User interaction patterns of a personal cooling system: a measurement study

(1)

User interaction patterns of a personal cooling system

Citation for published version (APA):

Verhaart, J. C. G., Li, R., & Zeiler, W. (2018). User interaction patterns of a personal cooling system: a measurement study. Science and Technology for the Built Environment, 24(1), 57-72.

https://doi.org/10.1080/23744731.2017.1333365

DOI:

10.1080/23744731.2017.1333365

Document status and date: Published: 02/01/2018 Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne Take down policy

If you believe that this document breaches copyright please contact us at: openaccess@tue.nl

providing details and we will investigate your claim.

(2)

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=uhvc21

Download by: [Eindhoven University of Technology] Date: 13 December 2017, At: 03:10 ISSN: 2374-4731 (Print) 2374-474X (Online) Journal homepage: http://www.tandfonline.com/loi/uhvc21

User interaction patterns of a personal cooling

system: A measurement study

Jacob Verhaart, Rongling Li & Wim Zeiler

To cite this article: Jacob Verhaart, Rongling Li & Wim Zeiler (2018) User interaction patterns of a

personal cooling system: A measurement study, Science and Technology for the Built Environment, 24:1, 57-72, DOI: 10.1080/23744731.2017.1333365

To link to this article: https://doi.org/10.1080/23744731.2017.1333365

Published online: 14 Jul 2017. Submit your article to this journal

Article views: 164

View related articles

(3)

Science and Technology for the Built Environment (2018) 24, 57–72

Published with license by Taylor & Francis ISSN: 2374-4731 print / 2374-474X online DOI:10.1080/23744731.2017.1333365

User interaction patterns of a personal cooling system:

A measurement study

JACOB VERHAART∗, RONGLING LI,andWIM ZEILER

Department of Built Environment, Eindhoven University of Technology, Building Services, P.O. Box 513 5600 MB Eindhoven, The Netherlands

Personal cooling systems provide cooling for individual office occupants to maintain thermal comfort at their workplace when cooling is needed. The indoor temperature of the office can be maintained at several degrees higher than is customary in offices today when personal cooling is available, which results in energy saving for office buildings as a whole. To better understand the individual cooling demand of building occupants and develop good control strategies for personal cooling systems, it is necessary to assess the interaction between the user and the personal cooling system. For this purpose, a personal cooling system was tested in a stable, slightly warm environment (27.5°C) in a climate chamber with 11 human subjects. The personal cooling system was controlled by the subject using a simple slider. The interaction of the user with the system was related to comfort level and perceived air quality. The subjects are categorized into groups based on gender, on comfort level, and on whether their comfort improved during the test or not. The results show that comfort level did not directly reflect in a difference in the number of interactions or level of the setting. The largest difference in setting was found between male and female subjects, where females required less cooling.

Introduction

The goal of any climate system in a building should be to opti-mize the environmental conditions for the occupants (Vischer

2008). In order to maintain the comfort levels of individual building occupants, local systems have to be provided to com-pensate for temperature drifts and extreme circumstances. However, even in a neutral environment, some people would like to feel cooler or warmer. A personal cooling system (PCS) is able to accommodate that and provide better comfort at lower total energy consumption (Verhaart et al.2015c).

User involvement in the control of PCS is essential. Avail-able personal control for the users leads to more realistic

Received November 1, 2016; accepted April 30, 2017

Jacob Verhaart, MSc, is a PhD Candidate. Rongling Li, PhD, is a Postdoctoral Researcher. Wim Zeiler, IR, is a Professor.

∗_{Corresponding author e-mail:}_{j.c.g.verhaart@tue.nl}

Color versions of one or more of the ﬁgures in the article can be found online atwww.tandfonline.com/uhvc.

C

Jacob Verhaart, Rongling Li, and Wim Zeiler.

This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and repro-duction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.

expectations about the thermal environment, which means that users tend to be more tolerant toward the local envi-ronment. This was shown by Brager (2004) for the case of openable windows. Boerstra et al. (2013) performed a large survey in oﬃce buildings and concluded that when building occupants perceive more control over noise, venti-lation, temperature, occupant satisfaction, and self-reported productivity increased. There was no relation found between perceived and available control in this study, except for the case with solar shading. However, it can be concluded that a high level of perceived control is a key factor for the sat-isfaction expected by building occupants (Hellwig 2015). From the conceptual framework presented by Hellwig (2015), guidelines for designing user control in oﬃce buildings can be distilled.

First, a high level of perceived control follows from the availability of effective controls. Luo et al. (2016) studied perceived control in a climate chamber where people did not have any control but could indicate dissatisfaction by ringing a bell. They were told that the operator would change the temperature when the bell was rung. This had a positive effect on thermal sensation (TS) and comfort, even though people did not have actual control and the temperature was not actually changed by the operator. It was noted that the test subjects, who participated in the test several times in different conditions, might be aware of the fact that the conditions did not change during the later tests. Even though this was a short study, the problems with providing so-called

(4)

dummy controls were already visible. In the long run, dummy controls lead to low self-eﬃcacy, lack of user’s belief in their own competences, and frustration with the building or the facility management (Hellwig2015). Eventually, this could lead to the misunderstanding, ignoring, and misuse of thermostats and other temperature controls as was found by Karjalainen and Koistinen (2007).

Second, a more responsive control is perceived as more sat-isfying (Leaman and Bordass1999). This related to the direct-ness of the control response. When applying this to the design of PCS, direct feedback of the control action toward the user is preferred over a delayed response. How direct the response is, is dependent on the type of PCS that is used. PCSs in all conﬁgurations have been tested in the past. This includes systems that were based on radiation, conduction, and con-vection. Radiation panels are generally slow in response and were, therefore, tested in combination with fans for cooling, for example (Melikov et al.2013). Systems based on conduc-tion, such as the hand cooler by Arens and Zhang (2008), the water-cooled chair employed by Pasut et al. (2013), the cooled desk plate (Verhaart et al.2015b) and the chair and desk plate tested by Pallubinsky et al. (2016) usually have a faster response. Most systems rely on the cooling eﬀect of air movement. This is because air-based systems do not require additional installations to move and mitigate energy and are able to respond quickly. Most typically, the air is provided from a supply unit on the desk (Bauman et al.1998; Dalewski et al. 2014; Kaczmarczyk and Melikov2006; Melikov et al.

2012) or integrated in the chair (Pasut et al.2014; Suzuki et al.

2010; Watanabe et al.2009).

The opportunity of using PCS as a means to give the user control and assess their demand for cooling is only hinted at up until now. In their patent for the heated and cooled chair, Arens et al. (2014) proposed a link that could be made between the chair and the global HVAC system. Boerstra et al. (2015) tested the effect of automated control for PCS by mimicking the user controls from the first test in a second test with the same subject. The main advantage found with respect to the control, was that in the second test, the subjects did not have to spend time and valuable attention to operating the personalized system. This led to an increase in measured productivity, most likely caused by the increase in focus they had to perform their task during this short test. Users might benefit from practice, where controlling the system becomes a habit, or part automation of the system through learning and interpretation by the system itself.

In most articles on PCS, the focus is on the thermal phys-iological response of a PCS on the body. This is tested using either thermal manikins (Arens et al.2013; Lipczynska et al.

2015) or human subjects (Pallubinsky et al.2016) in a con-trolled environment using fixed settings for the PCS. In other tests the PCS is user controlled, but the focus is more on the comfort reached by the systems. User control is usually only mentioned, not explained or analyzed. User control is shown in (Bauman et al. 1998), a field study using the first system described in (Demeter and Wichman1989). Results show that users changed the setting less often than once a day.

This study focuses on the interaction of the user with a PCS. How are control actions inﬂuenced by air supply temperature, personal characteristics, body composition, and

how the user perceives the system? Can user actions be inter-preted to determine their thermal preference? These research questions are investigated in this human subject experiment performed in the climate chamber. For this purpose, a PCS with cooling through air movement was built. The human interaction with the PCS is interpreted through logging the controls to determine the resulting cooling demand in an oﬃce environment. The way in which people operate the per-sonal cooling in a stable and warm environment was tested.

Methodology

Experiment design

The interaction of the user with a PCS is studied. A PCS is developed and installed in the climate chamber. The cli-mate chamber is furnished as a small two-person oﬃce. The environmental temperature is kept at a stable, slightly warm level, which is the same for all tests. In this condition, it is expected that a number of subjects feels uncomfortably warm and will be using the provided cooling system, which is avail-able for the subjects to use. User interactions with the system are monitored by logging the changes in setting continuously. Two aspects of the interactions that are analyzed are the num-ber of interactions and the level of the selected airspeed. This data is compared with the self-assessed comfort level to ﬁnd the relation between the two.

Test subjects

Eleven subjects were recruited among the students and staff of the faculty and consisted of six female subjects and five males and performed all tests. The details of the subjects are listed inTable 1. Each subject participated in four tests of 90 min-utes each in the climate chamber on 4 different days, perform-ing light computer work. The test subjects were advised to wear summer style clothing and wear similar clothes through the sessions. Most subjects complied with this; however, some subjects wore different outfits. This was caused by the outside conditions that fluctuated over the testing period. The cloth-ing insulation is collected from the questionnaire and shown in Table 2, together with the environmental temperature in the room during the tests and the outdoor temperature on the testing days. On average, the effect of this is minor. The sub-jects participated in a 4-site skin fold measurement (Jackson and Pollock1985) to determine their fat percentage. Using the fat percentage and the mass of the subject, the basal metabolic rate (BMR) is calculated according to Cunningham (1991). A clear difference in BMR was found between male and female test subjects. This BMR corresponds to 0.8 met.

PCS

The PCS that was available to use for the subjects is a local cooling system which uses moving cooled air for personal cooling. This system looks similar to the system tested by Dalewski et al. (2014). The purpose of the experiment by Dalewski was to test ductless ventilation, the air was taken from the room directly. In the present study, the air was taken

(5)

Volume 24, Number 1, January 2018

59

Table 1. The physical characteristics of the test subjects summarized for all male and all female test subjects.

Number Age, yr Height, m Weight, kg BMI, kg/m2 _{Fat percentage, %} _{BMR, W/m}2

Female 6 24.7 ± 3.4 1.68 ± 0.10 66.0 ± 12.0 23.3 ± 2.6 31.5 ± 6.0 37.2 ± 1.6 Male 5 24.8 ± 1.7 1.81 ± 0.14 87.4 ± 20.3 26.8 ± 5.4 17.8 ± 6.5 44.5 ± 0.6

from outside the climate chamber and heated or cooled to the pre-determined temperatures. The supply unit used in the current test has the same shape and size as Dalewski’s. How-ever, the air duct was wider and an aluminum honeycomb mesh with holes of 5 mm and a thickness of 5 cm was used to provide a laminar airflow at the nozzle. The air is sup-plied from 60 cm away, aimed at the face from the front and above (shown inFigure 1). The position of the supply nozzle was flexible in the test by Dalewski et al. (2014) and could be changed during the experiment. In this experiment, the posi-tion of the nozzle is adjusted to the test subject at the begin-ning of the test and fixed. The direction selected was the most preferred one among Dalewski’s test subjects.

The airspeed from the PCS is controlled directly by the user with the slider (shown inFigure 2), whereas subjects in Dalewski’s experiment used a turning knob. The design of slider makes use of the functionality matching strategy (Wever et al.2008). People are used to operating similar interfaces like sliders (for example in setting volume and light level on a com-puter or phone). A slider gives better visual feedback to the position than a turning knob. More complicated interfaces tend to distract the user, frustrate and induce abuse (Verhaart et al.2015a).

The relation between the position of the slider and the air-speed in m/s in the breathing zone is linear.Figure 2shows the relation of the control voltage and the airspeed in the breathing zone. Three supply temperatures were tested. Three different systems were used for that. The slider position relates to the control voltage between 0 and 10 V. The minimum and maximum values were adjusted to compensate for differences between the systems. This way, the same position of the slider represented the same airspeed for the user. Below 0.5 m/s, the airflow is hardly perceptible. Feedback from the slider posi-tion is immediate as the ventilator is able to adjust to the new airspeed in seconds. The slider design was chosen because it shows the user the current position and the available cooling capacity.

The supply air temperature was controlled using a ther-mostat with the thermistor in the nozzle. Through mixing between the nozzle and the face, the temperature of the air

at the face is diﬀerent from the temperature set-point at lower selected airspeeds. The relation between the air temperature at the face and the airspeed at the face is shown inFigure 3.

The PCS setting and air temperature is logged nine times per 10 seconds. User interaction with the PCS is assessed and the use patterns are categorized. The categories are ana-lyzed to ﬁnd the underlying causes for the behavior pattern by assessing comfort levels, body composition, and after-test interview responses.

Questionnaire

The TS and thermal comfort of the subjects was assessed using a computer-based questionnaire installed on the com-puter that was used by the subject. The same questionnaire was used by Veselý et al. (2017). The test subjects started their test right after entering the climate chamber by filling out the first part of the questionnaire. This part contains a checklist where the subjects mark all clothing items they are wearing. In most cases the cooling was started right away to compen-sate for the warm environment. Every 15 minutes, the next part of the questionnaire popped up for the subject to fill out. Using the questionnaire, the subjects reported their TS on the ASHRAE 7-point sensation scale as well as comfort for seven body parts (whole body, head, neck, arms, hands, legs, and feet) and answered seven questions related to sick building syndrome (SBS) symptoms such as perceived air quality (on a scale from stuffy to fresh). An example of the questions on TS and thermal comfort as well as an example of the per-ceived air quality is shown inFigure 4. After leaving the cli-mate chamber, the subjects were asked to fill out a form with open questions related to the overall conditions experienced and the functionality of the local cooling system.

The questionnaires in this experiment log the TS and com-fort votes on a continuous scale between the outer limits. The comfort scale is separated in the middle to indicate the dif-ference between comfortable and uncomfortable (Veselý et al.

2017). Both boxes are divided in two to create the four cat-egories: clearly uncomfortable, just uncomfortable, just com-fortable, and clearly comfortable.

Table 2. The general information of the tests: the air temperature in the room and the average clo-values. The format is mean± standard deviation.

Outdoor 24-hour Clo-value, clo

Test mean Room air Room radiant Relative

variant temperature,°C temperature,°C temperature,°C humidity, % Female subjects Male subjects Tref 12 27.6 ±0.2 27.6 ±0.1 25 ±4 0.64 ±0.14 0.49 ±0.04 T25 12.2 27.6 ±0.2 27.6 ±0.1 25 ±5 0.66 ±0.11 0.55 ±0.16 T23 7.3 27.3 ±0.3 27.6 ±0.2 23 ±8 0.65 ±0.11 0.51 ±0.01 T26 7.3 27.4 ±0.4 27.6 ±0.3 23 ±8 0.73 ±0.14 0.50 ±0.04

(6)

Fig. 1. Slider (on the left) used by the test subjects to control the airspeed of the PCS, which supplies cool air through the ducts on the right.

Room and testing environment

The tests were performed in the climate chamber located at Eindhoven University of Technology, previously described by Schellen et al. (2012) and Veselý (2017). The room is 3.6× 5.7 × 2.7 m3 _{on the inside. The walls, ﬂoor, and}

ceil-ing consist of aluminum panels with integrated water pipes. The room is also equipped with a dedicated air-handling unit that can be used to supply the room with air of a wide range of temperatures and relative humidity.

Figure 5shows the ﬂoor plan and sideview of the chamber during this experiment and it includes measures of the cham-ber itself. Two test subjects were placed in the chamcham-ber at the

Fig. 2. The calibration of the three PCSs. For each of the supply temperatures, a separate system was built. To make the systems feel the same for the subjects, this calibration was used to adjust the slider response.

same time. The desks are separated by a screen of 1.4 m height and 2.6 m width, which is high enough so that the subjects are unable to see each other when seated. Both desks were equipped with a PCS. The separator screen also ensures that the disturbance from the PCS from the other side of the wall is minimized.

The climate chamber is set to maintain an operative temperature of 27.5°C during all tests. The air temperature and radiative temperature are controlled separately. The air temperature is controlled by using the temperature measured at 1.1 m using a dedicated temperature sensor. The radiative temperature is controlled by maintaining the surface tem-peratures of all walls, the ﬂoor, and the ceiling. These are

Fig. 3. Relation between the air temperature and air speed in the breathing zone for the systems with diﬀerent supply tempera-tures, this can be seen as a measure for mixing between the supply nozzle and the breathing zone.

(7)

Volume 24, Number 1, January 2018

61

Fig. 4. Screenshots from the computer-based questionnaire: Table for selecting clothing items on the left, two example ques-tions on TS and thermal comfort in the middle and four of seven questions on SBS symptoms on the right.

controlled via water running through the aluminum panels. A temperature sensor is integrated in each of the walls, the ﬂoor, and the ceiling. The mean of these temperatures is used as the control temperature for the walls.

The climate is monitored using instruments mounted on the environmental measurement stand (EMS). The position of the EMS is marked inFigure 5. The instruments mounted are three anemometers (Dantec type 54R10), three air tem-perature sensors (type NTC, shielded) and three relative humidity sensors (type Humitter 50 U) mounted at three

Fig. 5. Layout of the climate chamber used in the experiment.

heights (0.1, 0.7, and 1.1 m). In addition, a black globe tem-perature sensor (15 cm black brass globe with NTC PR sen-sor) is mounted at 0.9 m. Environmental data of the climate chamber is logged at 10-second intervals.

Testing procedure

The subjects had personal cooling available in three of the four tests. The first stage of the experiment included reference test (Tref) and a test with supply air from the PCS available at 25°C (T25). The second stage included a test with cooling available at 23°C (T23) and another one with 26°C (T26) sup-ply air. These two tests were done in random order to com-pensate for the learning effect between Tsfigure T23 and T26.

Figure 6 shows the test schedule. Each test lasted for 90 minutes with 30 minutes accustomization prior to the test outside the climate chamber in a temperature between 22°C and 24°C, which can be considered a neutral environment. This procedure is in accordance with similar studies, such as Pallubinsky et al. (2016), who did a preparation before

(8)

Fig. 6. Test schedule during the experiment. Accustomization is done prior to the start of the test. At the start of the test, the subject moves to the climate chamber. Most interactions happen during the ﬁrst 30 minutes of the test, this is analyzed separately.

entering the climate chamber, 30 minutes of rest in the cli-mate chamber before starting the experiment, and repeated 30 minutes rest between every cooling mode. (Veselý et al.

2017), 30 minutes acclimatization in a neutral environment of 21°C (with a higher clo-value). Lipczynska et al. (2015) applied 30 minutes acclimatization in special room and sub-sequently moved to experimentation room. Dalewski et al. (2014) applied 30 minutes acclimatization in “anteroom” dur-ing which air quality assessment has already started. Pasut et al. (2012) applied a 15 minute preconditioning period in the test chamber, on a diﬀerent chair to get used to the tem-perature, then moved to the chair that is used for the test itself. During the stay in the climate chamber, the test subjects were allowed to work on their own computer. For analysis, the ﬁrst 30 minutes will be considered the settling period and the last hour is the stable period.

Results

Room environmental condition

The experiments were performed between September 28th and October 15th of 2015. The tests were conducted in two parts. The reference test and the test with cooling available at the supply temperature of 25°C were performed ﬁrst between September 28th and October 5th. The second part of the test, where the supply temperatures were 23°C and 26°C, was per-formed between October 7th and October 15th. InTable 2, the testing conditions are summarized. The outdoor temper-ature is the average of the 24-hour mean of the test days for a particular test variant measured at a Royal Netherlands Mete-orogical Institute weather station in Eindhoven. During the second part of the experiment, the outdoor temperature was lower. However, the change of clo-value of the four tests is insigniﬁcant.

Learning eﬀect

The first 30 minutes of the experiment is considered the set-tling period in which the users find the optimal setting for the PCS. The difference between the first 30 minutes and the last hour was analyzed using the Wilcoxon Signed Rank test (non-parametric dependent test) in MATLAB. Statistical sig-nificance is assumed of p< 0.05. Of all interactions, 79% are done in the first 30 minutes. In all cases, there is a signifi-cant difference (p< 1%) between the number of interactions in the first 30 minutes compared to the last hour. Therefore, this period is analyzed separately from the last hour of the experiment.Figure 7 shows the average setting per test con-dition used by all subjects over the whole testing period. The

tests show that the mean level of the selected airspeed does not change much after 30 minutes.

The tests were performed in a particular order. The refer-ence test and T25 were conducted ﬁrst and T23 and T26 in a second set. The order in which the subjects conducted the tests within the sets was random. To assess whether the order had an eﬀect on the behavior, the tests were analyzed in the order in which they were done. For half of the subjects, this order was T25, T23, and T26. For the other half, the order was T25, T26, and T23.

The number of interactions during these tests are shown in

Figure 8, graph on the left. All data is also shown inTable 3.

There is a significant difference in the number of interac-tions in all tests between the first 30 minute and the last hour; however, no significant difference between the tests. There was no significant difference found in the level of the selected air-speed after 30 minutes compared to the level at the end of the tests and there is only a slight significance to the difference between the level of the first and the last test after 30 min-utes. This effect was far less than the effect of the supply air temperature.

The eﬀect of supply air temperature

In the experiment, three supply temperatures for the PCS were tested. All results are included inTable 4. The eﬀect of the supply temperature on the interactions and the level of the selected airspeed was small. The diﬀerence in the number of

Fig. 7. Average setting over time per test variant, 79% of all inter-actions by the subjects are done in the ﬁrst 30 minutes, here indi-cated by the gray area.

(9)

T able 3 . Use char a cteristics of the tests in or der o f h o w they w er e conducted. Mean v a lues p -v alues M1 M2 M3 a t 30 min a t 9 0 m in M1 M2 M3 at at at at at at M 1 v M 2 v M 2 v M 1 v M 2 v M 2 v 3 0 v 3 0 v 3 0 v 30 min. 90 min. 30 min. 90 min. 30 min. 90 min. M3 M1 M3 M3 M1 M3 90 90 90 Settings 1 ,302 1,183 1,101 1,007 1,046 1,043 3.22% ∗ 27.83% 51.95% 19.82% 11.82% 81.45% 28.52% 67.19% 68.75% Inter a ctions 10,91 3,18 9,45 1,55 7,00 2,45 12.70% 78.91% 30.47% 45.31% 12.50% 38.28% 0.98% ∗∗ 1.95% ∗ 3.13% ∗ ∗p-v alue < 5%; ∗∗p -v alue < 1%.

63

(10)

Fig. 8. Number of interactions for each test and all subjects. For each test, the boxplot shows the ﬁrst 30 minutes and the last hour of the test. The circles indicate the mean values. On the left, sorted in time: tests performed ﬁrst, second, and third. On the right: sorted per supply air temperature.

Table 4a. Thermal sensation, comfort, SBS symptoms and use characteristics per supply air temperature, mean values.

Tref T25 T23 T26

at 30 min at 90 min at 30 min at 90 min at 30 min at 90 min at 30 min at 90 min 1 Whole body thermal

sensation

0.901 0.675 − 0.029 − 0.064 0.161 − 0.313 0.205 0.212 2 Whole body comfort 1.127 1.382 1.945 1.947 1.780 2.206 2.005 2.551 3 Thermal sensation of the

head

0.915 0.666 − 0.106 − 0.153 0.263 − 0.063 0.038 − 0.053 4 Comfort of the head 1.094 1.555 0.839 1.931 2.002 1.672 1.238 1.703 5 Thermal sensation of the

neck

0.717 0.498 0.180 − 0.069 0.096 − 0.039 0.254 0.042 6 Comfort of the neck 1.526 1.752 1.976 1.944 1.864 1.895 1.407 1.991 7 Thermal sensation of the

hands

0.894 0.707 0.456 0.173 0.330 0.145 0.361 0.067 8 Comfort of the hands 1.298 1.732 1.732 1.556 1.490 1.706 1.765 1.742 9 Thermal sensation of the

arms

0.677 0.348 0.060 0.032 0.103 − 0.345 0.266 − 0.171 10 Comfort of the arms 1.215 1.937 1.655 1.173 1.676 1.494 1.873 2.008 11 Thermal sensation of the

feet

1.065 0.985 1.021 0.848 0.663 0.765 0.759 0.783 12 Comfort of the feet 0.339 0.825 1.032 1.551 1.996 1.496 1.391 1.702 13 Thermal sensation of the

legs

0.702 0.732 0.465 0.534 0.585 0.502 0.326 0.505 14 Comfort of the legs 1.037 1.195 1.767 2.098 2.167 1.781 1.835 2.046 15 Perceive the air as stuﬀy

or fresh

36% 40% 57% 57% 52% 53% 53% 50% 16 Have a headache or not 76% 81% 74% 76% 83% 84% 66% 80% 17 Find it diﬃcult to think

or have a clear head

67% 71% 72% 72% 75% 76% 76% 76% 18 Feel tired or rested 46% 54% 63% 62% 68% 65% 65% 56% 19 Find it diﬃcult or easy

to concentrate

59% 66% 71% 72% 78% 74% 74% 69% 20 Feel sleepy or alert 51% 56% 61% 64% 73% 68% 64% 59% 21 Indicate the ability to

work in percentage

67% 71% 75% 72% 80% 76% 77% 76% Settings n.a. n.a. 1.302 1.183 0.603 0.571 0.871 0.803 Interactions n.a. n.a. 10.91 3.18 8.82 1.18 7.64 2.82

(11)

T able 4 b . Ther m al sensa tion, comf ort, SBS symptoms a nd use char a cteristics per suppl y a ir temper a tur e, p -v alues . at 3 0 m in a t 9 0 m in Tr ef v T25 Tr ef v T23 Tr ef v T26 T25 v T26 T23 v T25 T23 v T26 Tr ef v T25 Tr ef v T23 Tr ef v T26 T25 v T26 T23 v T25 T23 v T26 1 W hole bod y ther mal sensa tion 0.59% ∗∗ 5.47% 1.17% ∗ 15.63% 71.09% 65.23% 0.78% ∗∗ 0.98% ∗∗ 2.34% ∗ 7.42% 57.03% 2.93% ∗ 2 W hole bod y comf ort 2 4.02% 20.31% 20.31% 76.46% 96.58% 100.00% 35.94% 32.03% 17.48% 27.83% 46.48% 75.78% 3 Ther m al sensa tion o f the head 1.86% ∗ 2.54% ∗ 0.98% ∗∗ 70.02% 14.75% 10.16% 1.86% ∗ 4.88% ∗ 1.37% ∗ 100.00% 100.00% 44.92% 4 C omf o rt of the h ead 8 9.84% 4.20% ∗ 57.71% 33.40% 8.30% 74.22% 62.50% 70.02% 76.95% 82.03% 57.23% 73.44% 5 Ther m al sensa tion o f the neck 5.47% 0.78% ∗∗ 0.78% ∗∗ 74.22% 84.38% 15.63% 1.95% ∗ 12.89% 7.81% 67.19% 98.05% 84.38% 6 C omf o rt of the n eck 55.66% 55.66% 92.19% 55.47% 82.03% 61.72% 73.44% 82.03% 76.95% 100.00% 94.53% 57.03% 7 Ther m al sensa tion o f the hands 2.73% ∗ 9.77% 3.91% ∗ 74.22% 100.00% 93.75% 13.09% 19.34% 16.41% 54.10% 90.23% 98.05% 8 C omf o rt of the h ands 36.52% 69.53% 37.50% 44.92% 97.66% 51.56% 91.02% 100.00% 100.00% 16.02% 57.23% 98.44% 9 Ther m al sensa tion o f the ar ms 2.34% ∗ 0.39% ∗∗ 2.34% ∗ 68.75% 94.53% 16.41% 23.24% 3.91% ∗ 7.42% 32.23% 12.89% 83.59% 10 Comf ort o f the ar ms 43.16% 43.16% 49.22% 67.58% 74.22% 85.16% 25.00% 36.52% 84.57% 12.89% 43.16% 57.03% 11 Ther m al sensa tion o f the feet 76.95% 13.67% 7.81% 25.00% 38.87% 66.41% 63.77% 69.53% 55.66% 76.46% 96.58% 82.03% 12 Comf ort o f the feet 37.70% 3.91% ∗ 6.74% 10.55% 10.16% 70.02% 42.58% 43.16% 16.41% 69.53% 96.58% 43.16% 13 Ther m al sensa tion o f the legs 30.08% 35.94% 10.94% 43.75% 53.13% 1.56% ∗ 41.60% 33.59% 19.53% 100.00% 75.00% 84.38% 14 Comf ort o f the legs 3.13% ∗ 6.45% 27.54% 71.88% 51.95% 91.50% 20.31% 41.41% 25.00% 100.00% 76.95% 76.95% 15 P er cei v e the a ir as stuffy o r fr esh 0.68% ∗∗ 5.37% 14.75% 62.50% 69.53% 76.46% 5.08% 32.03% 30.86% 46.48% 83.11% 65.53% 16 Ha v e a h eadache o r not 81.25% 6.25% 46.09% 94.53% 21.88% 7.81% 29.69% 81.25% 100.00% 3.91% ∗ 16.41% 57.81% 17 Find it difficult to think or ha v e a clear h ead 70.02% 20.61% 32.03% 23.24% 23.24% 73.44% 89.84% 76.46% 76.46% 57.71% 57.71% 83.11% 18 F eel tir ed o r rested 1 0.16% 8.30% 5.76% 24.02% 24.02% 76.46% 32.03% 32.03% 76.46% 12.50% 57.71% 21.39% 19 Find it difficult o r easy to concentr a te 10.64% 1.37% ∗ 10.16% 63.77% 13.09% 19.82% 46.48% 27.83% 70.02% 89.84% 30.96% 62.21% 20 F eel sleep y o r a lert 32.03% 2.44% ∗ 10.55% 89.84% 6.64% 41.31% 59.28% 14.75% 46.48% 49.22% 51.95% 27.83% 21 Indica te the ability to w or k in per centa ge 36.52% 0.98% ∗∗ 8.30% 96.58% 36.52% 20.61% 89.84% 76.46% 76.46% 57.71% 57.71% 83.11% Settings n .a. n .a. n .a. 0 .29% ∗∗ 0.10% ∗∗ 0.68% ∗∗ n.a. n.a. n.a. 0.68% ∗∗ 0.20% ∗∗ 6.45% Inter a ctions n.a. n.a. n.a. 20.12% 47.66% 47.66% n.a. n.a. n.a. 60.94% 9.38% 9.38% ∗p-v alue < 5%; ∗∗p -v alue < 1%.

65

(12)

Table 4c. Per supply air temperature, p-values continued.

Tref T25 T23 T26

30 v 90 30 v 90 30 v 90 30 v 90 1 Whole body thermal sensation 39.06% 76.17% 12.89% 67.19% 2 Whole body comfort 57.03% 96.58% 14.75% 16.02% 3 Thermal sensation of the head 3.52%∗ 83.11% 13.09% 35.94% 4 Comfort of the head 5.08% 6.74% 94.53% 25.00% 5 Thermal sensation of the neck 16.41% 29.69% 91.02% 21.88% 6 Comfort of the neck 29.69% 73.44% 73.44% 35.94% 7 Thermal sensation of the hands 36.13% 21.88% 60.74% 29.69% 8 Comfort of the hands 38.28% 84.77% 62.50% 96.88% 9 Thermal sensation of the arms 3.91%∗ 94.53% 25.00% 2.34%∗ 10 Comfort of the arms 15.63% 84.38% 100.00% 84.38% 11 Thermal sensation of the feet 69.53% 43.16% 94.53% 100.00% 12 Comfort of the feet 47.66% 70.02% 76.95% 19.34% 13 Thermal sensation of the legs 81.25% 56.25% 32.81% 43.75% 14 Comfort of the legs 56.25% 26.56% 98.24% 67.58% 15 Perceive the air as stuffy or fresh 12.30% 92.19% 84.57% 63.77% 16 Have a headache or not 29.69% 57.81% 93.75% 1.56%∗ 17 Find it difficult to think or have a clear head 36.52% 96.58% 89.84% 100.00% 18 Feel tired or rested 5.76% 62.50% 43.16% 0.49%∗∗ 19 Find it difficult or easy to concentrate 10.16% 83.11% 16.89% 63.77% 20 Feel sleepy or alert 36.52% 76.46% 19.34% 73.44% 21 Indicate the ability to work in percentage 41.31% 51.95% 23.14% 46.48% Settings n.a. 28.52% 71.88% 57.81% Interactions n.a. 0.98%∗∗ 0.39%∗∗ 12.89%∗∗

∗_p-value_{< 5%;}∗∗_p-value_{< 1%.}

interactions between the tests for both the first 30 minutes and the last hour is not significant. There is a difference in the set-ting between all supply air temperatures. In addition to that, a significant difference was found between the number of inter-actions in the first 30 minutes and last hour of tests T25 and T23. This is shown inFigure 8, graph on the right.

The comfort and TS votes for all subjects at 30 minutes and at 90 minutes are compared for all three supply temper-atures. Figure 9shows the mean overall TS votes per ques-tionnaire, sorted for all supply air temperatures as well as for the reference test. The mean PMV is also shown. The TS of the reference test moves toward the PMV at the end, while the cases with cooling move toward neutral, or slightly below that for T23.

Only a few cases are found to have a significant differ-ence between two tests. Between the case without cooling and all cases with cooling, most aspects of TS and comfort show a significant difference, both after 30 minutes and after 90 minutes. When looking per test at the difference between 30 and 90 minutes, no parameters show significant differences.

The eﬀect of body composition on PCS use

The subjects are categorized according to three characteristics to determine the relation between comfort level, the number of interaction and the selected level of airspeed. Full results can be found inTable 5. The diﬀerence between groups was analyzed using the Wilcoxon Rank Sum test (non-parametric independent test, equivalent to Mann-Whitney U-test). The ﬁrst categorization is based on gender, because there was a

large difference in body composition and BMR between the male and female subjects (seeTable 1). There is no significant difference between male and female subjects when it comes to TS, comfort levels, and the occurrence of SBS symptoms except for a slight difference for the TS at the arms and the neck. This can be caused by the direction of the airflow and the difference in the type of clothing worn, where t-shirts of the female subjects tend to cover less of the upper arm.

Fig. 9. Mean PMV, calculated for all tests per timestep, with the mean TS as voted by all subjects over the whole session, sorted per supply air temperature.

(13)

Volume 24, Number 1, January 2018

67

Table 5. Thermal sensation, thermal comfort, SBS symptoms and use characteristics of all tests, subjects split in male and female subjects.

Mean values p-values

Male Female at 30 min. at 90 min. Male Female Male vs. Male vs.

at 30 min. at 90 min. at 30 min. at 90 min. Female Female 30 vs. 90 30 vs. 90 1 Whole body thermal

sensation

0.27 0.10 − 0.02 − 0.19 36.40% 12.81% 32.96% 16.71% 2 Whole body comfort 1.46 2.15 2.28 2.31 23.92% 91.35% 10.36% 34.90% 3 Thermal sensation of the

head

0.17 − 0.18 − 0.02 − 0.02 38.21% 52.56% 3.13%∗ 90.58% 4 Comfort of the head 1.24 1.97 1.46 1.60 92.74% 37.21% 4.80%∗ 64.84% 5 Thermal sensation of the

neck

0.33 0.05 0.05 − 0.09 3.15%∗ 40.92% 9.96% 53.15% 6 Comfort of the neck 1.76 1.87 1.74 2.01 82.59% 97.07% 100.00% 16.27% 7 Thermal sensation of the

hands

0.41 0.17 0.36 0.09 94.00% 71.63% 33.11% 13.15% 8 Comfort of the hands 1.88 1.66 1.48 1.68 68.84% 86.94% 58.79% 20.93% 9 Thermal sensation of the

arms

0.26 0.07 0.04 − 0.35 38.69% 2.76%∗ 53.13% 3.12%∗ 10 Comfort of the arms 1.76 1.60 1.72 1.53 88.41% 75.62% 89.84% 98.90% 11 Thermal sensation of the

feet

0.74 0.88 0.88 0.73 58.22% 100.00% 68.36% 22.36% 12 Comfort of the feet 0.90 0.80 1.95 2.24 24.64% 6.43% 85.80% 35.20% 13 Thermal sensation of the

legs

0.38 0.56 0.52 0.48 46.55% 61.68% 31.25% 53.13% 14 Comfort of the legs 1.38 1.62 2.38 2.27 37.48% 40.22% 21.63% 73.67% 15 Perceive the air as stuﬀy

or fresh

54% 56% 54% 51% 89.92% 49.21% 100.00% 93.06% 16 Have a headache or not 78% 86% 71% 75% 20.77% 4.20% 31.25% 4.93%∗ 17 Find it diﬃcult to think

or have a clear head

79% 80% 70% 70% 15.31% 8.26% 77.25% 94.79% 18 Feel tired or rested 71% 70% 61% 54% 16.39% 3.14% 68.48% 0.50%∗∗ 19 Find it diﬃcult or easy

to concentrate

79% 77% 71% 67% 41.56% 17.51% 67.88% 40.79% 20 Feel sleepy or alert 68% 69% 64% 60% 67.75% 23.27% 72.66% 42.27% 21 Indicate the ability to

work in percentage

81% 80% 74% 70% 36.60% 8.26% 84.69% 42.04% Settings 1.10 1.31 0.78 0.47 4.87%∗ 0.00%∗∗ 10.55% 0.07%∗∗ Interactions 9.20 1.67 9.06 3.00 56.14% 40.92% 0.16%∗∗ 0.27%∗∗

∗_p-value_{< 5%;}∗∗_p-value_{< 1%.}

The largest difference between male and female test sub-jects is found in the setting.Figure 10shows the significant difference in setting (p= 0.049) after 30 minutes and an even larger difference at the end of the experiment. This is mainly due to the fact that during the last hour, the male subjects tend to move the slider to increase the airspeed while female subjects tend to reduce the airspeed. The difference between the setting at 30 minutes and 90 minutes is significantly higher for the female subjects than for male subjects.

PCS performance

The second categorization is based on the overall comfort level. Full results are in Table 6. The comfort level at

30 minutes and at the end of the test are taken for the three test variants where the subjects had cooling available. These are assessed and when more than half was uncomfort-able or clearly uncomfortuncomfort-able, the subject is categorized in the uncomfortable group. The comfortable group contained eight people and the uncomfortable group three. The uncomfortable group consists of one male subject and two female subjects.

Figure 11shows that grouping according to comfort level results in a clear difference in the perceived comfort level for all body parts after both 30 and 90 minutes. In addition to that, there is a difference between groups with respect to the perceived air quality, which means they indicated that they perceived the air as stuffier (Figure 12). This shows that the experienced air quality relates to the overall thermal comfort. Some subjects reported that the air in the climate chamber

(14)

T able 6 . Ther m al sensa tion, comf ort, SBS symptoms , and u se char acteristics for all tests , “ comf orta b le” v ersus “ uncomf o rta b le”. Mean v a lues p -v alues Comf orta b le Uncomf o rta b le a t 30 min. a t 90 min. Comf orta b le v. C omf o rta b le v. Comf orta b le U ncomf o rta a t 30 min. a t 90 min a t 3 0 m in. a t 9 0 m in. Uncomf o rta b le Uncomf orta b le 3 0 v. 9 0 3 0 v. 9 0 1 W hole b od y ther m al sensa tion 0 .17 − 0.03 − 0.03 − 0.13 46.85% 65.20% 21.81% 12.11% 2 W hole b od y comf o rt 2.62 3.02 0.01 0.13 0.10% ∗∗ 0.01% ∗∗ 11.94% 36.72% 3 Ther m al sensa tion o f the head 0.00 − 0.21 0.25 0.24 19.26% 1.26% ∗ 20.54% 92.19% 4 C omf o rt of the h ead 2 .01 2 .46 − 0.38 − 0.09 1.09% ∗ 0.01% ∗∗ 16.17% 25.00% 5 Ther m al sensa tion o f the neck 0.11 − 0.07 0.35 0.11 13.85% 33.52% 24.61% 21.88% 6 C omf o rt of the n eck 2.44 2.60 − 0.10 0.19 0.03% ∗∗ 0.15% ∗∗ 50.34% 12.50% 7 Ther m al sensa tion o f the hands 0.39 0.14 0.35 0.09 29.30% 95.14% 13.29% 21.09% 8 C omf o rt of the h ands 2.27 2.23 0.03 0.17 0.21% ∗∗ 0.70% ∗∗ 78.60% 28.13% 9 Ther m al sensa tion o f the ar ms 0.19 − 0.11 0.03 − 0.30 80.08% 41.54% 17.72% 10.94% 10 Comf ort o f the ar ms 2.33 2.10 0.14 0.12 1.02% ∗ 0.58% ∗∗ 97.41% 62.50% 11 Ther m al sensa tion o f the feet 0.85 0.86 0.72 0.65 91.83% 72.99% 77.82% 26.56% 12 Comf ort o f the feet 2.00 2.02 0.06 0.43 1.27% ∗ 4.26% ∗ 96.58% 29.69% 13 Ther m al sensa tion o f the legs 2.00 2.02 0.06 0.43 23.76% 98.35% 47.29% 56.25% 14 Comf ort o f the legs 2.59 2.60 0.14 0.31 0.24% ∗∗ 0.53% ∗∗ 40.76% 40.63% 15 P er cei v e the a ir as stuffy o r fr esh 59% 58% 41% 40% 6.28% 3.21% ∗ 44.70% 46.09% 16 Ha v e a h eadache o r not 82% 89% 54% 55% 0.11% ∗∗ 0.02% ∗∗ 2.56% ∗ 91.02% 17 Find it difficult to think or ha v e a clear head 82% 78% 55% 65% 0.08% ∗∗ 2.49% ∗ 19.61% 9.77% 18 F eel tir ed o r rested 73% 69% 46% 40% 0.23% ∗∗ 0.30% ∗∗ 5.73% 5.47% 19 Find it difficult o r easy to concentr a te 81% 78% 57% 55% 0.34% ∗∗ 0.21% ∗∗ 45.76% 57.03% 20 F eel sleep y o r a lert 72% 71% 50% 46% 4.11% ∗ 0.56% ∗∗ 63.78% 46.09% 21 Indica te the ability to w or k in p er centa ge 80% 78% 69% 65% 6.01% 2.49% 71.03% 44.53% Settings 0 .95 0 .97 0 .85 0 .55 83.98% 6.87% 58.61% 15.63% Inter a ctions 8.50 2.50 10.78 2.11 19.41% 100.00% 0.23% ∗∗ 0.78% ∗p-v alue < 5%; ∗∗p -v alue < 1%.

68

(15)

Volume 24, Number 1, January 2018

69

Fig. 10. The diﬀerences found in the level of selected airspeed of the PCS for female and male test subjects after 30 minutes and at the end of the test. The circles indicate the mean values.

or from the personal cooling terminal smelled bad, this could be related to the air quality experienced by the test subjects. There is no diﬀerence in the selected airspeed or the num-ber of interactions between the comfortable and uncomfortable groups.

The last grouping is based on whether the comfort level was improved for the subject, results included inTable 7. A subject is categorized in the group improved when the age comfort rating after 90 minutes is higher than the aver-age after 30 minutes. The group improved contains four sub-jects and the group with the same comfort contains seven. The improved group contains one female subject.

Fig. 11. The diﬀerence in comfort level between the people who were comfortable and the people who were uncomfortable during most of the tests. The circles indicate the mean values.

The comfort level improved over the test period for the group of subjects improved, but stayed the same for the other group: the same (see Figure 12). The comfort level of the improved group is lower in general than the same group. This, however, was only a significant difference in the air quality and the comfort of the legs and feet. There is no difference in the number of interactions between the improved and the

same; however, the improved group kept the airspeed at the

same level during the whole test, when the the same group reduces the level between 30 and 90 minutes, leading to a sig-niﬁcant diﬀerence at the end.

Discussion

In this study, 11 subjects tested a PCS in four tests. The sub-jects had cooling available which they could control directly. The interaction between human and cooling system can pro-vide insight into the cooling demand of diﬀerent people. This might help the development of PCSs, interfaces and automated local climate control systems as well as providing insight into cooling demand on a room or an oﬃce level.

There is no clear progression visible from the occurrence of discomfort to an action taken to change the airspeed. This is due to the setup of the experiment where question-naires are presented at regular time intervals. This shows the progression of TS and comfort in a low resolution. The moment of interaction cannot be interpreted as the moment of discomfort. In a number of studies, the logic is used that the control action is directly related to experienced discom-fort (Haldi and Robinson2010). However, a user’s action is related to attention level, whether they are active or passive users, habituation, and the wide range of usually preferred environmental conditions. This can lead to a delayed action by the user which introduces diﬃculties when relating control actions directly to comfort level or current environmental conditions. The motivation of a speciﬁc control action can be asked by providing the subject with a questionnaire at the moment of interaction; however, this places a burden on the subject and could lead to an altered behavioral pattern.

The main diﬀerence between male and female subjects in this study was related to the setting at the end of the test, which was higher for the males. This is in accordance with the ﬁndings of (Amai et al.2007). The subjects in that study had the option to adjust the supply air temperature and found that the female subjects preferred a higher supply air temperature. The system interface is easy to understand and the feed-back of the fan is instantaneous which makes it easy for the user to assess the setting of the system and to adjust quickly.

The analysis of the difference in preferred temperature of the supplied air showed that there was no difference in selected airspeed level and no difference in TS and comfort. More important for the performance of the PCS is the air quality. During a small number of tests, the air quality was deemed insufficient by the subjects and they reported a bad smell. This resulted in an uncomfortable situation. The position of the supply unit and the direction of the supplied air, directly toward the face requires that the air is fresh and odorless.

(16)

T able 7 . Ther m al sensa tion, comf ort, SBS symptoms , and u se char acteristics o f a ll tests , subjects sorted in g roups “impr o v ed” and “ the same”. Mean v a lues p -v alues Impr o v ed the S ame a t 30 min. a t 90 min. Impr o v ed the S a t 30 min. a t 90 min. a t 30 min. a t 90 min. Impr o v ed v. the S ame Impr o v ed v. the Same 30 v. 90 30 1 W hole b od y ther m al sensa tion 0 .24 0 .23 0 .04 − 0.22 98.47% 1.59% ∗ 71.48% 1.47% 2 W hole b od y comf o rt 0.70 1.82 2.60 2.47 0.49% ∗∗ 33.90% 0.29% ∗∗ 93.07% 3 Ther m al sensa tion o f the head 0.19 − 0.13 − 0.01 − 0.07 46.23% 86.59% 27.54% 76.51% 4 C omf o rt of the h ead 0 .58 1 .71 1 .80 1 .80 14.64% 83.57% 2.44% ∗ 81.07% 5 Ther m al sensa tion o f the neck 0.27 0.06 0.12 − 0.07 46.20% 25.47% 49.61% 18.27% 6 C omf o rt of the n eck 1.02 1.60 2.17 2.14 16.59% 55.63% 31.25% 62.91% 7 Ther m al sensa tion o f the hands 0.64 0.28 0.24 0.04 9.03% 46.32% 17.58% 19.29% 8 C omf o rt of the h ands 1.09 1.63 1.99 1.69 18.66% 90.97% 27.54% 72.27% 9 Ther m al sensa tion o f the ar ms 0.54 0.08 − 0.08 − 0.30 0.76% ∗∗ 16.31% 20.31% 9.79% 10 Comf ort o f the ar ms 0.93 0.89 2.19 1.94 4.76% ∗ 12.11% 100.00% 96.53% 11 Ther m al sensa tion o f the feet 1.14 1.36 0.63 0.48 10.66% 1.88% ∗ 55.66% 21.44% 12 Comf ort o f the feet 0.35 0.08 2.11 2.44 3.74% ∗ 0.55% ∗∗ 90.97% 33.41% 13 Ther m al sensa tion o f the legs 0.81 1.02 0.26 0.22 2.22% ∗ 0.48% ∗∗ 29.69% 52.98% 14 Comf ort o f the legs 0.62 1.16 2.67 2.44 1.40% ∗ 10.91% 6.74% 94.79% 15 P er cei v e the a ir as stuffy o r fr esh 64% 68% 48% 45% 4.51% ∗ 0.40% ∗∗ 44.73% 56.62% 16 Ha v e a h eadache o r not 65% 78% 80% 81% 14.50% 86.20% 1.95% ∗ 49.73% 17 Find it difficult to think or ha v e a clear head 74% 78% 75% 73% 86.62% 34.94% 31.05% 50.16% 18 F eel tir ed o r rested 73% 70% 61% 56% 9.58% 6.39% 38.87% 0.57% 19 Find it difficult o r easy to concentr a te 70% 71% 77% 72% 32.10% 70.82% 79.10% 8.22% 20 F eel sleep y o r a lert 66% 69% 66% 61% 95.52% 33.05% 76.46% 19.08% 21 Indica te the ability to w or k in p er centa ge 76% 78% 78% 73% 76.46% 34.94% 51.86% 11.38% Settings 1 .09 1 .23 0 .83 0 .64 27.77% 0.53% ∗∗ 51.56% 1.34% Inter a ctions 11.00 3 .00 8 .05 2 .05 16.45% 42.60% 1.37% ∗ 0.07% ∗p-v alue < 5%; ∗∗p -v alue < 1%.

70

(17)

Volume 24, Number 1, January 2018

71

Fig. 12. The diﬀerence in rating of air quality. Left: people categorized as comfortable rate the air quality higher compared to the uncomfortable people. Right: the diﬀerence between the ratings for the people that were more content at the end of the test than at the beginning compared to the people for whom this did not change. Mean values indicated by the circles.

Conclusions

The experiment was conducted in a slightly warm environ-ment (27.5°C ± 0.3°C) with subjects that were accustomed to a neutral environment before entering the climate chamber. This test, therefore, creates a one-step change from neutral to warm and determines the subject’s response to that. The period of settling is clearly visible in the first 30 minutes of testing. After that, the setting of the PCS is more stable. This is reflected in the fact that in most cases, the subjects use the system significantly more often in the first 30 minutes when compared to the last 60 minutes of the test.

Three air supply temperatures are tested and all tempera-tures show an improvement to the TS, comfort and SBS symp-toms over the reference case without cooling. Between the cases with cooling, there is no difference in TS and comfort or number of interactions needed by the subjects; however, there is a significant difference between the selected airspeed. The supply air temperature of 23°C was used at the lowest setting on average.

The best determinant found for the difference in use is body composition. The selected airspeed is clearly different for the male subjects compared to the female subjects. The female subjects in all cases chose a lower airspeed and tended to lower the airspeed during the test (between 30 and 90 min-utes). This indicates that BMR is of great importance for the level of cooling needed. The comfort level between male and female subjects was not significantly different. This indi-cates that it was possible for most people to find the correct airspeed and that the selected airspeed reflects the cooling demand of the subject correctly, apart from the cases where people were uncomfortable. Discomfort with the system could be traced back to perceived quality of the air provided by the system.

The number of interactions did not change as a result of discomfort, body composition, or air supply temperature. No significant difference was found between any of the groups and any of the tests. There was no clear difference in the

number of interactions and setting level in the cases of

com-fortable subjects compared to uncomcom-fortable people. The

set-ting of people that were categorized in the improved group used a higher setting at the end of the test compared to the

same group. This means that, from this test, it cannot be

con-cluded that use frequency is an important factor related to comfort. On the other hand, the selected level of airspeed is a good indicator of desired level of cooling, when the air quality is suﬃcient.

References

Amai, H., S. Tanabe, T. Akimoto, and T. Genma. 2007. Thermal sensa-tion and comfort with diﬀerent task condisensa-tioning systems. Building

and Environment 42(12):3955–64.

Arens, E.A., and H. Zhang. 2008. Impact of a task-ambient ventilation system on perceived air quality. Proceedings of Indoor Air. Copen-hagen, Denmark, August 17–22.

Arens, E.A., H. Zhang, and W. Pasut. 2014. US 2014/0217785 A1. United States.

Arens, E.A., H. Zhang, W. Pasut, Y. Zhai, T. Hoyt, and L. Huang. 2013.

Air Movement as an Energy Eﬃcient Means Toward Occupant Com-fort. Report CA Air Resources Board, CBE Berkeley 2012.

Bauman, F.S., T.G. Carter, A.V. Baughman, and E.A. Arens. 1998. Field study of the impact of a desktop task/ambient conditioning system in oﬃce buildings. AHSRAE Transactions 104, 1153–71.

Boerstra, A.C., T. Beuker, M.G.L.C. Loomans, and J.L.M. Hensen. 2013. Impact of available and perceived control on comfort and health in European oﬃces. Architectural Science Review 56(July 2015):30–41.

Boerstra, A.C., M. te Kulve, J. Toftum, M.G.L.C. Loomans, B.W. Ole-sen, and J.L.M. Hensen. 2015. Comfort and performance impact of personal control over thermal environment in summer: Results from a laboratory study. Building and Environment 87:315–26. Brager, G.S., G. Paliaga, R.J. de Dear, B.W. Olesen, J. Wen, J.F. Nicol,

and M. Humphreys. 2004. Operable windows, personal control, and occupant comfort. ASHRAE Transactions 110(Part I):17–35. Cunningham, J. 1991. Body composition a synthetic review general

pre-diction of energy expenditure: As a determinant and a proposed.

American Journal of Clinical Nutrition 54(6):963–69.

(18)

Dalewski, M., A.K. Melikov, and M. Vesely. 2014. Performance of duct-less personalized ventilation in conjunction with displacement ven-tilation: Physical environment and human response. Building and

Environment 81:354–64.

Demeter, M., and P. Wichman. 1989. Personal environmental module. United States Patent no. 4,872,397. United States.

Haldi, F., and D. Robinson. 2010. On the uniﬁcation of thermal percep-tion and adaptive acpercep-tions. Building and Environment 45(11):2440– 2457.

Hellwig, R.T. 2015. Perceived control in indoor environments: A concep-tual approach. Building Research & Information 43(March):302–15. Jackson, A., and M. Pollock. 1985. Practical assessment of body

com-position. The Physician and Sports Medicine 13:76–90.

Kaczmarczyk, J., and A.K. Melikov. 2006. Human response to ﬁve designs of personalized ventilation. HVAC&R 12(2):37–41. Karjalainen, S., and O. Koistinen. 2007. User problems with

indi-vidual temperature control in oﬃces. Building and Environment 42(8):2880–7.

Leaman, A., and B. Bordass. 1999. Productivity in buildings: The “killer” variables. Building Research & Information 27:4–19. Lipczynska, A., J. Kaczmarczyk, and A.K. Melikov. 2015. Impact of

Per-sonalized Ventilation Combined with Chilled Ceiling on Sbs Symp-toms Intensity. Healthy buildings Europe 2015. The Netherlands:

Eindhoven.

Luo, M., B. Cao, W. Ji, Q. Ouyang, B. Lin, and Y. Zhu. 2016. The under-lying linkage between personal control and thermal comfort: Psy-chological or physical eﬀects? Energy and Buildings 111:56–63. Melikov, A.K., T. Ivanova, and G. Stefanova. 2012. Seat

headrest-incorporated personalized ventilation: Thermal comfort and inhaled air quality. Building and Environment 47:100–8.

Melikov, A.K., B. Krejciríková, J. Kaczmarczyk, M. Duszyk, and T. Sakoi. 2013. Human response to local convective and radi-ant cooling in a warm environment. HVAC&R Research 19(8): 1023–32.

Pallubinsky, H., L. Schellen, T.A. Rieswijk, C.M.G.A.M. Breukel, B.R.M. Kingma, and W.D. van Marken Lichtenbelt. 2016. Local cooling in a warm environment. Energy and Buildings 113: 15–22.

Pasut, W., H. Zhang, E.A. Arens, and Y. Zhai. 2014. Energy-eﬃcient comfort with a heated/cooled chair. Proceedings of the 13th

Inter-national Conference Indoor Air Conference 2014, Hong Kong, July

7-12, 2014.

Pasut, W., H. Zhang, S. Kaam, E.A. Arens, J. Lee, F.S.P.E. Bauman, and Y. Zhai. 2013. Eﬀect of a heated and cooled oﬃce chair on thermal comfort. HVAC&R Research 19(5):574–83.

Schellen, L., M.G.L.C. Loomans, M.H. de Wit, B.W. Olesen, and W.D. van Marken Lichtenbelt. 2012. The influence of local effects on thermal sensation under non-uniform environmental conditions— Gender differences in thermophysiology, thermal comfort and pro-ductivity during convective and radiant cooling. Physiology &

Behavior 107(2):252–61.

Suzuki, I., K. Washinosu, and T. Nobe. 2010. Adaptive eﬀect to thermal comfort of cool chair in ZEB oﬃce. 7th Windsor Conference: The

changing context of comfort in an unpredictable world, Windsor, UK,

April 12–15, pp. 12–5.

Verhaart, J.C.G., M. Vesely, R. Li, and W. Zeiler. 2015a. Climate cham-ber tests for measuring performance characteristics of a personal cooling system. In 2015 ASHRAE Annual Conference, Atlanta, GA, June 27–July 1.

Verhaart, J., R. Keune, R. Li, and W. Zeiler. 2015b. Personal cooling using thermal conduction on the desk. In CISBAT 2015, Lausanne, Switzerland, September 9-11.

Verhaart, J., M. Veselý, and W. Zeiler. 2015c. Personal heating: Eﬀec-tiveness and energy use. Building Research & Information 43(3): 346–54.

Veselý, M., P. Molenaar, M. Vos, R. Li, and W. Zeiler. 2017. Personal-ized heating—Comparison of heaters and control modes. Building

and Environment 112:223–32.

Vischer, J.C. 2008. Towards a user-centred theory of the built environ-ment. Building Research & Information 36(3):231–40.

Watanabe, S., T. Shimomura, and H. Miyazaki. 2009. Thermal evalua-tion of a chair with fans as an individually controlled system.

Build-ing and Environment 44(7):1392–8.

Wever, R., J. van Kuijk, and C. Boks. 2008. User-centred design for sus-tainable behaviour. International Journal of Sussus-tainable Engineering 1(1):9–20.