An Airflow Output Control to Maintain a Constant Body Heat Loss During Sleep on Temperature-Changing Nights: Implementation in a Ventilated Sleep Capsule

Abstract

Good sleep is essential for a healthy life. While airflow improves sleep in a hot environment, it may cause an excessive drop in body temperature because thermal-adaptive behavior is inactive during sleep. This study aims to propose an airflow control theory that prevents the excessive drop in body temperature while maintaining good sleep. The theory changes the heat transfer coefficient between the skin and the environment by the intensity of the fan operation to maintain a heat loss of 30 W/m² from a body with a skin temperature of 34.5 °C in a temperature-changing environment. We fabricated a ventilated sleep capsule in which this theory was embedded. Thermal manikin experiments were conducted to obtain the relationship between sensible heat transfer coefficients and the fan operating signal to establish control equations. We conducted a case study to evaluate whether the theory provided better sleep than sleeping with a regular fan running freely in homes in Ueda City, Japan, in one summer. Although the data used in the analysis were from only three subjects, the statistical analysis showed that sleeping in the ventilated sleep capsule provided better sleep, with a mean of 6% better sleep efficiency.

1. Introduction

We spend about one-third of our life sleeping. Problems with sleep cause decreased vitality, decision-making speed, and efficiency in daily life [1,2]. In terms of health, the problems can also lead to coronary heart disease and impaired immune function, cognitive function, and memory [1,3] and can also trigger heat disorders [4]. Good sleep is essential for a healthy and quality life.

Regarding the relationship between sleep and thermal environment, especially high-temperature environments, Okamoto-Mizuno et al. [5] conducted sleep experiments under four conditions: 29 °C 50% RH, 29 °C 75% RH, 35 °C 50% RH, and 35 °C 75% RH. They reported that the high-temperature and -humidity environment disturbed the body core temperature dropping, reduced slow wave sleep and rapid eye movement (REM) sleep, and promoted arousal, resulting in decreased sleep quality. Okamoto-Mizuno and Tsuzuki [6] measured sleep and the thermal environment in elderly homes in the summer, winter, and autumn. They then confirmed that sleep was most interrupted in the summer. Based on the high head skin temperatures measured in the summer, they suggested that core body temperature reduction was suppressed during summer sleep. Chevance et al. [7] reviewed the literature that met the meta-analysis guidelines on sleep and environmental temperature. Then, they reported that a high-temperature environment causes poor sleep, especially in elderly people living in hot areas and low-income districts. With global warming, heat islands, and population concentration in urban areas, outdoor climates are bound to become hotter and hotter. Therefore, it is critical to identify practical countermeasures to poor sleep for people who are socially and economically impoverished and do not have air conditioners.

Airflow can improve sleep in a hot environment. Lin and Deng [8] have shown how much airflow increases the indoor comfortable operative temperature. Xu and Lian [9] also reported that different combinations of air temperature, bedding, and airflow can lead to the same body temperature during sleep. Aijazi et al. [10] reported that airflow and electrically heated blankets are effective in reducing energy consumption in heating, ventilation, and air conditioning (HVAC) systems, based on an evaluation through thermal manikin experiments. Tsuzuki et al. [11] conducted sleep experiments in 26 °C 50% RH 0.2 m/s, 32 °C 80% RH 1.7 m/s, and 32 °C 80% RH 0.2 m/s environments. The 32 °C 80% RH environment with 1.7 m/s airflow was equivalent to the 26 °C 50% RH 0.2 m/s environment in terms of the sleep stage assessment. However, rectal and skin temperatures in the 32°C 80% RH 1.7 m/s environment were intermediate between those in the 26 °C 50% RH 0.2 m/s environment and the 32 °C 80% RH 0.2 m/s environment. They concluded that airflow can contribute to good sleep in hot and humid environments. Since airflow effectively mitigates hot discomfort during sleep, there are commercial products that utilize airflow, such as fans with shut-off timers and direct current (DC) fans with finely tuned outputs in response to air temperature change. Mattresses with airflow features are also commercially available.

Meanwhile, Morito et al. [12] compared sleeping with air conditioners with an average air velocity of 0.14 m/s and a maximum air velocity of 1.1 m/s with those with an average air velocity of 0.04 m/s and a maximum air velocity of 0.3 m/s in an environment of approximately 26 °C and 55% RH. Although there was no significant difference in the duration of each sleep stage, focusing on the time of the blow, the authors reported that the sleep stage shifted toward arousal when the air conditioner with the stronger blow was operated. These results showed that even in environments with average wind speeds of less than 0.2 m/s, high wind speeds had a negative effect. Akiyama et al. [13] reported that while airflow reduced sleep efficiency, there was no difference in the number of awakenings due to airflow. They then showed that the effect of airflow turbulence on sleep depended on temperature, with less effect at higher temperatures.

In addition to using airflow, there are other methods of adapting to hot environments: opening windows and adjusting clothing (during sleep, this is similar to adjusting blanket coverage) [14,15,16]. Opening windows makes the indoor environment more sensitive to the outdoor weather. As the outdoor temperature decreases in the early morning, the indoor operative temperature also decreases. Regarding the human body side, Ohnaka et al. [17] reported increased body movement in hot environments in the summer as a body response to the environmental temperature change. Wang et al. [18] also reported adjusting the blanket coverage to approach thermal neutrality. Here, we consider the combined effects of airflow utilization, bedding coverage adjustment, and a decrease in outdoor temperature. The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) [19] recommends that the use of airflow should be limited if the occupants are unable to adjust the airflow themselves, even for occupants who are awake and capable of adaptive actions. Since occupants cannot adjust airflow during sleep, airflow utilization during sleep must be applied carefully. Here, we considered the combined effects of airflow utilization during sleep, bedding coverage rate adjustment, and indoor temperature reduction based on the illustration describing the heat interaction among the human body, clothing, and environment using a difference in water head and flow rates [20]. Figure 1 shows the interaction modified for sleep to include bedding.

When the blanket removal and airflow cooling are applied simultaneously (valves [a] and [b] in Figure 1 are open), the total thermal resistance between the skin (water head [B]) and the indoor environment (water head [C]) decreases. In this case, the decrease in the indoor operative temperature (water head [C]) intensifies the heat transfer between the human body (water head [B]) and the indoor environment (water head [C]). Therefore, continuous airflow use (opening valves [a] and [b]) during sleep results in an excessive drop in body temperature (water heads [A] and [B]) when the indoor operative temperature (water head [C]) drops in the early morning. Since the activity of the human body’s immune system decreases as the body temperature drops [21], the excessive drop makes the body more likely to catch infections (e.g., a summer cold). While people may avoid the excessive drop, the quality of sleep must deteriorate significantly. Okamoto-Mizuno and Mizuno [22] reported that sleeping naked in a low-temperature environment seriously deteriorates sleep quality, although it is usually not a serious problem because people can avoid sleeping in such a condition by using blankets. However, in regard to the defect caused by the indoor operative temperature drop during sleep, people cannot add blankets as a countermeasure. Lin and Deng [8] reported that the use of airflow during sleep should be limited because people cannot consciously perform adjustment behaviors during sleep. This limitation may be explained by the severe effect of the body temperature drop associated with airflow utilization.

Applying a methodology based on heat transfer is effective for analyzing the relationship between the thermal environment and sleep. Xu and Lian [9] analyzed sleep by calculating the heat transfer between the skin and the environment, considering the uneven distribution of hair, pajamas, and bedding. They demonstrated the close relationship between sleep and the calculated heat transfer and proved the effectiveness of taking heat transfer into account when analyzing the relationship between the thermal environment and sleep quality. In addition, various methods have been reported to determine a comfortable thermal environment during sleep by modifying Fanger’s comfort equation or the predicted mean vote (PMV) [8,23,24,25]. A method to control the set-point temperature of the air conditioner using the PMV, evaluated based on heat transfer, has also been proposed [26]. This method took individual differences into account. However, no method has been established to control airflow during sleep based on heat transfer.

The main objectives of this study are to propose a theory of controlling the airflow output to maintain a constant body heat loss in a temperature-changing environment and to show a way to implement this theory. This theory corresponds to controlling the opening of two valves [a] and [b] between the skin and the indoor environment (water heads [B] and [C]) in Figure 1 to maintain a constant flow rate at a constant skin temperature (water head [B]) in the temperature-changing indoor environment (water head [C] changing condition). The focused points of our control theory are highlighted in red in Figure 1. In this paper, we describe (1) the derivation of the control theory, (2) the fabrication of a sleep ventilation capsule into which the theory is embedded, (3) the implementation method of the control theory, (4) the effect confirmation through a human case study in the summer, and (5) the positioning of the result and further expected developments. The above background and research gap of this study are summarized in Figure 2. If we can obtain good sleep by controlling the airflow to keep the body’s heat loss constant during sleep in response to fluctuating temperatures, this could be a breakthrough that solves the problem of airflow utilization during sleep. This theory encourages the establishment of an implementable solution to the deterioration of sleep quality for low-income people who do not have access to air conditioning. In this study, we propose a method for controlling airflow during sleep based on heat transfer and verify its effectiveness.

2. Methods

2.1. Airflow Control Theory

Equation (1) expresses the sensible heat loss from the human body to the indoor environment using the heat transfer coefficient between the skin and the environment (h^*).

Q = h^*(t_s − t_o)

(1)

where Q: sensible heat loss per unit skin area [W/m²], h^*: heat transfer coefficient between skin and indoor environment [W/(m²·K)], t_s: skin temperature [°C], and t_o: operative temperature [°C].

The metabolic heat during sleep is reported to be about 0.7 met (40.7 W/m²) [19]. If the same amount of heat as this metabolic heat is lost during sleep at neutral skin temperature, the body can maintain thermal equilibrium in a thermally neutral state. In thermally neutral sleep, we assumed that 20–30% of the metabolic heat is removed by respiration and perspiration, and the remaining 70–80% is removed by sensible heat transfer processes: convection, radiation, and conduction. The airflow control theory proposed in this study controls the sensible heat loss at 30 W/m², corresponding to 70–80% of the metabolic heat, so as not to cause excessive drops in body temperature during sleep. Based on a literature review, Xu and Lian [27] reported 32.8–35.4 °C as an average skin temperature during sleep. In addition, Rohles and Munson [28] reported that the skin temperature in a 21.1 °C environment where their subjects slept well was 34.5 °C. Nagano et al. [29] also reported 34.5 °C as the thermally neutral skin temperature during sleep. Therefore, we used 34.5 °C as the neutral skin temperature during sleep. For the conditions where the t_o changes, the h^* that maintains the thermal equilibrium in the thermally neutral state can be expressed by Equation (2).

h^* = 30/(34.5 − t_o)

(2)

In this study, we replace the t_o in Equation (2) with a measured air temperature (t_a) assuming that the t_a is almost equal to t_o in the environments where airflow can be utilized. Then, we proposed an airflow control method that measures t_a, determines the target h^* value by Equation (2), and controls the airflow to achieve this h^*. Figure 3 summarizes the structure of this study from the theory proposal of Equation (2), theory implementation processes, and case study to confirm its effectiveness.

2.2. Ventilated Sleeping Capsule

A ventilation capsule implementing this airflow control theory was fabricated using a one-person tent to provide a wind path (Figure 4). An air intake opening was placed at the head end of the tent, and exhaust fans were installed at the foot end to provide airflow from head to feet. A three-dimensional air-permeable cushion (Figure 5) was placed inside the tent as a mattress. As the exhaust fans, three silent fans, whose speed can be controlled by pulse width module (PWM) signals, were installed at the foot end of the tent to extract air and two at the foot end of the mattress (Figure 6). Two fans at the foot end of the mattress created an airflow path at the rear. A blanket was placed over the top of the tent to provide thermal insulation when the ambient temperature dropped. When the fans were not operating, the thermal insulation of the blanket reduced the h^*. A slatted grid was used as the base of the tent.

During sleep, there is no active body motion. Therefore, we evaluated the relationship between h^* and the PWM signal sent to the fan based on motionless thermal manikin experiments. We conducted the experiment in an artificial climate chamber maintained at 20 °C. The humidity was not controlled. The airflow in the artificial climate chamber was the minimum necessary to control the temperature of the chamber. We set the surface temperature of the thermal manikin at 34.5 °C. A short-sleeved shirt, shorts, and underwear were selected as a representative summer sleep ensemble, and a bath towel was used as a pillow. The six-step PWM signals from minimum to maximum were applied to all the fans. From the steady-state data, we found that this ventilated sleep capsule can adjust h^* from 3.2 to 9.3 W/m²K. From the h^* range, it can be concluded that the ventilated sleep capsule was capable of maintaining the thermal equilibrium of the human body in the thermally neutral state at a t_a between 25.2 and 31.3 °C.

We used a programmable microcomputer to measure t_a, calculate the PWM signal according to our control theory, and output the PWM signal to the five exhaust fans. To measure t_a, we placed an SHT31 sensor module, whose accuracy was ±0.3 °C, at the top of the space inside the one-person tent. The microcomputer calculated the target h^* from the measured t_a based on Equation (2). Then, the microcomputer calculated the PWM signal resulting in the target h^* based on the relationship between h^* and the PWM signal determined by the thermal manikin experiments. Finally, the microcomputer outputted the PWM signal to the five silent DC fans. This process was repeated every 10 s.

2.3. Field Survey and Analysis Method

We conducted a human case study to confirm the effectiveness of the airflow control theory using the fabricated ventilated sleep capsule. An experiment was conducted from 9 August to 16 September 2021. Seven males in their twenties (age: 23.3 ± 1 years, height: 171.3 ± 4.6 cm, and weight: 65.2 ± 4.5 kg) without sleep disorders participated in the experiment. We used the Pittsburgh Sleep Quality Index (PSQI) [30] to confirm that the subjects did not have sleep disorders. All the subjects lived in Ueda City, which is located in central Japan. Four ventilated sleep capsules were prepared for the experiment. The experiment was conducted in two sessions: three subjects for the first and four for the second. The ventilated sleep capsules were distributed to each subject’s home, and the experiment was conducted at each subject’s home. We investigated two conditions: the fabricated ventilated sleep capsule in operation (hereafter referred to as ‘condition V’) and the use of a fan and towel blanket as a comparison condition (hereafter referred to as ‘condition C’). In condition C, the subjects were free to adjust the blanket coverage and the operation and location of the fan. Each subject assembled the ventilated sleep capsule on the night of condition V, or set up the bedding provided for condition C in his home. The experiment was conducted with the approval of the Ethics Committee for Human Subjects Research of Shinshu University (approval number 308). Prior to the experiments, we obtained each subject’s consent to participate by explaining the research procedures. The subjects received rewards for their participation. The experiment flow is shown in Figure 7.

The subjects wore underwear and the same T-shirts and shorts used for the thermal manikin experiment in both conditions V and C. They used the same bath towel as a pillow. Each subject was provided with washed T-shirts, shorts, and towels. The mattress used in condition C was the same as that placed under the ventilated sleep capsule in condition V. The fan used in condition C was a DC fan that could adjust the flow rate in seven steps and had functions to rotate the airflow direction and to fluctuate the speed; an off timer at 1, 2, or 4 h; and an on timer at 4, 6, or 8 h. In both conditions, the window opening and the operations of air conditioners and kitchen exhaust fans were prohibited for one hour before going to bed. However, toilet and bathroom fans were allowed to operate to maintain the house’s necessary ventilation. No methods were used to control indoor t_o during sleep. If the subject felt cold in condition V, he was allowed to use the towel blanket provided for condition C. If he felt cold in condition C, he was allowed to use the quilt provided for condition V. However, the nights when the quilt was used in condition C or the blanket used in condition V were excluded from the analysis because the nights were not considered hot summer nights.

The experimental period consisted of 16 consecutive days. However, 2 of the 16 days were designated as days when the subjects were free from the experiments. The first day was designated as the acclimatization night for condition V, and the second day was designated for condition C. Sleep time was set at 7 h and 30 min for all days. The same bedtime, determined by each subject, was set for the 14 days except the two free days. The experimental procedures were as follows.

(1)

Finish meals at least two hours before bedtime. Avoid caffeine consumption immediately before bedtime and alcohol consumption on the day of the experiment. Assemble the ventilated sleep capsule for condition V or the mattress, fan, and towel blanket for condition C in preparation for bedtime. There were no specific instructions for bathing or showering.

(2)

At 30 min before bedtime, change into the provided T-shirt and shorts and complete the pre-bedtime questionnaire described later.

(3)

At bedtime, go to bed and sleep.

(4)

Wake up after 7.5 h.

(5)

Complete the post-awakening questionnaire described later.

(6)

Change from the provided clothing to his own.

(7)

In principle, conditions V and C should alternate every other day. However, if the weather forecast predicted that there would be more hot or cold days for one condition, the authors specified the conditions for that day to reduce bias.

During sleep, sleep efficiency (SE), sleep latency, number of awakenings, wakefulness after sleep onset (WASO), and total sleep time were monitored using a wristwatch-type sleep monitor (Philips, Actiwatch2) worn on the subject’s non-dominant wrist. Data were recorded every 30 s. For the sensory ratings in

Table 1. Reported sensory rating items and their scales in addition to OSA.

	Before Sleep and After Waking Up			Before Sleep	After Waking Up
	Thermal Comfort	Cold–Hot Sensation	Sleep Feeling	Fatigue	Sleep Satisfaction
3	Very uncomfortable	Very hot	Very bad	Strong	Very satisfied
2	Uncomfortable	Hot	Bad	Moderate	Satisfied
1	Slightly uncomfortable	Slightly hot	Slightly bad	Weak	Slightly satisfied
0	Comfortable	Neither	Good	Not at all	Neither
−1		Slightly cold			Slightly dissatisfied
−2		Cold			Dissatisfied
−3		Very cold			Very dissatisfied

, the subjects completed questionnaires about thermal comfort, cold–hot sensation, and sleep comfort before sleep and after waking up. In addition, they responded to a questionnaire about sleep satisfaction and completed the Oguri–Shirakawa–Azumi (OSA) Sleep Inventory [31] after waking up. They reported a fatigue feeling before sleep. We also asked the subjects to freely describe their impressions of sleep before sleep and after waking up. All of these questions were on paper, and they completed all of the questionnaires while lying on the bed. A flashlight was provided so that they could complete the questionnaire in the dark. In condition V, they reported whether the towel blanket was used to cope with the cold, while in condition C, they reported how the fan was used and whether the quilt was used to cope with the cold after waking up.

Based on the meteorological data for Ueda City during the experiment [32], days with the lowest outdoor temperature less than or equal to 20 °C were excluded as non-hot summer nights. Data from subjects whose results were available for only one condition were also excluded from the analysis to eliminate the effects of individual differences.

We first performed the F-test to evaluate whether the variances were equal. Then, we applied the t-test for unequal variances to the items with unequal variances and the t-test for equal variances to the items considered to have equal variances. For both t-tests, we set the significance level to 5%. We used a two-tailed test when evaluating the difference between means, while we used a one-tailed test when assessing whether condition V provided better or worse sleep than condition C.

3. Results

3.1. Data Screening

A total of 98 data points were collected in the experiment. We applied the following four filters to screen the data before analysis. Filter 1: The first two days of the experiment, corresponding to the acclimatization nights for conditions V and C. Filter 2: The days that did not correspond to hot summer nights (days when the quilt was used for condition C or the blanket was used for condition V to cope with coldness). Filter 3: The days that did not correspond to hot summer nights (days for which the minimum outdoor temperature [32] was 20 °C or less), and Filter 4: Data from the subjects with only one condition remaining after the previous filters. In the second session of the experiment, which started on August 25, the outdoor temperature was already low, and no data passed the filters. After applying this screening criteria, 11 data points for condition V and 13 for condition C from only three subjects were retained.

Table 2. Data screening process to exclude cold nights.

Subject	Session	Condition	Sample Size After Filter
			Before Filters	Filter 1	Filter 2	Filter 3	Filter 4
A	1	V	7	6	3	3	3
		C	7	6	6	5	5
B		V	7	6	4	3	3
		C	7	6	4	3	3
C		V	7	6	6	5	5
		C	7	6	6	5	5
D	2	V	7	6	0	0	0
		C	7	6	3	2	0
E		V	7	6	0	0	0
		C	7	6	1	1	0
F		V	7	6	0	0	0
		C	7	6	6	0	0
G		V	7	6	3	0	0
		C	7	6	6	0	0

shows the processes of data filtering before data analysis.

Figure 8 shows the average outdoor temperature for each hour from 22:00 to 9:00 of the analyzed 11 data points for condition V and 13 data points for condition C. For each hour when subjects were asleep, we tested whether there was a significant difference in the average outdoor temperature between conditions C and V. As a result, p > 0.05 for all hours. Then, no significant difference was confirmed for the average outside temperature between the two conditions.

3.2. Sleep in Fabricated Ventilated Sleep Capsule Condition and Comparison Condition

Table 3. Average and standard deviation for OSA and other perceptions.

Items			Cond. V	Cond. C	One-Tailed p-Value
Before sleep	Fatigue (0: not at all, 3: strong)		0.7 (0.3)	0.8 (0.3)	0.26
After waking up	Thermal comfort (0: comfortable, 3: very uncomfortable)		0.3 (0.1)	0.6 (0.3)	0.07
	Cold–hot sensation (−3: very cold, 3: very hot)		0.0 (0.2)	0.1 (0.2)	0.29
	Sleep feeling (0: good, 3: very bad)		0.6 (0.1)	0.8 (0.2)	0.15
	Sleep satisfaction (−3: very dissatisfied, 3: very satisfied)		0.1 (1.2)	−0.5 (1.8)	0.10
	OSA (higher is better)	Drowsiness at waking up	43 (9)	44 (12)	0.47
		Good initiation and maintenance of sleep	44 (9)	43 (12)	0.45
		Less frequency of dreaming	56 (8)	55 (8)	0.37
		Refresh feeling	44 (9)	42 (10)	0.23
		Subjective sleep length	46 (8)	44 (11)	0.28

shows the mean and standard deviation of OSA, fatigue before sleep, thermal comfort rating, cold–hot sensation, sleep comfort, and sleep satisfaction after waking up for all the data, as well as the p-value of the one-tailed test. As the results of the F-test showed that the variances were equal for all the items, we applied the t-test for equal variances. For all the items, the mean values for condition V were either better or neutral compared to condition C. However, there was no significant difference in the one-sided test.

Table 4. Means of sensory ratings and sleep variables for each subject.

Subject	Cond.	n	Cold–Hot Sensation	Thermal Comfort	Sleep Feeling	Sleep Satisfaction	Sleep Latency [min]	Number of Awakenings [Times]	Wakefulness After Sleep Onset WASO [min]	Sleep Efficiency SE [%]
A	V	3	− 0.1	0.4	0.8	− 0.2	3	28	30	85.1
	C	5	− 0.1	0.6	0.7	− 0.6	12	31	48	80.9
B	V	3	0.2	0.2	0.3	0.7	7	18	13	95.2
	C	3	0.0	0.7	0.7	0.0	17	19	13	92.6
C	V	5	− 0.1	0.4	0.7	0.0	19	22	21	88.5
	C	5	0.3	0.5	0.9	− 0.8	48	32	40	80.4
Total	V	11	(See Table 3)				11	22	21	89.4
	C	13					27	28	37	83.4

shows the mean values of the thermal comfort, cold–hot sensation, sleep feeling, sleep satisfaction, sleep latency, number of awakenings, WASO, and SE for each subject. Condition V generally produced mean scores that were better than or equal to condition C for the thermal comfort, sleep satisfaction, sleep latency, number of awakenings, WASO, and SE. However, we could not conclude that condition V provided better sleep than condition C for the number of subjects.

Using all the data from three subjects, we evaluated whether condition V provided statistically better sleep than condition C. Figure 9 shows the sleep latency, number of awakenings, WASO, and SE of the three subjects. Only the number of awakenings could be assumed to have equal variance for conditions V and C. For the other items, equal variance could not be confirmed. The results of the t-test showed that p > 0.05 for the sleep latency (Figure 9a) and the number of awakenings (Figure 9b). We then concluded that sleep in condition V was not significantly better than in condition C in terms of sleep latency and number of awakenings. The results for WASO (Figure 9c) and SE (Figure 9d) were both p = 0.03, and we concluded that sleep in condition V was significantly better than sleep in condition C in terms of WASO by 16 min and SE by 6%. Subjects might have difficulty sleeping in condition C because controlling the airflow was difficult. The different variance and better mean WASO and SE of condition V suggest that condition C resulted in disturbed sleep on some nights and good sleep on some nights, whereas subjects in condition V, embodying the developed airflow control theory, usually slept well.

The subjects reported their impressions of sleep comfort in the ventilated sleep capsule. The most important problem related to airflow utilization during sleep was thirst or a dry throat. The airflow in the ventilated sleep capsule from the head to the feet suggested that sleeping in this capsule did not allow users to rebreathe exhaled humid air. Breathing only non-humidified ambient air could cause thirst or a dry throat. In this regard, changing the airflow inside the capsule is desirable so that users can rebreathe some of the exhaled air.

4. Discussion

Based on the heat balance of the human body and the determined relationship between the heat transfer coefficient and the PWM signal from a thermal manikin experiment, this study proposed an airflow output control theory for temperature-changing environments and indicated the process of implementing it in a system. In a case study conducted one summer in Ueda City, we found that condition V, the system in which the theory was implemented, produced statistically better sleep than condition C in terms of WASO and SE.

Lin and Deng [8] modified Fanger’s comfort equation and proposed a method to determine the combination of environmental factors and total thermal resistance that provides comfort during sleep. Fanger’s comfort equation describes the heat balance between the human body and the environment for the human body in a comfortable state [33]. Since the PMV, developed based on Fanger’s comfort equation, has been adopted in several standards [19,34] to evaluate thermal comfort, the human body heat balance is a fundamental theory for analyzing thermal comfort. As an application of the heat balance model to control the thermal environment during sleep, Song et al. [26] proposed a theory that learns the subject’s metabolic rate, the thermal resistances of clothing and bedding, and other individual differences and then established a method to control the set-point temperature of the air conditioner based on the PMV. Although they mentioned the possibility of controlling airflow and humidity with the same strategy, they did not indicate a method for controlling airflow during sleep. As values for evaluating the heat transfer of the human body during sleep, the total thermal resistance, including bedding [35], and the convective and radiative heat transfer coefficients of the human body in a sleeping position under a still airflow [36] have been reported. Here, we consider the effect of airflow based on the case of awake occupants by replacing bedding with clothing. Lu et al. [37] clarified that the thermal resistance of clothing changed depending on body movement and airflow, and the thermal resistance of clothing should be corrected accordingly. Although there is less body movement during sleep than during wakefulness, the thermal resistance of bedding still changes depending on the airflow. Therefore, the heat transfer from the human body under airflow during sleep is complex. During sleep, behavioral thermoregulation becomes inactive, making it difficult to respond to changes in the thermal environment. Thus, more precise automatic control of the thermal environment is required during sleep than during wakefulness. This complexity and the need for high accuracy have been a significant obstacle in deriving a theory for the automatic control of airflow output during sleep.

To solve these problems, in this study, considering that the airflow utilization is limited to the hot season, we adopted T-shirts and shorts as sleepwear. Then, without distinguishing convection, radiation, and conduction, we obtained a relationship between the heat transfer coefficient for the total sensible heat loss and the control signal of the fan operation by referring to the result of a thermal manikin experiment. While the applicable condition was limited to occupants wearing T-shirts and shorts, based on the relationship, we proposed a theory for controlling the airflow to keep the sensible heat loss constant during sleep in a temperature-changing environment. With this airflow control method, the airflow does not change much unless the air temperature changes suddenly. Therefore, this method can solve the problem of sleep stages shifting to arousal when a strong cold breeze blows, as reported by Morito et al. [12]. Since there is less body movement during sleep than during wakefulness, heat transfer coefficient evaluation using a thermal manikin is particularly effective for analyzing human body heat transfer. Although the cold weather during the experiments limited the data used in the analyses, we still confirmed that the ventilated sleep capsule with the newly developed airflow control provided statistically better sleep than when the fan was freely controlled.

While our theory is derived based on the human heat balance, the proposed theory can also be interpreted as a control to maintain a constant whole-body equivalent temperature (t_eq,whole) [38]. Equivalent temperature is one of the standard methods for evaluating the thermal environment of vehicles [38]. t_eq,whole evaluates the combined effect of t_a, radiation, and air velocity. The whole-body equivalent temperature is defined by Equation (3) using the whole-body skin temperature (t_sk,whole), the measured convective and radiative heat loss in the evaluated condition (Q_,whole), and the combined heat transfer coefficient determined in the calibration in a standard condition (h_cal,whole).

t_eq,whole = t_sk,whole − Q_,whole/h_cal,whole

(3)

According to the proposed theory, the fan operation is controlled to keep the t_sk,whole and Q_,whole at 34.5 °C and 30 W/m², respectively, for a temperature-changing environment. Then, once h_cal,whole is given by calibration, our control theory keeps the t_eq,whole constant. However, as with the characteristics of the equivalent temperature, the environment with a higher operative temperature than the temperature capable of maintaining the neutral thermal equilibrium of the human body causes sweat secretion, and discomfort due to humidity may occur.

As reported by Aijazi et al. [10], the use of airflow and electric blankets effectively reduces energy consumption in HVAC systems. Nagasawa et al. [39] developed an add-on sweating system for the local part of a dry thermal manikin. They presented an example of evaluating the evaporative cooling performance of a ventilated mattress. Song et al. [40] combined local foot heating with a ventilated mattress that removes contaminants as a countermeasure for local foot cold discomfort. They evaluated the effectiveness of this combination of ventilated mattress and local foot heating using human subjects. Similar to these examples, we foresee potential future research on cooling systems that utilize airflow during sleep. However, each system has a unique relationship between the total sensible heat transfer coefficient and the airflow control signal. Therefore, thermal manikin experiments for each system are necessary to implement the airflow control theory for the temperature-changing environment proposed in this study.

As for the mean SE of our case study, it was 89.4% for condition V and 83.6% for condition C. Even for condition V, the SE was lower than the reported mean SEs of more than 93% for the thermally neutral still-air conditions of Tsuzuki et al. [11] and Okamoto-Mizuno et al. [5]. In addition, Tsuzuki et al. [11] reported that the SE for the 32 °C 80% RH 1.7 m/s environment was almost the same as that for the 26 °C 50% RH 0.2 m/s thermally neutral still-air condition. However, the SEs of conditions V and C of our study were higher than the SE of 78.1% for the 32 °C 80%RH 0.2 m/s environment. This suggests that the experimental environments in each subject’s home of our study were better than the 32 °C 80% RH 0.2 m/s environment. However, it could not be concluded from these SEs that condition V did not provide as comfortable an environment as 32 °C 80% RH 1.7 m/s. SEs of 86.7% [41], 88.7% [42], and 89.1% [43] have also been reported for thermally neutral still-air conditions. It is still uncertain whether the 89.4% SE of condition V was worse than that of the thermally neutral still-air condition. Further studies are needed to evaluate this.

One expected application of the proposed airflow control theory would be its implementation in a DC electric fan. Air conditioners that control airflow based on a detected occupant position are already on the market. By implementing a similar technology in a DC electric fan, which is characterized by its fine flow rate control, we foresee the development of a fan that controls airflow according to air temperature and the relative position of the occupants to the fan to provide constant body heat loss. Another application is a DC ceiling fan [44]. The ceiling fan has a larger airflow area than DC electric fans, and the distance between the ceiling and the sleeping occupant is not much different. This means that the DC ceiling fan in which the theory is implemented can control the heat transfer coefficient h^* without detecting the occupant’s position. These fans would be an effective solution to the deterioration of sleep quality, especially for low-income people who cannot afford to install air conditioning while living in hot areas. Our theory could also be applied to cardboard beds used in evacuation shelters. The maximum power consumption of one unit of the silent DC fan used in our ventilated sleep capsule was only 1.1 W. When all fans of the ventilated sleep capsule are running at maximum, the power consumption is 6 W. This rate is equivalent to 48 Wh for 8 h of sleep. While power shortages sometimes occur after disasters, this amount of power can be covered by combining the power bank with solar power generation during the day. In addition, by changing the shape of the cardboard bed, it is possible to create an airflow path inside the bed.

Here, we describe the limitations of this study and topics for future research. Due to the weather conditions, data from only three subjects were considered data from a hot summer night. Statistical tests showed that sleep in the ventilated sleep capsule with the proposed airflow control theory was significantly better than sleep with the free operation of a regular electric fan. However, since only three subjects were involved, further experiments with a larger number of subjects are desirable to verify whether the theory can be generally applied. In addition, we could not record the indoor thermal environment during the experiment due to equipment malfunction. As a result, we could not indicate the appropriate indoor environment range for our theory based on the measurement with human subjects. The conditions in which airflow is utilized to alleviate hot discomfort are various. It ranges from the summer of the temperate zone, which has four seasons, to the year-round heat of the tropics. The change in thermal environment at night varies depending on the regional climate. The case study was conducted only in Ueda City, whose central part’s altitude is around 450 m. It is then necessary to confirm whether the method can be applied to residents in other regions. In addition, the thermal manikin experiment requires a steady-state measurement. Although it is not common to use thick bedding when applying airflow for cooling, bedding with a high heat capacity, such as winter bedding, is unsuitable for our airflow control theory, because heat storage continues for a long time and causes difficulty in establishing a steady state. Also, the sleep stage changes from the beginning of sleep to waking up. Based on this change, Lan et al. [45] proposed a strategy to change the indoor environment to achieve a better night’s sleep. Yu et al. [46] also proposed a theory to change the airflow output by sleep periods for a fixed temperature environment. The heat dissipation control proposed in this study continuously controls the amount of heat loss from an occupant at a constant (30 W/m²). It was not a control that provided a better environmental change based on the shift in the sleep stage. In turn, the heat transfer characteristics determined by the thermal manikin did not take into account individual differences. In the future, it is better to incorporate a learning system that takes individual differences into account, as suggested by Song et al. [26]. In addition, as an issue related to the utilization of airflow, some subjects complained of discomfort related to thirst. Thirst is a problem that can occur with other systems that utilize airflow during sleep.

5. Conclusions

In this study, we proposed a theory for controlling airflow in response to a changing air temperature to maintain a constant body heat loss for achieving good sleep in a hot summer. To demonstrate whether this control theory provides good sleep, we fabricated a ventilated sleep capsule and conducted experiments on human subjects in their homes in Ueda City, Japan. The control theory, the process of implementing this theory, and the study to confirm its effectiveness are summarized below.

The theory of airflow control is to adjust the heat transfer coefficient between the skin and the environment by changing the airflow. By doing so, the sensible heat loss of the body with a skin temperature of 34.5 °C is kept constant at 30 W/m² in a temperature-changing environment.

We fabricated a ventilated sleep capsule that sucked in air at the head end and exhausted it at the foot end. Silent DC fans with PWM control were installed at the foot end of the capsule.

We conducted a thermal manikin experiment for the ventilated sleep capsule with multi-stage fan operation. Based on this experiment, we determined the relationship between the fan control PWM signal and the sensible heat transfer coefficient between the skin and the environment.

Because of the cold weather conditions during the experimental period, the analyzed data for sleep in hot summer conditions were collected from only three subjects. However, the statistical analysis showed that the ventilated sleep capsule with the developed airflow control improved the subjects’ sleep compared to the free operation of a regular electric fan.

During sleep, behavioral thermoregulation is inactive. Therefore, increasing the heat transfer coefficient between the skin and the environment for body cooling will result in an excessive drop in body temperature when the ambient temperature drops from a comfortable level. For this reason, the airflow must be applied carefully during sleep.

Airflow utilization has great potential to reduce the energy associated with air conditioner operation and to mitigate hot discomfort for people who do not have access to air conditioners. Since human heat balance is a fundamental theory for analyzing thermal comfort, and the airflow output control during sleep proposed in this study is based on the human heat balance, the proposed theory can be a fundamental theory for further development of devices that provide airflow to adjust thermal comfort without causing the excessive drop in body temperature during sleep for temperature-changing environments. The proposed theory can be a breakthrough in solving the problems associated with airflow utilization during sleep and can be expanded to provide an affordable solution to mitigate hot discomfort during sleep in areas where air conditioning is not widely available.