sup>φ</sup>* is the effective diffusion coefficient of *φ*, and *S<sup>φ</sup>* is the source term.

Lagrangian particle tracking was used for the simulation method of particle tracking. The bioaerosol was injected into the space with the discrete phase model and simulated transiently. The particle size was 1–5 μm with a median size of 2.5 μm with a density of 1000 kg/m3, which is approximately equal to the density of water. It was simulated as spherical particles. The discrete phase for supply air and outlet air was set up with "escape" boundary conditions, while the remaining surfaces such as walls, medical equipment, etc., were set up with "trap" boundary conditions. Saffman lift force and thermophoretic force for the particle phase were used in this study. The equation is as follows.

$$\frac{du\_{pi}}{dt} = \frac{18\mu}{\rho\_{p}d\_{p}^{2}} \frac{C\_{D}Re}{24} \left(\mathcal{U}\_{i} - \mathcal{U}\_{pi}\right) + g\_{i} \left(1 - \frac{\rho}{\rho\_{p}}\right) + F\_{ai} \tag{4}$$

where *Ui* and *Upi* are the velocities of the fluid and particles, respectively; *μ* is the molecular viscosity of the fluid; *ρ* and *ρ<sup>p</sup>* are the densities of the fluid and particles, respectively; *dp* is the diameter of the particles; *Re* is the particle Reynolds number; *CD* is the drag coefficient; *gi* is the gravitational acceleration in the *i* direction; *Fai* is the additional force exerted on the particles.

A tracer gas method was carried out in this study, with simplified pathogen or particles without considering biological characteristics. Carbon dioxide (CO2) was selected as a pollutant to assess indoor ventilation efficiency and environmental conditions in the operating room with different models and conditions. The boundary conditions of numerical simulation are shown in Table 2. The CO2 concentration in the outdoor atmosphere is about 400 ppm, which will be used as the concentration value for supplying air from HEPA [26]. The recommended indoor CO2 concentrations should be maintained at or below 1000 ppm [27]. The exhaled air from patient and personnel are set with a concentration of around 38,000 ppm [28]. Heat flux generated from each patient and surgeon are at 17.45 W/m<sup>2</sup> and 33.55 W/m2, respectively [29]. The walls and door were assumed to be adiabatic, which have no heat transfer.


**Table 2.** The boundary condition for numerical simulations.

#### *3.4. Grid Independence Test and Validation*

The parameters could change the level of accuracy during the simulation process [30]. Increasing the number of mesh elements influences the accuracy of the simulation results. However, this requires a long time and sufficient resources. The grid independence test and validation of the simulation are illustrated in Figure 4. Three different meshes with 894,474, 1,543,686, and 2,303,385 were generated and simulated to obtain the optimal number of elements that can meet the appropriate meshing process. Furthermore, the grid independence test was validated for accuracy with the field measurement data of temperature and velocity. There were seven temperature measurement points in the field measurements, and then compared with the numerical simulation results. The velocity data were analyzed from the height of 0 m to 3 m. The type of mesh that is closest to the measurement data is 2,303,285 elements. However, it required sufficient resources and time. Therefore, considering the number of 1,543,686 elements with an error of less than 10% could be optimum for the subsequent simulation.

**Figure 4.** Grid independence test and validation of the measurement and simulation. (**a**) Grid independence test; (**b**) Validation.

#### **4. Results and Discussion**

#### *4.1. Experimental Results*

Field measurement tests were conducted in at-rest occupancy state conditions, and the results were also retrieved in accordance with the operating room standards. The results can represent the operating room quality with the data parameters: ventilation rate, temperature, relative humidity, pressurization, and particle counts. The measurement results revealed that the total air change per hour in this operating room was 22 ACH, which qualifies to the operating room design based on ASHRAE Standard 170 with a minimum of 20 ACH. The other indoor environmental parameters met good agreement with the standard: temperature average of 21.6 ◦C, relative humidity of 51.4% (shown in Figure 5), and pressurization of 10.6 Pa. The operating room is classified as ISO 7 (class 10,000). According to the field measurement results, particles at size 0.5 μm were counted less than the standard of 352,000 particles/m3, and also the particles at size 5 μm were counted less than 2900 particles/m3. The field measurement points of 2, 4, and 6 are located under the supply HEPA filter, resulting in a lower temperature and fewer particles in contrast to the outer HEPA location.

**Figure 5.** Field measurement test results, (**a**) temperature and relative humidity, (**b**) particle counts and air velocity.

#### *4.2. Airflow Pattern Distribution*

The airflow distribution in these three different cases needs to be reviewed in more detail. This will certainly affect the flow of clean air supplied by the HVAC system through the HEPA filter to reduce the amount of contamination in the operating room. Figure 6 shows the airflow distribution in different schemes. The airflow in case 1 can spread eventually in the room. In contrast to case 2, the addition of patients, personnel, and medical staff in a critical area becomes very influential in the airflow spread. Some airflow that hits the surface of the human body will cause the variation of velocity. Not only that but the airflow also does not appropriately spread at the bottom of the patient's bed. This is caused by the obstruction of airflow. Overall, the airflow in the room for case 2 can be well distributed, although it has not reached the entire room properly. The HVAC system recirculates the air in the operating room through the return air grilles. Case 3 shows the importance of paying attention to the airflow direction in the room blocked by the medical trolley. Putting a trolley near the return air grille causes the air suction process in the operating room to obstruct. The airflow collides on the top surface of the trolley and makes the air flow in a reverse direction. The air is more turbulent in the room. Therefore, airflow has difficulty reaching areas outside of the critical zone.

**Figure 6.** The airflow distribution in different schemes: (**a**) case 1: at-rest condition; (**b**) case 2: operational condition; (**c**) case 3: actual condition.

#### *4.3. Contamination Removal Analyses in Different Cases*

Different cases certainly have different results of concentration. Case 1 analyzes the performance of the HVAC system to reduce the amount of contamination during at-rest conditions. Furthermore, the identification was made in case 2 by adding occupants as a source of contamination in the operating room. Most likely, the amount of contamination is no longer the same as in case 1. However, conditions in the room are not always ideal conditions. Sometimes, health workers put surgical equipment anywhere and, therefore, could block the path of air return. Therefore, case 3 should be examined more comprehensively to determine whether there is a difference from the other cases.

The airflow in the operating room also has a function to dilute contamination. CO2 was assumed as a source of contamination in the operating room. Figure 7 shows the concentration profile in three different cases. Contamination in case 1 has a low concentration because the contamination could be diluted quickly, and also clean air is evenly distributed in the room (Figure 7a). Case 1 shows a decrease in contamination and is more efficient compared to other cases. This is because the operating room in case 1 was unoccupied and without any medical equipment. In contrast, in case 2, additional patient and surgery personnel generated some contaminants. Concentration increases in case 2 due to the addition of occupants placed in the middle of the room (critical zone) so that the airflow flowing in the room is slightly obstructed (Figure 7b). The air that spreads in the room has a concentration of air exhaled from the patient and surgery personnel. The cross-section from that figure shows the highest concentration is in the ceiling because airflow cannot reach that part. With the object near the air return, it means the dilution of concentration in the room is inhibited so that the concentration in the room is higher than in other cases. In case 3, it can be clearly seen that the top of the trolley has a high concentration value (Figure 7c). The clean airflow causes this area to be obstructed by the object. The results show the differences between the schemes given in the rooms in case 1, case 2, and case 3. In the final condition, case 1 has the lowest concentration level of 404 ppm, case 2 has an average concentration of 420 ppm, and case 3 has a higher concentration than other cases of 474 ppm. This analysis shows that using the ventilation rate of 22 ACH can efficiently reduce contamination in the different operating room cases.

Plane Z = ƺ3 m Plane Y = 1 m

**Figure 7.** The concentration profile in different schemes: (**a**) case 1: at-rest condition; (**b**) case 2: operational condition; (**c**) case 3: actual condition.

#### *4.4. Effect of Ventilation Rate on the Operating Room Concentration*

The ventilation rate certainly affects the amount of higher or lower concentrations in the operating room. Lower ventilation rates can result in energy savings but concentrations may increase. In contrast, increased ventilation rates produce fewer particles in the operating room but require more energy. Therefore, the optimal ventilation rate must be adjusted to match the energy consumption and concentration. Increasing the ventilation rate does not always result in a lower concentration, but air pattern distribution is one of the essential things. This study investigated an obstruction near the return air grilles by some medical trolleys, resulting in more turbulent airflow patterns. The increase and decrease in ACH number were carried out in this study with 15 ACH, 22 ACH, and 29 ACH, respectively.

Figure 8 illustrates the concentration contamination decay in different schemes that were monitored for 500 s. The results revealed that the air distribution spread eventually in the operating room for case 1 and case 2 with 22 ACH and could remove the contamination. For case 3, objects near the return air grilles made the concentration higher than the others. The average concentration in the operating room with 22 ACH was 474 ppm. Compared to ideal conditions, the result is higher when the medical equipment is located in the wrong area that could obstruct the airflow. In addition, studies with increasing ventilation rates were also carried out in order to result in lower concentrations. The results revealed that when ventilation rates increased to 29 ACH, the concentration became lower and could be reduced to 446 ppm. The concentration is close to case 2 when the operating room is in ideal conditions (no obstruction in the return air grilles).

**Figure 8.** Concentration decay effect in different schemes.

In order to have some energy saving, the reduction in ventilation rates was conducted in this study with 15 ACH. The average concentration in the room increased along with the reduction in the velocity inlet. The average concentration was 495 ppm. The lack of air distribution causes the dilution of concentration in the area to be inhibited. Reducing the ventilation rates or velocity of air supply entering the room cannot be tried by guessing the numbers because velocity is closely related to airflow and the dilution of contamination. Hence, velocity reduction also has a limit. The ventilation rate reduction could be possible for the operating room with at-rest (unoccupied) and ideal conditions.

#### *4.5. Ventilation Efficiency*

Ventilation efficiency is the ratio between the contaminant concentration in the occupied spaces and the concentration in the outlet air. It measures how effectively the air present in a space is replaced by fresh air from the ventilation system [31]. The ventilation efficiency is expressed by the Equation (1)

$$
\varepsilon = \frac{\mathbf{C\_{c}} - \mathbf{C\_{s}}}{(\mathbf{C}) - \mathbf{C\_{s}}} \times 100\% \tag{5}
$$

where ε is ventilation efficiency, Ce is pollutant concentration at the outlet air, Cs is pollutant concentration at supply, and (C) is the average pollutant concentration in the room.

The ventilation efficiency results are shown in Table 3. The results revealed that case 1 has the highest efficiency because of no concentration generated inside the operating room, while case 2 and case 3 decreased due to the additional personnel inside the operating room and the blocked return air grilles, resulting in lower ventilation efficiency.



#### *4.6. Bioaerosol Flow Path Model*

Several things that can affect the distribution of bacteria-carrying particles (BPCs) are occupancy state condition, ventilation rates, and operating room design and condition. In this study, two different operating room conditions were investigated during the ideal and actual conditions. The bioaerosol particles were injected into the space from the exact location. Particle sizes were 1–5 μm and median 2.5 μm. The particles were released for the same amount of time in the simulation. Figure 9 illustrates the results of the different particles birth time of 50 and 300 s in two different conditions.

**Figure 9.** Distribution of bioaerosol particles in the operating room. (**a**) case 2: particle birth at 50 s; (**b**) case 3: particle birth at 50 s; (**c**) case 2: particle birth at 300 s; (**d**) particle birth at 300 s.

The model carried out in the ideal condition presents a good flow path of the bioaerosol particles model. Particles' birth time nodes at 50 s could be faster diluted through the outlet air (Figure 9a). Along with the time period, the airflow distribution could carry the particles reaching almost the ceiling corner (Figure 9c). In addition, the different conditions were conducted to know the effect of the actual condition in the operating room. The placement of the medical equipment was located blocking the outlet air. The bioaerosol particles were injected. The particles spread to the upper corner of the operating room when simulated for 50 s of particles' birth time, as shown in Figure 9b. It also had deposition particles below the surgical table. The medical table location affects the air pattern and obstructs the removal flow path to the return air grilles. Considering not to put the medical equipment near the outlet air grilles could make for better particles' removal.

#### *4.7. Pressurization*

In order to maintain the quality of the air in the operating room, it should have sufficient clean air supplied to dilute and remove the airborne contamination generated within the room. Pressurization is critical to the proper functioning of the cleanroom. Thus, the contamination can be prevented during the surgery process. Figure 10 depicts the results of the pressurization in a different scheme. The field measurement was conducted with the pressurization at 10.6 Pa, compared to the numerical simulation with the pressurization of 10.8 Pa, which has been validated, and the results were close to the experimental results. The design specification of the pressure is 5 Pa, and excessive design air supply creates high pressure. This study was conducted in three different models and different ventilation rates. Case 1 had a pressure of 10.8 Pa, followed by case 2 with 11.1 Pa, and case 3 with 10.9 Pa. The difference model conditions in the operating room did not show a significant change in the results of room pressurization. In addition, the different ventilation rates in case 3 made a quite significant change compared to the existing design. The results of operating room pressurization with ventilation rates at 15 ACH, 22 ACH, and 29 ACH were 9.2 Pa, 10.9 Pa, and 13.6 Pa, respectively. Increasing the ventilation rates makes the pressurization higher, but when it decreases, it is still larger than the design requirement minimum at 5 Pa.

**Figure 10.** Pressurization effect with different schemes.

#### **5. Conclusions**

This research investigates indoor environmental parameters for the operating room through field measurement tests. CFD simulations were also conducted to investigate and analyze the operating room performance in different schemes. The conclusions are as follows:


420 ppm. Then, some medical equipment blocked the outlet air in case 3, resulting in the highest concentration with an average concentration value of 474 ppm.


**Author Contributions:** Conceptualization, F.W., I.P. and D.R.; Data curation, I.P. and B.Y.P.; Formal analysis, I.P., D.R. and B.Y.P.; Investigation, F.W., I.P. and B.Y.P.; Methodology, I.P.; Validation, I.P.; Visualization, I.P.; Writing—original draft, I.P.; Writing—review and editing, F.W. and D.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by the Ministry of Science and Technology under the grant no. MOST 109-2622-E-167-002-CC3.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **A Performance-Based Window Design and Evaluation Model for Naturally Ventilated Offices**

**Hardi K. Abdullah 1,\* and Halil Z. Alibaba <sup>2</sup>**


**\*** Correspondence: hardi.abdullah@su.edu.krd

**Abstract:** This study proposes a performance-based window design model for optimised natural ventilation potential by reducing the level of indoor carbon dioxide (CO2) concentration and improving thermal comfort, consequently minimising supplementary heating/cooling loads. The model consists of several stages: (1) Knowledge acquisition, (2) establishing a relationship between window design and natural ventilation, (3) identifying performance criteria and the design of experiments (DOE), (4) conducting performance-based dynamic simulations, (5) evaluation of findings, and (6) making informed design decisions. The study also proposed an evaluation method by which assessments of indoor CO2 concentration and adaptive thermal comfort are performed using the threshold suggested by the World Health Organisation (WHO, Geneva, Switzerland) and the acceptability categories of the British/European standard BS EN 15251:2007. The proposed model was applied to a single office inspired by the staff offices at the Department of Architecture, Eastern Mediterranean University, Famagusta, North Cyprus. The findings show that the developed model of performance-based window design enables the handling of various window design variables along with different performance criteria to determine the near-optimal window design alternatives for effective natural ventilation (NV) and mixed-mode (MM) offices. This model can guide architects in making informed decisions in the early stages of office window design.

**Keywords:** window design; natural ventilation; indoor air quality; indoor CO2 concentration; adaptive thermal comfort; performance-based design

## **1. Introduction**

Air movement for habitable spaces has an important impact on perceived indoor air quality [1]. Studies claim that air tightening within an occupied zone may result in complaints of unsatisfactory indoor air, particularly in air-conditioned (AC) spaces. Recent field studies suggest that elevated airspeed can achieve thermal comfort even at higher temperatures and improve perceived indoor air quality [2].

The importance of indoor air quality (IAQ) is reflected in the increased number of researchers studying various aspects of this topic. Due to the increasing demand for energy-saving and energy-efficient buildings, research into IAQ requires adopting various passive alternatives. In recent studies, the utilisation of natural ventilation to remove indoor pollutants and maintain indoor air quality, along with the indoor thermal comfort of various building types, has been challenged [3]. However, past attempts examined one goal at a time (e.g., indoor air quality, thermal comfort, energy consumption, productivity, etc.) and assessed the ideal environmental conditions for optimising that single target. The findings of previous studies recommend conflicting objectives and emphasise the need to pursue a more integrative approach to indoor environmental quality (IEQ) by tackling more than one criterion simultaneously [1].

Natural ventilation through window openings is the most common means to deliver fresh air indoors [4]. An effective method for maintaining indoor air quality and thermal

**Citation:** Abdullah, H.K.; Alibaba, H.Z. A Performance-Based Window Design and Evaluation Model for Naturally Ventilated Offices. *Buildings* **2022**, *12*, 1141. https:// doi.org/10.3390/buildings12081141

Academic Editor: Xi Chen

Received: 11 July 2022 Accepted: 29 July 2022 Published: 1 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

comfort is window openings controlled by building occupants. It has been proven that window-based NV can profitably replace mechanical ventilation, as well as ventilative cooling techniques using windows, and it can be harvested during free-running periods instead of using AC systems [5]. Therefore, a significant amount of energy consumption and carbon dioxide emissions can be reduced [6,7].

These discussions often note that window design has a strong relationship with NV performance regardless of the building type. Evidently, window design is an early decision for architects, who need adequate knowledge supported by quantitative data and experiments concerning airflow and heat transfer in buildings [8]. This study attempts to bridge the gaps in window design, natural ventilation, indoor air quality, and thermal comfort in a holistic, performance-based design approach that can guide architects in early design decisions.

#### *1.1. Aim and Objectives of the Study*

An appropriate window design can maximise the free-running period and thus save a considerable amount of energy and reduce CO2 emissions. Thus, architects need to understand the elements of window design decisions in terms of NV performance. The primary aim of this study is to develop a performance-based window design model that can optimise natural ventilation performance in terms of reduced indoor CO2 concentration and supplementary heating/cooling loads, as well as improved ventilation rates and thermal comfort in NV and MM offices.

Accordingly, the objectives of the study are:


#### *1.2. Architectural Considerations for Natural Ventilation*

The relationship of natural ventilation with a building is developed using various aspects of architectural design, which Kleiven [9] defined as characteristic elements in his concept of a "natural ventilation system". The decisions on these aspects are mainly made in the early architectural design process, including site selection (building location), planning, landscaping, building form, and envelope-related components [10].

Overall, building envelope elements have a greater impact on natural ventilation performance [11] due to the fact that most of these components are directly related to natural ventilation design, such as openings, shadings, orientation, thermal mass, etc. This study focuses on the effects of the building envelope, particularly those of window design on natural ventilation performance; thus, more details are provided on these topics.

#### *1.3. Window Design Parameters*

The glazed envelope is located at the opening of the building's façade and provides a visual connection between the outdoor environment and the indoor spaces. In addition to providing aesthetic value and a view to the outside, windows are the most critical components that affect building performance in terms of indoor air quality, natural ventilation, thermal comfort, daylight, visual comfort, and, essentially, energy performance. According to state-of-the-art research in the reviewed literature, including but not limited to [12–18], the most important window design variables identified are: size, orientation, type, opening, shape, position, separation, glazing, frame, and the availability of shading.

The impact of the window-to-wall ratio on different building performance goals has been studied more frequently, such as in the cases of [19–24]. The reviewed studies report that window size has a significant impact on natural ventilation conditions [25] and indoor environmental quality [26]. An investigation of windows located at the east and west orientations in a hot–humid climate showed that a 25% WWR provided better indoor thermal comfort conditions than a 50% WWR [27]. According to building regulations in North Cyprus, the minimum window size is defined as a 10% window-to-floor area ratio [28]. However, the question of whether this window size is sufficient to sustain the indoor air and thermal conditions of naturally ventilated offices needs to be answered.

Window orientation is considered a significant design parameter in terms of wind direction and solar radiation. A suitably placed window in a specific wall orientation can maximise ventilative cooling potential and minimise direct solar radiation, which is highly important in warm and hot climates. Therefore, window orientation is one of the critical energy-efficient design decisions that influence building envelope energy performance. The results of one study [29] investigating the effect of orientation and envelope insulation appliances found an up to 43% reduction in the resulting cooling load. Researchers [27] conducted an experimental study in a hot–humid climate; they reported that rooms with east-orientated windows had less thermal comfort hours than west-oriented windows in the case of 50% WWR, while both rooms performed similarly when they had a 25% WWR. The optimum window size depends on the window orientation and weather conditions; for instance, WWRs ranging from 10–70% are suggested for different window orientations and climates in Iran, where the difference between the minimum and maximum energy consumption rate is between 20–100% in its hot–humid climate [30].

Window type and natural ventilation are closely related to each other. The basic window types, performance ratings, and glossary of window-related terms are described in the AAMA/NWWDA/CSA 101/I.S.2/A440-08—North American Fenestration Standard/Specification for Windows, Doors, and Skylights [31]. Wang and Chen [32] investigated the impact of different window types, namely, casement, awning, and hopper windows, on single-sided natural ventilation with different opening angles using computational fluid dynamics (CFD) as an airflow prediction method. The findings suggest that the impact of the window type on the ventilation rate varied with the wind direction, whereby the windows and the turbulence effect created different flow patterns. These conclusions were also reported by a similar study [33]. Another study [34] evaluated the influence of different window types on ventilation performance in the residential buildings of Hong Kong using air change per hour (ACH) to quantify natural ventilation. The authors claimed that casement windows are the most effective design solutions, followed by awnings and sliding windows, in that order. It has been reported that casement windows are preferable in warm months, while hopper windows are preferable in cold months for both single-sided and cross ventilation [35]. Moreover, the natural ventilation performance of hopper windows also improves with a different opening angle [36], while the discharge coefficients of casement and hopper windows do not vary significantly [35]. Casement windows allow higher airflows for windward conditions compared to hopper and awning windows; however, hopper windows perform better in terms of overall airflow rates for all wind directions due to fewer obstructions [37].

In naturally ventilated buildings, window-opening behaviour significantly affects indoor air quality, thermal comfort, and energy consumption [38,39]. Closed windows increase the concentration of indoor particles (e.g., PM2.5) emitted by indoor particle sources [40]. Window-opening behaviour relies on both subjective sensations, particularly physiology and psychology, and objective factors, which include indoor air and thermal comfort; thus, it is subjected to a fair degree of randomness and uncertainty [41]. It has been found that the duration of window-opening in warm climates is significantly higher than in cold climates, especially during working hours (9:00–17:00) on weekdays, even in residential apartments [42]. Researchers [38,41] identified the major variables in determining the probability of window-opening as the level of indoor CO2 concentration and outdoor temperature. Furthermore, window-opening prediction models and occupant behaviour have recently come under consideration [43,44], including questions concerning the reliability

of simulation tools in handling this matter [45,46]. A few studies claim that occupantcontrolled window operation leads to insufficient natural ventilation performance; instead, they recommend automated ventilation control schemes [47–50].

Window shape (or window aspect ratio) is another important parameter that can affect the flow pattern of air indoors. The commonly used window shapes are rectangular (vertical or horizontal) and square shapes. One study [15] tested a number of vertical and horizontal rectangles and square windows with cross ventilation. The square window performed better than both the vertical rectangle and horizontal rectangle windows.

Opening position (or window location) is considered a significant factor that can affect the indoor airflow pattern. Shetabivash [13] studied the effect of various window positions and configurations on natural cross ventilation performance. The window positions the study investigated included placing the windows at the top and bottom of a room in opposite directions (windward and leeward sides). When the windows were placed at the same level but near the bottom of the wall, this presented the least effective scenario. However, window positions perpendicular to each other can improve natural cross ventilation performance [16]. Ventilation flow rate also depends on window separation in a way that low separation (S'~0.1)—aperture separation scaled by building width (S')—can boost single-sided natural ventilation performance, while a larger separation (S' > <sup>1</sup> <sup>2</sup> ) inhibits the realisation of this added benefit [51].

A window's thermal performance is typically a function of the glazing, frame, and perimeter details, with the overall goal of achieving the most effective natural ventilation (in the case of openable windows) to maintain IAQ and TC, as well as the best possible daylight transmission with the least heat transmission (e.g., heat gain and heat loss). Overall, glazing thermal performance relies on controlling the level of radiative heat transfer, which is mostly transferred through solar radiation and longwave infrared radiation [52]. One of the most effective ways of improving window thermal performance is the use of low-E coatings on the glass pane. Window frame conductivity is a function of the frame material, geometry, and use of thermal breaks inside the frames. Aluminium, vinyl (PVC), wood, and fibreglass are the common materials used for window frames in the building construction industry.

External window shade is another envelope component that is mainly applied to envelope openings. It is a form of solar control that can be utilised to optimise the amount of solar gain and daylight entering a building. Therefore, it can reduce energy use and, eventually, CO2 emissions. Window shade has a significant influence on the thermal and visual comfort of occupants, protecting them from overheating and glare. Numerous studies focus on the role of window shades on the energy usage, thermal comfort, and visual performance of buildings [20,22,53–56]. Overall, well-thought-out window parameters (including window size, orientation, and shades) lead to a significant improvement in natural ventilation conditions and thermal comfort, increasing the airspeed by six times and reducing the air temperature by 2.5% [12]. The most effective way to realise the full potential of natural ventilation in the Mediterranean climate is to determine the appropriate window-to-wall area for optimal thermal performance, the appropriate material for glazed windows, and the right shading devices when deciding on the building envelope so that the reliance on active systems is minimized [55].

#### **2. The Proposed Model of Window Design and Evaluation Relative to Natural Ventilation Performance**

#### *2.1. Rationale of the Proposed Model*

Architectural design is an iterative process of understanding, exploration, and validation in which design assumptions are continuously modified and assessed against the intended performance criteria. Using iterations, designers have the ability to go back and forth through the cyclical process until the design solution achieves a lower risk of failure. Therefore, architects need comprehensive frameworks to explore and evaluate their early design decisions, which eventually affect the upcoming design stages, construction stage, and post-occupancy building performance. The concept of the proposed model originated

from a performance-based design approach within the digital design process. In the PBD paradigm, "performance" is defined as "the desirability of the confluence between form and function in a given context" [57]. Unlike generative design (another approach to the digital design process), in the PBD paradigm, the computer does not generate design solutions but "acts as a partner with the designer during the design process" [58]. Hence, a performance-based design approach facilitates structuring the architectural design process to enable architects to make informed decisions in the early design stages [58,59].

Numerous studies have investigated the impact of window design on indoor environmental conditions [11–17,23–27,30,32,34–41,43–49,51,60,61]. Certainly, these attempts confirm the crucial role of window design on occupant health, comfort, and productivity, especially in naturally ventilated buildings. The concept of proposing a comprehensive, performance-based window design model is intended to provide architects with informative feedback about potential design decisions aimed at simultaneously improving IAQ and TC performance. Another significance of the proposed model is that it overcomes the limitations of previous methods in terms of reducing the required time and effort by adopting a practical approach in conducting a minimal number of experiments, called Taguchi design of experiments, to determine the impact of each design parameter on the performance criteria. For reference, in the case of eight parameters, each with three levels (38), the full factorial design method requires 6561 runs to test all combinations; in contrast, only 18 runs are necessary (less than 3%) for the Taguchi orthogonal arrays used in the proposed model. In addition, the proposed model facilitates the trade-off selection of design solutions among multiple objective functions as an alternative to the assumed optimal solution for a particular criterion.

#### *2.2. Components of the Proposed Model*

The proposed model is a performance-based model encompassing procedural methods aimed at ensuring architects make educated decisions early on in the design stage concerning office envelope design, particularly window- and NV-related design parameters. The major stages include (1) knowledge acquisition, (2) establishing a relationship between window design and natural ventilation, (3) identifying performance criteria and the design of experiments (DOE), (4) conducting performance-based dynamic simulations, (5) the evaluation of findings, and (6) making informed design decisions.

#### 2.2.1. Knowledge Acquisition

To start any architectural design process, the predesign stages involve data collection and knowledge acquisition about the project and its requirements. Therefore, the first stage of the proposed model is referred to as the "knowledge acquisition" of the space under design, such as the building location, information about the context and environment, and the building type and function, as well as relative local or international building regulations and codes. These pieces of information serve as design constraints, not variables, and should be considered by designers in defining design parameters in the proposed model.

#### 2.2.2. Establishing a Relationship between Window Design and Natural Ventilation

A well-designed window paves the way for efficient NV performance to improve indoor air, occupant thermal comfort, and, consequently, a reduction in the use of mechanical ventilation and cooling [25]. In addition, airflow rate, windspeed, and indoor temperature are directly proportional to the various window design variables [60,61].

This stage combines the design of envelope-related components and a natural ventilation strategy. The model concentrates on the design of wall glazing in relation to NV performance within early building envelope design; nevertheless, other envelope-related design parameters can also be studied using the proposed model. Natural ventilation types (i.e., wind-driven and buoyancy-driven) and classifications (i.e., single-sided and cross ventilation) are defined by the window design parameters, for which the amount of airflow that enters and leaves the space is determined accordingly. Therefore, this stage establishes

a relationship between window design and natural ventilation by developing a correlation between various parameters affecting the ventilation rates and, consequently, indoor air and thermal conditions.

#### 2.2.3. Design of Experiments and Identifying Performance Criteria

Design of experiment is proposed as an alternative to full factorial design (FFD), in which the number of necessary experiments can be minimised to a reasonable amount while obtaining all the required information about the sensitivity of the design variables under study. Among the available DOE methods, this study suggests the use of the "Taguchi orthogonal arrays" method [62] as a standard method of experimental design. Furthermore, the data analyses include the analysis of variance (ANOVA) approach and the signal-tonoise (S/N) ratio [63]. Using this performance-based model, architects can select intended environmental performance objectives in the domains of indoor environmental quality and energy efficiency goals. However, in this model, the considered performance criteria are limited to ventilation rates, the indoor CO2 concentration level, and occupant comfort.

#### 2.2.4. Performance-Based Dynamic Simulations

The British/European standard 15251:2007 recommends "whole year computer simulations" as a reliable method to study and evaluate the indoor environment and energy performance of new and existing buildings. Studies on computer modelling and simulations have shown that computer simulations play a vital role in building design, influencing resident comfort and energy performance by helping to solve building performance issues [64]. Computer simulations of energy modelling require substantial knowledge about the physical and operational characteristics of the building, as well as precise input data on the building and climate. During the application of the proposed model, any validated simulation software can be used, such as computational fluid dynamics tools. In this study, Tas Engineering software version 9.4.4—developed by Environmental Design Solutions Limited (EDSL) [65]—was used to conduct the computational dynamic thermal simulations and fulfil this stage of the study.

#### 2.2.5. Evaluation and Decision-Making

This stage covers the evaluation of the analytical and numerical findings from simulation experiments, on the basis of which informed decisions can be made. The evaluation method comprises the assessment of each measurement indicator of the selected performance criteria, namely ventilation rate, carbon dioxide concentration, thermal comfort, and supplementary heating/cooling loads using a relevant and recommended calculated indicator. Following the evaluation of findings and data analysis, architects can make informed decisions, taking into account whether they are satisfied with the performance of the initial design or the evaluated results, and suggest improvements through the modification of envelope-related parameters, particularly wall glazing variables and NV design. Accordingly, the framework of the proposed model is developed and illustrated in Figure 1.

#### *2.3. Evaluation Method of the Findings*

The BS EN 15251:2007 standard in Annex I (see Table 1) contains a classification of indoor environmental assessments based on building status [66]. The developed model addresses the early design of office spaces by assessing the impact of various architectural design variables on the indoor environment, as well as the energy performance of a mixed-mode strategy (if applicable). Consequently, it applies a year-round hourly dynamic computer simulation based on the classification method suggested in the BS EN 15251:2007 standard. The objective is to guide decision-making in the early design phases and apply building performance simulation (BPS) at the outset of the design process in a PBD approach. The effectiveness of window design and its implications for NV performance were assessed in terms of the ventilative cooling potential for IAQ and TC and the additional HVAC load needed to maintain indoor environmental conditions when natural ventilation proved insufficient due to extreme weather conditions.


**Table 1.** Classification of methods used for indoor environmental assessment [66].

According to the BS EN 15251:2007 standard [66], the "calculated indicators of indoor environment method include the (1) simple indicator, (2) hourly criteria, (3) degree hours criteria, and (4) overall thermal comfort criteria (weighted PMV criteria)". The hourly criteria indicator was adopted in this study, which allows building performance to be assessed based on the percentage of time (%) and/or number of hours (h) during which the intended criteria were met.

This research is limited to examining and evaluating the performance of window-based natural ventilation in diluting indoor carbon dioxide and maintaining acceptable indoor air and thermal comfort for the building occupants. Hence, the considered measurement criteria are the ventilation rate and CO2 level, thus assessing indoor air performance and predicting the thermal sensation of occupants using the adaptive comfort model to evaluate indoor thermal comfort in free-running buildings while also lowering HVAC loads in mixed-mode spaces. The evaluation model for assessing the potential findings from the proposed model is illustrated in Figure 2.

**Figure 2.** Evaluation model used to assess findings from the proposed window design model.

#### 2.3.1. Assessment of Indoor Air Performance

The assessment of indoor air is limited to ventilation rates and carbon dioxide levels. Other common measurements of IAQ include concentrations of formaldehyde (HCHO) and volatile organic compounds (VOCs), which were not considered in this study. The concentration of carbon dioxide in an indoor space is often a reliable indicator of the quality of the space. CO2 concentration has also been used in previous studies to evaluate the ventilation performance of indoor spaces using the "gas tracer method" in field experiments or through dynamic building simulations. The benchmark limits of acceptable carbon dioxide concentrations in indoor spaces are defined by multiple standards and guidelines, including: the WHO [67], ASHRAE 62.1 [68], BS EN 15251 [66], and EN 13779 [69] standards. The World Health Organisation [67] recommends 1000 ppm as the upper limit of CO2 concentration, after which higher concentration levels are an indication of poor ventilation, significantly increasing the likelihood of indoor air quality problems and resulting in sick building syndrome [70].

In the same vein, the BS EN 15251:2007 standard [66] classifies indoor CO2 levels into different categories. The ASHRAE 62.1 standard similarly endorses the 1000 ppm threshold specified by the WHO, which is within the Category II range of indoor carbon dioxide concentration specified by the BS EN 15251:2007 standard. The 1000 ppm threshold recommended by the WHO was utilised in this study to evaluate the natural ventilation performance of different types of offices. Table 2 outlines the various standards addressing the level of indoor carbon dioxide concentration.


**Table 2.** Indoor carbon dioxide concentration thresholds defined by relative standards.

#### 2.3.2. Assessment of Ventilation Rates

Natural ventilation efficiency can be evaluated based on the amount of fresh air delivered to indoor spaces from the outdoor environment. The airflow rate can be evaluated through the relevant standards for determining the acceptability of indoor air quality and ventilation rates, including the ASHRAE 62.1 [68], BS EN 15251 [66], and EN 13779 [69] standards. The minimum ventilation rates outlined in these standards are determined based on the type of building, occupancy, and/or floor area. The breathing zone outdoor airflow (*Vbz*) in the ASHRAE 62.1 standard is calculated using Equation (1). Similarly, the BS EN 15251:2007 uses Equation (2) to calculate the overall ventilation rates (*qtot*) for indoor spaces based on the building emission ventilation rates (*qB*). It is noteworthy that, despite the fact that both standards adopt similar logics, they do not necessarily produce identical outputs. The ventilation rate calculation method suggested in the BS EN 15251:2007 standard was utilised in the proposed evaluation model. Table 3 outlines the recommended ventilation rates for office spaces. It is worth mentioning that the ventilation rate for smoking was omitted due to the prohibition on smoking in offices.

$$V\_{bz} = R\_p \cdot P\_z + R\_a \cdot A\_z \tag{1}$$

where *Rp* is the airflow rate per person (L/s·pers), *Pz* is the number of occupants, *Ra* is airflow per unit area (L/s·m2), and *Az* is the zone floor air (m2).

$$q\_{\text{tot}} = \mathbf{n} \cdot q\_p + \mathbf{A} \cdot q\_B \tag{2}$$

where *qtot* is the total ventilation rate of the space (L/s), *n* is the number of occupants, *qp* is the airflow rate per person (L/s·pers), *<sup>A</sup>* is the zone floor air (m2), and *qB* is the airflow rate for building emissions (L/s·m2).

**Table 3.** Ventilation rates (L/s·m2) for non-low polluted offices defined by the BS EN 15251 standard [66].


2.3.3. Assessment of Indoor Thermal Comfort

Indoor thermal comfort is another significant performance criterion that needs to be evaluated when assessing IEQ, especially in warm and hot climates. As stated in the previous sections, the scope of this research is limited to NV—including mixed-mode buildings; therefore, to achieve more reliable results, the most precise and suitable thermal comfort model should be employed. Fanger's PMV and PPD model [71] is widely used to assess the thermal comfort status of airconditioned spaces, although some researchers claim that the PMV and PPD method overestimates the percentage of occupant discomfort in hot and warm conditions for naturally ventilated spaces [72]. Furthermore, field studies have proved that the adaptive thermal comfort model is better suited to addressing the thermal comfort of users in free-running and MM buildings, owing to the fact that this

method takes into account human adaptation mechanisms as a reaction to changes in the outdoor environment [1,73].

The field studies under review take a negative position regarding the classification of the MM system with respect to AC buildings in current thermal comfort standards (i.e., ASHRAE 55 and BS EN 15251), arguing instead that natural ventilation is in use for most of the occupied hours in office spaces. Natural ventilation is described as being synonymous with free-running buildings in the aforementioned thermal comfort standards, for which the adaptive thermal comfort model has been developed using information generated by a variety of field studies. Recent field surveys have found that occupant thermal sensations in NV and MM buildings are better represented using the adaptive model relative to the PMV/PPD model, which does not adequately account for the various ways in which residents can adapt to variations in outdoor weather conditions. Furthermore, adaptive thermal comfort can also be used in conducting climate change impact studies on mixed-mode office spaces [74].

In mixed-mode buildings, indoor thermal comfort involves NV and AC systems, which can be assessed individually using the adaptive and steady-state thermal comfort models, respectively. This study implements an adaptive method to quantify occupant thermal sensations in terms of being comfortable or not in a given period, thereby evaluating the space based on acceptable adaptive model comfort ranges suggested by the relative standards. The British/European adaptive comfort model, stated in the BS EN 15251:2007 standard [66], is used on account of its being less restrictive when explaining the model's applicability conditions compared to the American adaptive model (i.e., ASHRAE 55).

However, because this study focuses on the potential benefits of natural ventilation in office spaces (as a free-running building or under a mixed-mode strategy), the evaluation of indoor thermal comfort is limited to the natural ventilation period by the adaptive thermal comfort of the BS EN 15251:2007 standard shown in Equation (3). The optimal indoor operative temperature is defined relative to an exponentially weighted outdoor running mean temperature, which is calculated for the previous 7–30 days using Equation (4). Depending on the value of constant α, the significance of the resulting temperatures declines over time. The three categories defined in the standard are I (*To* ± 2), II (*To* ± 3), and III (*To* ± 4), respectively representing high, normal (for new buildings), and moderate (for existing buildings) expectations. Table 4 reports the details of the adaptive thermal comfort model of both the American (ASHRAE 55) and British/European (BS EN 15251) standards. Based on the upper and lower limits of the intended category, the number of comfort hours during the occupancy period can be utilised as an indicator in evaluating the thermal comfort performance of a design scenario, and it is formulated by the BS EN 15251:2007 standard as follows:

$$T\_o = 0.33 \cdot T\_{rm} + 18.8 \tag{3}$$

$$T\_{rm} = (1 - a)T\_{ad-1} + aT\_{ad-2} + a^2T\_{ad-3} + a^3T4\dots,\tag{4}$$

where *To* is the indoor optimal operative temperature (◦C); *Trm* stands for the exponentially weighted running mean temperature (◦C) for the last 7–30 days; *α* represents a constant between 0 and 1; and *Tod*−<sup>1</sup> is the daily mean outdoor temperature for the day before, the day before that (*Tod*−2), the day before that (*Tod*−3), and so on.

The significance of the temperatures declines over time, with the speed of decay depending on the value of the constant, α. The equation developers suggested α = 0.8 as an appropriate value according to their SCAT database [75].


**Table 4.** The differences between American and British/European standards for an adaptive thermal comfort model.

#### 2.3.4. Assessment of Heating, Ventilation, and Airconditioning Loads

The aim of the mixed-mode strategy is to realise the full potential of natural ventilation using operable windows and maintain the quality of indoor thermal performance by utilising supplementary heating, ventilation, and air-conditioning (HVAC) in extreme weather conditions. This results in significant energy savings, along with a reduction in GHG emissions.

Natural ventilation is typically used in a hot or warm climate when the outdoor temperature ranges between 20 ◦C to 24 ◦C [76]. To amplify the impact of ventilative cooling and ensure compliance with the occupants' window-opening preferences, as outlined in the adaptive thermal comfort model, NV operation can be predicted or, alternatively, designed based on automation. Such an automated design will allow the windows to start opening when the indoor air temperature is at 21 ◦C and fully open when this rises to 24 ◦C. Practically speaking, the building management system (BMS) will need to be integrated with the necessary control mechanism [76,77].

To reduce the chance of overcooling, the operation of window openings conforms to the cooling/heating temperature ranges suggested by the BS EN 15251:2007 standard for a particular category, such as Category II for normal expectations, as shown in Table 5. The maximum temperature required for cooling in AC spaces is 26 ◦C, while the minimum indoor temperature for heating is 20 ◦C. However, occupants in naturally ventilated buildings are able to adapt to a wider range of temperatures relative to the outdoor temperature using a variety of adaptive behaviours [78]. The operation of air-conditioning within the mixed-mode system is regulated by the minimum heating temperature setpoint for Category II (20 ◦C), while the cooling temperature setpoint is defined by the Category II upper limit of the European adaptive model, as shown in Equation (5). For reference, cooling begins when the outdoor running mean temperature is 30 ◦C and the indoor operative temperature reaches 31.7 ◦C.

$$T\_{o,u-ii} = 0.33 \cdot T\_{rm} + 21.8 \tag{5}$$

**Table 5.** Heating and cooling temperature ranges for hourly calculation in Category II of the BS EN 15251:2007 standard [66].


Lastly, the annual comfort hours provided by natural ventilation (free-running period) are represented by the number of hours when the indoor operative temperature is within the acceptability limits of the adaptive model. Thermal satisfaction can be provided for the remaining office working hours (discomfort period) through mechanical air-conditioning in the mixed-mode system. The total HVAC load of the air-conditioning period is calculated for each design alternative. A comparative study for a particular design solution can be

conducted to contrast the performances of the mixed-mode system and full air-conditioning based on the heating and cooling temperature ranges, as defined in Table 5. Therefore, the assessment of HVAC in MM offices is based on maximising the free-running period (only NV in operation) and minimising the AC period using the number of hours, in which a specific mode is in operation during office working hours (occupation), as the calculated indicator.

#### *2.4. Validation of the Model Using Ventilative Cooling Methods*

Developed by the National Institute of Standards and Technology (NIST) [79] and further advanced in the International Energy Agency (IEA) Annex 62 [80] framework, the ventilative cooling (VC) method is used in validating natural ventilation performance in comparison to the comfort hours forecasted by the dynamic building simulation. The prevalence of this method is partly due to the growing interest in energy-efficient buildings and reducing greenhouse gas emissions. The VC method is useful for evaluating the potential benefits of natural ventilation during early design stages by accounting for internal heat gains (i.e., lighting, occupancy, solar radiation gains, and equipment gains), the thermal properties of the building envelope, and the airflow rate required to maintain IAQ and TC based on the relevant standards and regulations. Based on local climatic conditions, such an analysis is particularly useful for designer decision-making as it relates to the configuration of the building envelope and layout.

The algorithm used by the model considers the intended thermal comfort criteria and processes annual climatic conditions on an hourly basis. The model is derived from the energy balance of a well-mixed single zone, accepting that the accumulation term of the energy balance could be insignificant in the event that either the space's thermal mass is negligible or the internal temperature is maintained at a relatively constant level. In such an instance, the steady state model defines the thermal response of the zone based on an approximation of the particular climate's ventilative cooling potential, calculated using Equation (6).

$$T\_{o-hbp} = T\_{i-hsp} - \frac{q\_i}{\dot{m}p\_{\min}}\tag{6}$$

where *To*−*hbp* is the heating balance point temperature (◦C), *Ti*−*hsp* is the internal heating setpoint temperature (◦C), *qi* is the total internal and solar heat gains (W/m2), . *mmin* is the minimum required mass flow rate (kg/s), *cp* is air capacity (J/kg·K), ∑ *UA* is envelope thermal conductance (w/K), *<sup>U</sup>* is average U-value of the envelope (W/m2·K), and *<sup>A</sup>* is the area of the envelope exposed to outdoor conditions (m2).

According to this method, heating must be introduced when the outdoor air temperature falls below a certain level in order to preserve the indoor air temperature at a required internal heating setpoint temperature (*Ti-hsp*), which is determined by the heating balance point temperature (*To-hbp*). Direct ventilative cooling can be introduced when the outdoor temperature is higher than the heating balance point temperature as a means to counterbalance internal heat gains and maintain IAQ and TC within the required range. However, the utility of VC diminishes when the outdoor temperature is at or below *To-hbp*, although acceptable and healthy indoor air requires the provision of the minimum required ventilation rate suggested by the relevant standards, including BS EN 15251:2007 and ASHRAE 62.1.

In AC buildings, the steady-state values constitute the minimum and maximum *Ti-hsp*, taking into consideration the building type, such as the indoor temperature ranges suggested for cooling and heating in office spaces, as previously outlined in Table 5. However, the development of the adaptive comfort model progresses relative to variations in outdoor temperature; consequently, the acceptability limits (ASHRAE 55) or categories (BS EN 15251) for adaptive comfort are used to calculate *Ti-hsp*. As was pointed out earlier, Category II (for new buildings) of adaptive thermal comfort forms the primary focus of this study, the conditions for which are also applied to the analysis of ventilative cooling.

To compare the results of both the VC method and the dynamic simulations, it is necessary to calculate the amount of direct ventilative cooling resulting from an increase in the airflow rate. This can guarantee comfort conditions when the outdoor temperature falls inside the limits set for the comfort zone temperature, taking into consideration the temperature range of the particular category (i.e., Category II of the BS EN 15251 standard). If we accept that conductive losses that occur in the warm months are relatively small compared to the internal gains (i.e., ∑*UA (Ti-max* − *To-db)<qi*), the ventilation rate required for the provision of thermal comfort can be calculated using Equation (7).

$$
\dot{m}\_{\rm cool} = T\_{i-\rm hsp} - \frac{q\_i}{c\_p \left(To - db\_{i-\rm max}\right)}\tag{7}
$$

where *Ti*-max is the upper limit temperature of Category II (calculated by Equation (5) and *To-db* is the outdoor dry bulb temperature.

#### **3. Model Application: Window Design of a Single Office with Single-Sided Natural Ventilation**

#### *3.1. Knowledge Acquisition*

In this study, a hypothesised single office with single-sided natural ventilation was proposed, inspired by the academic staff offices at the Department of Architecture, Faculty of Architecture, Eastern Mediterranean University, Famagusta, North Cyprus. The application of various open-plan offices with cross ventilation can be found in [81]. The office floor area is 16.8 m2, and the floor aspect ratio was taken to be 1:1 (4.1 m × 4.1 m). The clear ceiling height was fixed at 3 m in accordance with the normal floor-to-ceiling height recommended in local building codes and regulations [28]. To examine the effect of an exclusively window-based NV design on the predefined performance criteria, the layout and form configuration, as well as the properties of the vertical and horizontal opaque features, were fixed in all design scenarios. These offices are designed to accommodate just one person; however, the provided space is often used by two persons, or even more, in some situations for a limited period. In this research, it was assumed that two occupants use the space during office hours (i.e., 9 am to 5 pm). Therefore, the floor area per person exceeds the suggested 10 m2 per user in single offices [66,82], resulting in elevated internal heat gains and, eventually, higher CO2 releases from occupants.

Due to the size of single offices, the majority of cases utilising such office designs have only one wall with an external condition or exposed to the outdoor environment. Hence, there might be a limit to the amount of fresh air permitted into the indoor space through a window (or windows) from this particular external wall, which is known as single-sided natural ventilation. It is worth mentioning that in North Cyprus, the minimum ratio of the WFR is 10%, and the minimum provided window-opening area is 5% or half of the minimum WFR [28].

#### *3.2. Establishing a Relationship between Window Design and Natural Ventilation*

The considered window design variables included window size, orientation, type, glazing property, aspect ratio, location, and shading availability. The levels of window size were 10%, 20%, 30%, and 50% (e.g., an approximately fully glazed external wall) windowto-floor area. The window orientations studied were north, south, east, and west, while the remaining available orientations were excluded. As explained in the previous sections, there are various types of windows relative to their operation. Of these, four common types were investigated in the present study, namely: casement, sliding, double-hung, and single-hung. The selected window types offer different natural ventilation scenarios depending on the driving forces of the NV, such as wind-driven and buoyancy effects. The glazing property is considered one of the most sensitive parameters affecting window performance in terms of indoor thermal comfort. Single-pane glass, double glass, double glass with low emissivity (low-E) coating, and triple glass with low-E coating were tested as various levels of glazing properties. The window aspect ratios of 1:1 (square shape) and 1:2 (rectangle shape), as well as the location of the window placement (i.e., middle or side) in the wall, were other studied variables and their particular levels were taken into account. The availability of shading is another studied parameter that can have a significant influence on window performance. Different design scenarios with either fully shaded windows during office hours or no shading mechanisms were examined as parameter levels to determine the role of shading in the summer period. Shading can be provided using any external or internal means, vegetation, solar shading devices, internal curtains, etc. In this research, external shading devices using horizontal fins (for south-oriented windows) or vertical fins (for east- and west-oriented windows) were implemented. The fins were designed in a way such that they can prevent excessive solar gains during office working hours in the warm months, specifically, May, June, July, August, and September.

The hypothesised office for a single-office design comprises a single thermal zone, which is located on the ground floor. The wall containing the window was defined as an external wall, whereas the other walls were assumed to be internal walls, and the ceiling was also considered an internal surface. Table 6 summarises the construction specifications used in the building performance simulations. The selection of materials and their properties were identical to the case study office building (determined by field observations), representing common construction systems in the study location (determined by studying local building construction guidelines). However, the glazing material was considered one of the window design variables in order to test different compositions.


**Table 6.** The construction materials and their U-values.

#### *3.3. DOE and Selection of Performance Criteria*

Table 7 outlines the studied window design parameters and their considered levels. Based on the number of design parameters and their levels, the most appropriate Taguchi orthogonal array is *L*16 (4ˆ4 2ˆ3) for which the Taguchi-based DOE suggests sixteen experiments to understand the whole study as well as the effect of each variable on the intended performance objectives. Thus, Table 8 reports the required design scenarios and the specific levels of each factor.


**Table 7.** The studied single-office window design variables and their levels.

**Table 8.** Simulation design scenarios based on the Taguchi *L*16 (4ˆ4 2ˆ3) standard orthogonal array.


Using analysis of variance, the effect of the design parameters on the intended performance criteria was evaluated, including the DF, the SSV, the SSTO, the MSV, the MSE, and factor effectiveness. The S/N ratio was used to identify the near-optimal level combinations of the design variables through a logarithmic transformation of the mean square deviation, where the signal-to-noise ratio of larger-is-better was employed for performance criteria related to NV, and smaller-is-better was applied for supplementary AC loads.

The intended measurement criteria for assessing window design in relation to NV performance were the airflow rates, CO2 concentration, adaptive thermal comfort, and mixed-mode loads. The calculated indicator for the NV-related measurements was the number of hours in which the criteria were met. That is, the total number of hours at which airflow rate and adaptive comfort are within Category II of the BS EN 15251:2007 standard and the number of hours in which the CO2 concentration level is equal to or less than the WHO threshold of 1000 ppm. Furthermore, the number of electricity loads (kWh/m2) required to maintain indoor thermal conditions when NV is not adequate was calculated to evaluate MM air-conditioning loads.

#### *3.4. Performance-Based Simulation*

#### 3.4.1. Setting Weather Data

The international climate zone classification provided in ANSI/ASHRAE/IES 90.1- 2019 [83] and the Köppen–Geiger climate system [84] classify Famagusta (35.1149◦ N, 33.9192◦ E) weather under warm–humid or the Csa: Mediterranean climate, respectively. This climate is characterised by cold, rainy, rather changeable winters and dry, hot summers in which July and January are the warmest and coldest months of the year, as described in Figure 3. Figures 4 and 5 show the monthly average diurnal temperature swing and global horizontal radiation and the wind rose of the study location, respectively.

The moderate climate of Famagusta facilitates the adoption of the mixed-mode system to preserve IAQ and TC conditions, which maximises the use of natural ventilation and energy-saving potential. The International Weather for Energy Calculations (IWEC) [85] offers typical metrological year (TMY) hourly datasets, which can be used for dynamic computational simulations. For verification purposes, the TMY datasets were compared to hourly weather data for 2019, measured by an official local metrological office. The comparison indicated the relative consistency and accuracy of the TMY datasets, which represent real conditions.

**Figure 3.** Famagusta climate characteristics on (**a**) 21st January and (**b**) 21st July.

**Figure 4.** Monthly average diurnal temperature swing and global horizontal radiation.

**Figure 5.** Wind speed and wind directions in Famagusta.

3.4.2. Benchmark Values for Internal Heat Gains and Schedules

The empirical-based benchmark values suggested by the Chartered Institution of Building Services Engineers (CIBSE) Guide A: Environmental Design [82] were employed to define the internal heat gains of the single thermal zone (office space); refer to Table 9. For occupancy, electrical equipment, and 500 lux artificial light schedules, the internal gains of the highest possible scenario (*k* = 1.0) were accounted for, corresponding to 45.0 W/m2, the average total internal heat gain (*Qint*). Finally, 0.3 ach was set for infiltration, and in order to determine only natural ventilation potential, no mechanical ventilation was assigned to the mixed-mode system.

The ASHRAE 55 standard [86] and ASHRAE standard [87] predict a metabolic rate of 1.2 met for office activities (e.g., sedentary and light office work), which corresponds to 125.7 W/person. Based on the Du Bois method [88], an average-sized adult releases 0.0052 L/s carbon dioxide, which is described in the ASHRAE 62.1 standard (ventilation for acceptable indoor air quality) [68]. In accordance with the 8.4 m<sup>2</sup> office area per person in this specific case study, the total CO2 generation rate was 2.22 l/h/m2.


**Table 9.** Schedules and loads assigned to calculate internal heat gains for the study of a single office.

#### **4. Results and Discussion**

#### *4.1. Impact of Window Design Variables on the Studied Performance Criteria*

To appraise the window design variables and their respective levels, the annual acceptable hours, specified in the category ranges, for ventilation rate, carbon dioxide concentration, and adaptive thermal comfort were calculated. In addition, the annual air-conditioning loads for each design experiment, defined by the Taguchi *L*16 (4ˆ4 2ˆ3) orthogonal array, was measured and are displayed in Table 10 and Figure 6.

The sixteen representative runs indicate that scenarios 15 and 11 provide more acceptable comfort hours in terms of the ventilation rate, CO2, and thermal comfort compared to other simulated cases. In scenario 15, airflow rates were inside Category II for about 1573 occupancy hours (75.3%), carbon dioxide less than 1000 ppm was recorded for 1740 h (83.3%), and thermal comfort was within the Category II range of adaptive comfort for 1391 h (66.6%). The initial interpretation for this case could be the suitability of a larger window size, which provides more fresh air and ambient air-cooling potential, particularly when the window is placed at a southern orientation. In contrast, for example, these combinations required a higher energy demand for mechanical cooling (14.7 kWh/m2) than scenario 9 (12.9 kWh/m2), which means that larger window sizes contribute to a higher internal heat gain by allowing for a greater amount of solar radiation, particularly when solar shading does not exist.

**Table 10.** The total annual acceptable hours for VR, CO2, and TC, as well as air-conditioning loads for the set of the Taguchi *L*16 (4ˆ4 2ˆ3) simulation scenarios.


Using the analysis of variance method, the factor effect (percentage contributions) of the window design variables were perceived, as outlined in Tables 11–14. It can be concluded that window size has the highest impact on airflow and CO2 concentration at 81.59% and 73.54%, respectively, followed by the window orientation and type. Moreover, the window aspect ratio and location have the least influence on the studied performance objectives, for which the factor effect does not surpass 1.1% in any cases.

Contrarily, the factor effect of the window design parameters indicates different patterns when the acceptability hours of the adaptive thermal comfort are considered: window orientation comes in first at 58.12%, followed by window size (24.25%) and shading (6.85%). The air-conditioning load needed to maintain indoor thermal conditions is highly affected by glazing property (29.36%), window orientation (26.52%), window size (14.79%), window type (11.44%), and the availability of external shading devices or other useful shading means (8.09%). Thus, the role of solar radiation is crucial to indoor thermal comfort, as well as AC loads, particularly in the absence of solar shading. Window location and aspect ratio have a lesser influence compared to other design variables, in which the percentages of contribution were calculated at 3.72% and 6.08%, respectively.


**Table 11.** Factor effect percentages to acceptable hours of VR.

**Table 12.** Factor effect percentages to acceptable hours of CO2.


**Table 13.** Factor effect percentages to acceptable hours of adaptive TC.



**Table 14.** Factor effect percentages to AC loads.

After determining the percentage contributions of each design parameter using the ANOVA approach, the signal-to-noise ratio method was then used to identify the most appropriate factor levels, thus obtaining the near-optimal design scenarios that can support early design decision-making.

#### *4.2. Identifying Optimal Design Alternative*

Using the signal-to-noise ratio method, the most significant level combinations of each design parameter were determined, which represent a near-optimal design scenario. Although this level combination does not necessarily correspond to the optimal case, as discrete levels of the parameters were implied in the analysis.

Figure 7 shows the S/N ratios for the tested design variable levels relative to each intended performance criterion. Figure 7a–c is based on the signal-to-noise of greateris-better, while Figure 7d applies smaller-is-better. By observing Figure 7a,b, it can be seen that the optimal level combinations for both ventilation rate and CO2 performance are almost similar, specifically for the factors that represent the most influential variables, confirming the direct proportionality relationship between the amount of delivered airflow and indoor air pollutants. For **ventilation rate** performance, the optimal level combinations are as follows:


Concerning **carbon dioxide concentration**, similar level combinations are preferred, except for the glazing property and aspect ratio, in which triple glazing and square windows show better results for the performance of this criterion. By looking at the S/N ratio plot of **thermal comfort**, shown in Figure 7c, the selection of optimal level combinations is as follows:


**Figure 7.** Signal-to-noise (S/N) ratio plots showing the effectiveness of each parameter and optimal factorial levels for (**a**) ventilative rate, (**b**) CO2 concentration, (**c**) adaptive thermal comfort, and (**d**) AC loads.

The S/N ratio plot of the studied variable levels relative to the performance of **mechanical air-conditioning loads**, shown in Figure 7d, indicates significant differences compared to the performance of the rest of the criteria. Discovered by analysis of variance, the most influential variable was glazing property, followed by window orientation and size. The optimal level combinations include:


#### *4.3. Trade-off Selection Based on Near-Optimal Level Combinations*

In the multi-objective optimisation approach, the near-optimal level combinations are prescribed by selecting trade-offs between distinct objective functions. Consequently, the most effective level combinations (trade-offs) and their overall performance results for each criterion are outlined in Table 15, followed by their visual illustration in Figure 8.

Based on the S/N ratio results, the trade-off window orientation is south-facing windows with square shapes placed in the middle of external walls. Offices with small windows normally require less energy demand; however, larger-sized windows were found to be the most appropriate scenarios when consciously designed by considering optimal factorial level combinations. For reference, trade-off options 1 and 6 had the same window design features, but a larger-sized window (50% WFR) was assigned to the former, and a smaller window (20% WFR) was provided for the latter; thus, the MM supplementary loads were recorded at 11.66 kWh/m2 and 12.94 kWh/m2, respectively. Consequently, the larger-sized window can be a considerably more energy-efficient solution by 10.4% compared to the 20% WFR. In addition, large windows can have a better outside view and aesthetic appearance, while visual comfort risks can be eliminated or lowered using a novel solar shading design.


**Figure 8.** The selected trade-off options for a detailed study of the intended performance criteria.

The same window design characteristics were applied to options 1 through 4, although window types varied. Double-hung windows offer the best possible results for each performance criteria, followed by sliding, casement, and single-hung windows. Such a window design, with trade-off option 1 attributes, provides 72.3% of occupancy hours inside Category II ventilation rates, an 83.7% CO2 concentration level below the WHO threshold (1000 ppm), and a 70.2% adaptive comfort Category II, and it maintains indoor conditions for 29.8% of hours; altogether, an annual AC load of 11.66 kWh/m<sup>2</sup> is needed. Since double-hung and sliding windows allow effective air circulation, particularly in both the wind-driven and buoyancy effects, natural ventilation might occur through doublehung windows. These results are tangible evidence that needs to be considered by architects when making early decisions concerning the window design of offices in the Mediterranean region and similar climatic conditions.

Shading negatively affects NV performance relative to VR and CO2 concentration performance, as can be seen in trade-off option 5, which performs better than the previous design scenarios. Nevertheless, solar shading improves indoor thermal comfort and reduces AC loads. In this situation, a double glass window with low-E coating can be more profitable than triple glass. Conversely, if shading does not exist, a triple glass window is essential if high-performance offices are intended.

218

**Table 15.** Results of the trade-off design solutions for different window design parameters.

#### *4.4. Results of Airflow Rates*

Table 15 reports the total annual number of hours at which the ventilation rates, for both occupancy and building pollution, were higher than the lower limit of Category II (VR ≥ 2.1) for the selected trade-off designs. Despite the constant window size (50% of floor area) and other window design features (apart from window type) assigned to scenarios for trade-off options 1–4, the double-hung window provided more acceptability hours (1511 h) of VR than sliding (1502 h), casement (1501 h), and single-hung (1398 h) windows. Therefore, double-hung windows facilitate effective NV to allow fresh air to enter the space, while sliding and casement windows perform similarly relative to airflow rates.

The optimal design solutions for each of the double-hung, sliding, casement, and single-hung windows offer 72.3%, 71.9%, 71.8%, and 66.8% Category II ventilation rate hours annually during office working hours. Due to cold outdoor conditions, which keep windows closed most of the time, January and February recorded lower airflows than the threshold. Therefore, a minimum airflow rate for acceptable indoor air quality needs to be provided using mechanical ventilation, or alternatively, windows should be opened regularly for a short time to replace exhausted indoor air. In general, the window aspect ratio had a minimal impact on the airflow performance; nevertheless, longitudinal (e.g., rectangle) windows were found to be better than the square shape.

Figure 9 shows the monthly ventilation rates for the trade-off designs selected through the analysis of variance and signal-to-noise ratio approach, in which the double-hung, sliding, and casement windows can accomplish Category II minimum amounts of ventilation rates for all months except January and February using the proposed window-opening scheme and MM cut-off temperature. By comparing the VR of trade-off option 3 to trade-off option 5, one can see that external solar shading (in this case, horizontal fins) reduces the NV potential for the airflow rate by 4.8%, but it can simultaneously enhance ambient air's ventilative cooling potential. The amount of VR reaches 10 L/s·m2 in the spring and autumn months, when windows are open during most of the occupancy hours; thus, the ventilative cooling potential of ambient air facilitates passive cooling. Finally, the small-sized window, namely, 20% WFR, offers 1235 h of Category II VR, 29.2% less effective in bringing fresh air indoors compared to the same window design inputs in the large window (i.e., 50% WFR).

**Figure 9.** Monthly airflow rates for the studied trade-off design scenarios.

#### *4.5. Results of Carbon Dioxide Concentration*

The number of hours for which the level of CO2 concentration is below 1000 ppm during occupancy time is presented in Table 15. The shaded double-hung window (50% WFR) provides around 1749 h out of 2088 h per annum, corresponding to approximately 83.7% of the time. Moreover, sliding, and casement windows offer approximately 83%, while the single-hung window provides 78.1% of the office hours within the CO2 threshold. The mixed-mode cut-off temperature of 31.7 ◦C closed the windows during the harsh summer days, which resulted in increased CO2 concentration. In the warm and cool periods, the average carbon dioxide concentration was below the WHO threshold. However, when the windows are closed during office working hours, the level of CO2 concentration exceeded the recommended threshold. For example, CO2 concentration rose to over 1400 ppm in July and August when the office window was closed all the time due to hot outside temperatures, regardless of the window type, as shown in Figure 10.

**Figure 10.** The level of carbon dioxide concentration for the studied trade-off scenarios.

Different window types offered similar results in terms of indoor CO2 concentration. Conversely, window size had a significant effect on the level of carbon dioxide concentration; for instance, a 20% WFR can only provide 69.0% (1441 h) compared to the same scenarios for a large-sized window (83.7%). In addition to a high indoor concentration in the warm months, a small-sized window can cause health-related problems in the cold months. Overall, larger window sizes, with greater opening fractions, allow more airflow to enter indoors, which can lower the level of CO2. The availability of solar shading does not make a considerable difference in regard to CO2 contamination, such as in the case of trade-off options 1 and 5.

#### *4.6. Results of Adaptive Thermal Comfort*

In this research, the potential of natural ventilation alone for thermal comfort was studied and reported, and TC during air-conditioning hours was excluded. In other words, the discomfort hours require the operation of mechanical air-conditioning within the MM system. By looking at Figure 11, specifically trade-off options 1–5, the total annual number of comfort hours through NV reaches 90%, meaning that the NV strategy can provide acceptable comfort conditions for nearly all the occupancy time in the cold period. In the

other words, these months constitute a free-running period. In June and September, it can cover approximately 40% to 60% of the office working time. However, the minimum number of comfort hours can be found during July (less than 10%) and August (less than 15%) in the summer. Therefore, the AC mode should be working most of the time during July and August compared to the other months.

**Figure 11.** Monthly percentages of comfort hours based on the Category II adaptive comfort limits for the studied trade-off design scenarios.

Nearly all window types with double glass coated with low-E and shading offer similar thermal comfort hours—for reference: double-hung, 70.2%; sliding, 69.5%; casement, 66.04%; and single-hung, 65.8%. In addition, triple glass without shading can offer identical results with a small difference, such as with a casement window at 66.9%. However, a smallsized window (i.e., 20%) can only provide 57.2% comfort hours during office occupancy time. Window location does not have a significant effect on indoor thermal comfort, while a window with an aspect ratio of 1:1 performs better than a window with a 1:2 proportion. Figure 12 illustrates the scatter plot of hourly indoor operative temperature in accordance with an outdoor running mean temperature for each month, employing the Category II upper and lower limits of the BS EN 15251:2007 standard for the optimal design scenarios: (a) O-1 and (b) O-6 (large and small windows, respectively). The hours appearing in between both limits represent the acceptable thermal comfort hours for Category II. The hours exceeding the upper limit correspond to the "too warm" hours in the summertime, particularly in July and August, while those below the lower limit are "too cool" hours in the winter occupancy time.

**Figure 12.** Hourly indoor operative temperature for the adaptive comfort Category II in the case of (**a**) O-1 50% WFR and (**b**) O-6 20% WFR.

#### *4.7. Air Conditioning Loads of the Mixed-Mode Strategy and a Fully Airconditioned Case*

The operation of air-conditioning within the mixed-mode system began when the indoor operative temperature was higher than 31.7 ◦C in the warm period or lower than 20 ◦C in the cool period. These approximately correspond to the upper and lower boundary limits of Category II in the British/European adaptive comfort standards. All the design variables affect AC loads as well as different factorial levels. Generally, the north façade receives a lesser solar ratio; thus, a lesser amount of air-conditioning loads will be required, especially in the absence of solar shading in the cases of the other window orientations that receive more annual solar radiation. Hence, the S/N ratio showed that smaller windows might spend less on MM air-conditioning compared to unshaded large-sized windows.

Large windows (i.e., 50% WFR) with double-hung, sliding, or single-hung properties are more energy-efficient solutions than windows with size a 20% window-to-floor ratio, as well as with respect to the other studied criteria. A 50% WFR with a double-hung

shaded square window located in the middle of the wall and double glass low-E utilises 11.66 kWh/m<sup>2</sup> annually, whereas a 20% WFR, having the same design variables as the large-sized window, needs a 12.94 kWh/m2 AC load per annum. However, a large-sized shaded casement window with double glass low-E seems to be an inefficient window type in relation to AC load, requiring 14.94 kWh/m<sup>2</sup> annually, which is even more than the unshaded casement window with triple glass (14.56 kWh/m2). When a designer does not apply a solar shading device, a high-performance window property (e.g., triple glass) must be used to achieve results nearly equal to a shaded window with a higher glazing U-value. Regardless of the window size, glazing property, location, or proportion, windows in southern and the northern external walls constitute the most efficient window orientations; therefore, these windows allow a greater amount of natural ventilation to be harnessed, thus facilitating less dependence on active AC systems.

The monthly air-conditioning loads for trade-off design scenarios are presented in Figure 13. High outdoor running mean temperatures cause elevated indoor operative temperatures in July and August, in which the Category II upper limit, 31.7 ◦C (cooling setpoint), is surpassed during most office working hours; thus, the maximum AC loads were recorded in these months. In nearly all the design scenarios, the cool period represents the free-running (no mechanical systems in operation) months, while in the rest of the months, both the natural ventilation and air-conditioning modes of the mixed-mode system were alternated. Unshaded high-performance windows (trade-off option 5) and shaded small-sized windows (trade-off option 6) utilise a small amount of AC load in the cool months. Conversely, if the air-conditioning is controlled by the adaptive comfort upper and lower limits of a particular category, the results might not be identical to the previous cases. This study used constant cooling and heating setpoints for the activation of AC; this was due to the limitations of the current dynamic simulation software. In this case, the results of the "comfort hours" indicator can better define the free-running hours. Overall, double-hung and sliding windows are more efficient window types than single-hung and casement windows, respectively.

**Figure 13.** Monthly AC loads for the trade-off design scenarios.

In order to assess the performance of the mixed-mode system against a fully airconditioned scenario, the air-conditioning loads of the O-1 trade-off design were compared to an identical design case with a fully AC system, using 20 ◦C and 26 ◦C for the heating

and cooling temperatures, respectively, as suggested in Category II of the BS EN 15251:2007 standard, illustrated in Figure 14. In July and August, the fully AC scenario used more than 11.0 kWh/m2, nearly 7.0 kWh/m2 more compared to the MM system. In the heating season, particularly January, February, and March, both MM and AC systems performed similarly due to assigning the same heating setpoint temperature (i.e., 20 ◦C) to both systems, although the fully AC system consumed more energy. The total annual cooling and heating loads for the fully AC and MM cases were 56.63 and 11.66 kWh/m2, respectively. Accordingly, the mixed-mode system can reduce cooling and heating loads by 79.41% compared to a fully AC cellular office, taking into account the design specifications of the O-1 trade-off design in the climatic conditions of Famagusta, North Cyprus. An almost similar reduction in air-conditioning loads was also observed in the results of a field study [77], in which the mixed-mode office consumed less than a quarter of the energy required by a similar fully air-conditioned space; a nearly 45% reduction was reported in another study [7].

**Figure 14.** Monthly AC loads for the O-1 design scenario in mixed-mode and fully AC systems.

#### **5. Conclusions**

This study presented a performance-based window design and evaluation model for NV and MM offices. The applicability of the proposed model was tested on the window design of a naturally ventilated single office with additional cooling and heating (mixedmode conditioning) in a Mediterranean climate. Multiple window design variables and levels were assessed using the Taguchi orthogonal array, ANOVA analysis, and S/N ratio approach, which are suggested in the model. The investigations included the study of window size, orientation, window type, glazing property, aspect ratio, location, and window shading in relation to the potential of NV to achieve acceptable indoor air and thermal comfort with significantly reduced air conditioning loads using a mixed-mode strategy. Suggested in the model stages, an hourly dynamic simulation method was utilised to measure the CO2 concentration levels, airflow rate, adaptive thermal comfort, and cooling/heating loads, taking the hours in which a specific criterion was satisfied as the calculated indicator. The analysis of variance results revealed the effectiveness of each variable on the selected performance criteria, as stated below.

#### *5.1. Contribution of Window Design Parameters to Airflow Rate and CO2 Concentration*

• Window size was in the first rank or scored the highest percentage of contribution (81.59% and 73.54%, respectively), followed by window orientation and type.

#### *5.2. Contribution of Window Design Parameters to Adaptive Thermal Comfort*

• Window orientation plays a vital role in providing comfortable indoor conditions, with a percentage of contribution of 58.12%. Window orientation is significantly correlated with the position of the sun and the direction of the wind, determining the amount of air and solar radiation permitted into the space.

#### *5.3. Contribution of Window Design Parameters to the Supplementary Air Conditioning Loads*


Accordingly, trade-off designs with near-optimal combinations were selected and further studied. The outcome of the O-1 trade-off design revealed that the ventilation rate met the minimum VR ≥ 2.1 for approximately 72.3% of the annual office working hours. The level of CO2 concentration did not exceed the 1000 ppm threshold for 83.7% of the time. The indoor operative temperature was within the Category II temperature ranges of the adaptive comfort approximately 70.2% of the occupancy time, constituting the free-running period, while air-conditioning was required for the remainder of the time to sustain indoor thermal comfort conditions, requiring 11.66 kWh/m2. Up to 90% of the office working hours in January, February, March, April, May, October, November, and December constitute the free-running period based on the number of comfort hours designated by the BS EN 15251 standard adaptive model.

Conversely, as a result of the elevated outdoor air temperature, ventilative cooling could only offer 5–15% adaptive comfort hours in July and August, as well as 40–60% in June and September. Nonetheless, the mixed-mode system resulted in a 79.41% reduction in cooling/heating loads relative to a fully air-conditioned scenario, considering the conditions of this study. The reduction in air-conditioning loads is also similar to the results reported by a reviewed field study.

**Author Contributions:** Conceptualization, H.K.A. and H.Z.A.; methodology, H.K.A.; software, H.K.A.; validation, H.K.A. and H.Z.A.; formal analysis, H.K.A.; investigation, H.K.A.; resources, H.Z.A.; data curation, H.K.A.; writing—original draft preparation, H.K.A.; writing—review and editing, H.K.A. and H.Z.A.; visualization, H.K.A.; supervision, H.Z.A.; project administration, H.Z.A.; funding acquisition, H.Z.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Article*
