key: cord-0942169-6hxk1iil
authors: Blocken, B.; van Druenen, T.; Ricci, A.; Kang, L.; van Hooff, T.; Qin, P.; Xia, L.; Ruiz, C. Alanis; Arts, J.H.; Diepens, J.F.L.; Maas, G.A.; Gillmeier, S.G.; Vos, S.B.; Brombacher, A.C.
title: Ventilation and air cleaning to limit aerosol concentrations in a gym during the COVID-19 pandemic
date: 2021-02-04
journal: Build Environ
DOI: 10.1016/j.buildenv.2021.107659
sha: 6df77a5cdf53548ef85f15056312ed1dc4eb1887
doc_id: 942169
cord_uid: 6hxk1iil

SARS-CoV-2 can spread by close contact through large droplet spray and indirect contact via contaminated objects. There is mounting evidence that it can also be transmitted by inhalation of infected saliva aerosol particles. These particles are generated when breathing, talking, laughing, coughing or sneezing. It can be assumed that aerosol particle concentrations should be kept low in order to minimize the potential risk of airborne virus transmission. This paper presents measurements of aerosol particle concentrations in a gym, where saliva aerosol production is pronounced. 35 test persons performed physical exercise and aerosol particle concentrations, CO(2) concentrations, air temperature and relative humidity were obtained in the room of 886 m³. A separate test was used to discriminate between human endogenous and exogenous aerosol particles. Aerosol particle removal by mechanical ventilation and mobile air cleaning units was measured. The gym test showed that ventilation with air-change rate ACH = 2.2 h(−1), i.e. 4.5 times the minimum of the Dutch Building Code, was insufficient to stop the significant aerosol concentration rise over 30 min. Air cleaning alone with ACH = 1.39 h(−1) had a similar effect as ventilation alone. Simplified mathematical models were engaged to provide further insight into ventilation, air cleaning and deposition. It was shown that combining ventilation and intensive air cleaning can reduce aerosol particle concentrations by factors of 2.3 up to 3.7 depending on aerosol size, compared to ventilation alone. This combination of existing ventilation supplemented with air cleaning is energy efficient and can also be applied for other indoor environments.

particles. These particles are generated when breathing, talking, laughing, coughing or sneezing. It can be 23

assumed that aerosol particle concentrations should be kept low in order to minimize the potential risk of 24 airborne virus transmission. This paper presents measurements of aerosol particle concentrations in a gym, 25

where saliva aerosol production is pronounced. 35 test persons performed physical exercise and aerosol 26 particle concentrations, CO 2 concentrations, air temperature and relative humidity were obtained in the room of 27 886 m³. A separate test was used to discriminate between human endogenous and exogenous aerosol particles. 28

Aerosol particle removal by mechanical ventilation and mobile air cleaning units was measured. The gym test 29

showed that ventilation with air-change rate ACH=2.2 h -1 , i.e. 4.5 times the minimum of the Dutch Building 30

Code, was insufficient to stop the significant aerosol concentration rise over 30 minutes. Air cleaning alone 31 with ACH=1.39 h -1 had a similar effect as ventilation alone. Simplified mathematical models were engaged to 32 provide further insight into ventilation, air cleaning and deposition. It was shown that combining ventilation 33 and intensive air cleaning can reduce aerosol particle concentrations by factors of 2.3 up to 3.7 depending on 34 aerosol size, compared to ventilation alone. This combination of existing ventilation supplemented with air 35 cleaning is energy efficient and can also be applied for other indoor environments. 36 37

Keywords: COVID-19; Aerosol; Sports club; Fitness center; Building ventilation; Air purifier 38 39

1. Introduction 40

In the second week of 2021, the European Centre for Disease Prevention and Control reported 94,582,873 41 cases of SARS-CoV-2 including 2,036,713 deaths, world-wide [1] . It has been suggested that this virus can be 42 transmitted by respiratory droplets and by contact routes [2] [3] [4] [5] [6] [7] . Direct transmission can occur when infective 43 droplets produced by activities such as talking, laughing, coughing or sneezing reach the mucosae (mouth and 44 nose) or conjunctiva (eyes) of another person. Indirect or contact route transmission can occur via handrails, 45 keyboard buttons and other objects, where virus is deposited after contact with an infected person. There is 46 mounting evidence that the virus can also be transmitted by inhalation of saliva aerosol particles because the 47 virus has been found in small aerosol particles that can remain in the air for hours, and it has been shown to 48 maintain viability in such aerosols [8] [9] [10] [11] [12] . Therefore, precautionary measures should not only be applied for the 49 direct transmission route and the contact route, but also for the airborne route. 50

Respiratory droplets are generated from the fluid lining of the respiratory tract during expiratory 51 activities such as breathing, talking, laughing, coughing and sneezing [13] [14] [15] [16] . A single sneeze can produce 52 10,000 droplets or more [17] . A cough can produce from 100 up to 1000 droplets or more. Talking can produce  53 about 50 droplets per second [18] . On an hourly or daily basis however, normal mouth breathing is assumed to 54 generate more aerosol particles than coughing or sneezing because the latter are intermittent events [13, [19] [20] [21] [22] . 55 Expired droplet sizes can range from about 0.1 µm to 1 mm [16] . Large droplets will generally settle 56 rather quickly due to gravity and therefore can only contribute to virus transmission between individuals in 57 close proximity. This is why "social distancing" has been introduced in countries around the world, although 58 there is no strict consensus on the distance to be kept and the currently used 1.5 m, 1.8 m or 2 m distance is 59 in answering the first and third question. The study does not explicitly focus on infection risk but on ventilation 120 and air cleaning as measures to limit the build-up of aerosol concentrations in the indoor environment of a gym. 121

The paper is structured as follows. In section 2, some more information about the state-of-the-art in 122 aerosol particle production during physical exercise and about the state-of-the-art in ventilation and in air 123 cleaning is provided. Section 3 contains a short study to discriminate between human endogenous and 124 exogenous aerosol particles. Section 4 presents the measurement set-up and associated measurement results in 125

the gym under study. In sections 5 and 6, simplified mathematical modeling is applied to provide insight into 126 the effective ventilation, air cleaning and deposition fluxes, and to extrapolate the findings to scenarios with 127 longer exercise sessions and more air cleaning units. Sections 7 (discussion) and 8 (conclusions) conclude the 128 paper. 129 130 2. Aerosol production, ventilation and air cleaning in gyms 131 132

2.1. Gyms and aerosol particle production during physical exercise 133 134

A gym is an environment that houses equipment and services for the purpose of physical exercise. A gym was 135 selected as a case study for several reasons. First, respiratory aerosol particle production and aerosol particle 136 inhalation in gyms is expected to be more pronounced than in many other indoor environments. Although there 137

are only a few studies that provide some indirect indication of how physical exercise influences the emission of 138 respiratory droplets, these studies are consistent in indicating an overall substantial increase in aerosol 139 expiration due to more intensive breathing compared to tidal breathing. Johnson and Morawska [13] found that 140 deep exhalation resulted in a 4 to 6-fold increase in aerosol particle concentration. Rapid inhalation produced a 141 further 2-to 3-fold increase in concentration, while rapid exhalation had little effect on the measured 142 concentration. Almstrand et al. [14] studied the effect of airway opening on aerosol particle production. Test 143 subjects performed different breathing maneuvers in which the initial lung volume preceding an inhalation to 144 total lung capacity was varied between functional residual capacity (FRC; the volume of air in the lungs at the 145 end of passive expiration) and residual volume (RV; the volume of air in the lungs after full exhalation person-based approach in which the minimum fresh air ventilation rates in dm³/s per person are stipulated. The 211 minimum values for different types of buildings are given in Table 1 , where a distinction is made between new 212 and existing buildings. Table 1 shows that the required flow rates for indoor sports centers are higher than for 213

shops but -for new buildings -lower than for educational buildings and identical to those of industrial and 214 office buildings. This does not seem to be aligned with the expected higher aerosol particle production during 215 physical exercise [13, 14] . 216

In 2008, the Dutch "Guidebook for Sports Accommodations" was published by the NOC*NSF [77]. 217

The NOC*NSF (Dutch Olympic Committee * Dutch Sports Federation) is the overarching organization for all 218 sports activities, professional and recreational, in the Netherlands. In 2014, specific guidelines for sports 219 facilities for people with disabilities were published by a consortium of organizations including the NOC*NSF 220

[78]. These guidelines stipulate a minimum ventilation flow rate of 11.1 dm³/s per exercising person for sports 221 halls, which is 70% above the minimum required value in the Dutch Building Code for new buildings and even 222 3.2 times higher for existing buildings (see Table 1 ). The Guidebook [77] even suggests a total of 6 air change 223 rates per hour (ACH) for fitness spaces, which implies that the volume of air in the room is replaced by fresh 224 air 6 times per hour. For aerobics and martial arts spaces, it advices ACH = 8 h -1 and for indoor cycling ACH = 225 10 h -1 . These higher values seem better aligned with the expected higher production of heat, vapor, CO 2 and 226 aerosol particles by people during physical exercise. 227

In view of the COVID-19 pandemic, ASHRAE, the American Society of Heating, Refrigerating and 228

Airconditioning Engineers, has acknowledged the potential for aerosol transmission of SARS- are also referred to as professional ACs. ACs should have a sufficiently high aerosol particle removal 244 efficiency and a sufficiently high volume flow rate, in comparison to the room volume to be treated. Fisk et al. 245 [82] stated that filter efficiencies above 85% provide only modest gains in performance. Several authors 246 mentioned that the air flow rates must be at least several ACH to obtain substantial particle reductions [82] [83] [84] [85] . 247

The To the best of our knowledge, at the moment of writing this paper, the application of air cleaning in 265 public spaces in the Netherlands and many other European countries is rather rare, even though AC technology 266

is not new and the COVID-19 pandemic is already more than a year old. This is partly attributed to the 267 sometimes less good reputation of commercially available ACs and the inferior performance of some of these 268

ACs. Measurements were conducted to provide a first indication on the amount of endogenously versus exogenously 283 generated aerosol particles during physical exercise. It is known that the amount of endogenously generated 284 saliva aerosols is small compared to the particle concentrations typically found in outdoor and indoor 285 environments [92, 93] . So breathing only provides small additions to PM concentrations although it is these 286 small amounts of saliva aerosol particles that are of concern in view of the spread of infectious diseases. Tests 287 were performed in a 3.9 x 2.7 x 2.3 m³ = 24.2 m³ airtight stainless steel test room (Fig. 1 ). The room was 288 equipped with a stationary bicycle in the center, an AC, a fan for generating well-mixed indoor conditions and 289 three Grimm 11D aerosol particle sizers (APS) with a measurement range from about 0.25 to about 30 µm [94]. 290

There was no supply or exhaust of air from the room. Three healthy human volunteers, aged 20 to 22 years and 291 accustomed to regular physical exercise, participated in this study. Approval for use of human subjects was 292 obtained from the Ethical Review Board of Eindhoven University of Technology with file number 293

ERB2020BE58. The subjects signed an informed consent form prior to participating in the study. Every subject 294 performed two times a session of 30 minutes exercise on the stationary bicycle in heart rate zones 3 and 4. In 295 the first session, the subject released its breath freely into the room and the APS measured the aerosol particles 296 from the different sources. In the second session, the subject released its breath via a mask into a tube that was 297 connected to the outside environment. The mixing fan was operated during the two sessions. Prior to every 298 session, the mask and tube were disinfected and air cleaning was performed while operating an additional fan 299 inside during at least 30 minutes to reduce the aerosol particle concentration. Assuming that the amount of 300 endogenously and exogenously generated particles was similar in both sets by the same individual, the 301 difference between both sessions provided an indication of the amount of endogenous aerosol particles. The 302

subjects were given at least 30 minutes rest between the two exercise sessions and were provided with a bottle 303 of drinking water to be consumed in the rest period. The temperature was 21°C and the RH ranged between 55 304 and 65%. Subject 1 had short hair, a short beard, wore a short-sleeved shirt and short trousers and applied a 305

pedaling frequency of about 90 rpm. Subject 2 had medium long hair, no beard, wore a short-sleeved shirt and 306

short trousers and applied a pedaling frequency of about 70 rpm. Subject 3 had short hair, a short beard, wore a 307

short-sleeved shirt and long trousers and applied a pedaling frequency of about 80 rpm. 308 309

3.2. Results 310 311

Because only two measurements sessions were performed for only three persons, the results only provide a 312 first indication on the proportion of exogenous versus endogenous particle concentrations. Table 2 lists the 313 resulting aerosol particle concentrations in five size fractions: 10 to 2.5 µm, 2.5 to 1 µm, 1 to 0.5 µm, 0.5 to 314 0.25 µm and below 0.25 µm, averaged over the three measurement locations (Fig. 1 µm diameter per m³. Additional variations in the exogenous particle emission among the subjects could be 324 attributed to the different pedalling frequencies, clothing and hair style. 325

• The results indicate a very high inter-subject variability for the endogenous particle emission. This is in line 326

with previous studies that also showed very large variability [13,14,20,31]. The first subject yielded only 327 very low concentrations of saliva aerosol particles, only significant in the size ranges below 1 µm. The 328

second subject yielded higher concentrations of saliva aerosol particles, only detectable in the size range 329 above 1 µm. For the largest particles (> 2.5 µm), the concentration of endogenous particles was in the order 330 of magnitude of that of the exogenous particles. Finally, the third subject emitted saliva aerosol particles in 331 all size ranges, with the same order of magnitude as the exogenous particles, for all size ranges. 332

• Apart from the largest size range, both the exogenously and endogenously generated particle concentrations 333 showed an increasing trend over time in the 30-minute sessions. 334

Note that the APS itself did not allow to discriminate between solid and liquid particles and that all 335 concentrations were obtained by assuming the particles had a density similar to that of saliva (1002-1006  336 kg/m³), as the density of the actual solid particles was unknown. Therefore, it could be assumed that solid 337

(exogenous) particle concentrations as mentioned in Table 2 The measurements were performed in the fitness 3 room of the Student Sports Center at Eindhoven University 345

of Technology in the Netherlands. Figure 2 shows the plan view. The room was split in two parts by a vertical 346 screen and only the south part was used for this study. The floor area of this part is 173.7 m² and the height was 347 5.1 m, yielding a room volume of about 886 m³. The ventilation system in the fitness room was a mechanical 348 mixing ventilation system by which fresh air was supplied into the room by openings with swirl diffusors in 349 the ceiling (indicated with p1-p10 in Fig. 3 ). The openings p4 to p8 were situated in the half of the room used 350

in the present study. The exhaust openings were present on the west side of the room, near the ceiling (Fig. 3) . 351

The J o u r n a l P r e -p r o o f existing buildings, assuming a near-full occupancy with 35 persons (see Table 1 ). However, note that this ACH 357 is considerably lower than the recommended value of ACH = 6 h -1 in [77]. 358 Figure 4 shows a perspective view of the measurement set-up in the gym with the cardio and weight 359 machines. We focused on a gym with cardio equipment consisting of stationary exercise bicycles and 360 treadmills and with workout equipment consisting of weight-based exercise machines. When using this 361 equipment, the people exercising are not moving throughout the room but instead remain confined at a rather 362 fixed position in the room, which was aimed at limiting resuspension of particles. [97] , the CADR is reduced as the particle size increases because larger 371

particles fall under the influence of gravity and have a relatively higher deposition rate. Test room 372 measurements indicated that the ACs did not produce substantial amounts of harmful byproducts NO x and O 3 . 373

The Grimm sensors were mounted at measuring heights of 1.367 m and 1.247 m. The 110 APS were mounted 374 on vertical poles at heights of 0.5, 1.0 and 1.5 m (Fig. 4) rooms in a hotel in Eindhoven city center and finally subjected to additional safety precautions on the 386 measurement day of 11 July 2020. While all 55 persons tested negative for COVID-19, after stringent 387 application of the safety protocol, 5 test subjects were excluded from the measurement campaign and 35 388 subjects remained. These 35 subjects performed cardio and/or weight machine exercises in sessions of 30 389 minutes (Fig. 4, 5) . 390

Six of the experimental scenarios or 30-minute sessions are listed in Table 3 . The scenarios can be 391 grouped in three sets: Set 1: ventilation on and air cleaning off; Set 2: ventilation off and air cleaning off; Set 3: 392 ventilation off and air cleaning on. Within every set, the parameter is the physical exercise: present or not. In 393 those scenarios when physical exercise was not conducted, all subjects were removed from the room and 394 directed to a large sports hall where they waited for the next exercise session. In between sessions, the subjects 395

were provided with drinking water and sandwiches. 396

All exercise sessions were performed in the same way. The subjects were divided into two groups: 397 cardio workout (CW) (16 subjects) and strength training (ST) (19 subjects). Within the strength training group, 398 10 subjects followed the protocol for muscle endurance (STME) and 9 subjects followed the protocol for 399 muscle mass (STMM). There were three 30-minute sessions in which the subjects had to follow a cardio 400 workout (once or twice) and performed a strength training (one or twice). The CW performed their training for 401 30 minutes at an intensity between 60-75% of heart rate reserve. This was measured by a TICKR heart rate belt 402

connected to the machine. An additional task was that subjects should be able to continue talking to each other 403 not to end up with a too high exercise intensity. Both the STME and the STMM performed three sets of 20 or 404 10 repetitions on three different machines. Each repetition started with a start signal and lasted 3 minutes. After 405 the execution of the exercise, the subjects were given rest until the next start signal. On the last machine they 406

performed an extra set to complete the 30 minutes. In the CW the subjects had a choice of machine. The 407

following CW machines were used in every session: 10 treadmills (LifeFitness, Elevation series), 2 Powermill 408 climbers (LifeFitness, Elevation series) and 4 upright exercise bikes (LifeFitness, Elevation series). STME 409 performed this protocol on the following machines (LifeFitness, Circuit Series): leg extension, seated row, 410

chest press, seated leg curl, ab crunch, lat pulldown, triceps press, squat, shoulder press, biceps curl. STMM 411 performed this protocol on the following machines (LifeFitness, Optima Series): leg extension, seated leg curl, 412

chest press, seated row, hip abduction, hip adduction, biceps curl, shoulder press, machine fly. Every row of two figures represents one set, as outlined above. The results are presented as concentrations in 420 the size fractions 10-2.5 µm, 2.5-1 µm, 1-0.5 µm, 0.5-0.25 µm and the fraction below 0.25 µm. To aid in 421

interpreting the semi-logarithmic graphs, results are shown in Figure 7 . The following observations are made: 427

• As a general comment, evaporation is not considered a factor here because this process is very fast, 428 therefore all measured concentrations are expected to be those of the droplet nuclei. 429

• Figure 6a shows that when physical exercise was performed and the ventilation system was engaged (with 430

ACs off), the concentrations of aerosol particles in all size fractions increased almost monotonically. The 431

ventilation system was clearly not effective in avoiding the rise in aerosol concentrations within the 30-432 minute period. After 30 minutes, the subjects ceased their exercise. 433

• Figure 6b demonstrates that after the physical exercise had ceased and after everybody had left the room, 434

the ventilation system was effective in reducing the aerosol concentrations -almost monotonically -in all 435 size fractions during the period of 30 minutes. 436

• Figure 6c depicts the rise in aerosol concentrations when physical exercise was performed and neither 437 ventilation nor ACs were engaged. The increase in the fraction 2.5 to 10 µm was most pronounced in the 438 first 10 minutes, while afterwards the concentration in this fraction remained quite constant. In the other 439 size fractions there was an almost monotonic increase. 440

• Figure 6d shows that when physical exercise had ceased, people had left the room and ventilation remained 441 turned off, there was a substantial concentration decrease especially in the largest size fraction versus a 442 much more limited decrease in the smaller fractions, both of which are attributed to natural deposition in 443 the calm indoor environment. 444

• Figure 6e shows the increase when exercise was performed, ventilation was turned off but the ACs were 445 engaged. It is clear that also air cleaning alone, at the flow rate provided, was not sufficient to limit the rise 446 in aerosol concentrations within the 30-minute time period. 447

• Figure 6f shows that the ACs were also effective in reducing the aerosol particle concentrations after the 448 exercise had halted, people had left the room but the two ACs remained active. 449

• Comparing Figure 6b with 6f and rows 2 and 6 in Table 4 , the aerosol particle concentration reductions by 450 ventilation versus ACs were quite similar, with the ACs appearing to have been even more effective than 451 ventilation in several size fractions. This in spite of the fact that the ventilation ACH was 2.20 h -1 while the 452 air cleaning ACH was 1.39 h -1 , which is a 58% difference. Note however that ventilation also injects a 453 small portion of PM into the room (i.e. the concentration in the outdoor air after filtering in the mechanical 454 ventilation system -see section 6). 455

• Figure 7 depicts the 5-minute CO 2 concentrations throughout each of the six sessions as an average of the 456 values measured by the 110 AQS2020PRO APS at 0.5, 1.0 and 1.5 m height. The first session shows an 457 almost doubling of the concentration due to the physical exercise in spite of the ventilation system being 458

active. The second session shows the concentration decay due to ventilation. In the third and fifth session, 459

there are strong rises in concentration due to the absence of ventilation. Note that the ACs do not affect the 460 CO 2 concentration, as shown for sessions 4 and 6 where this concentration remains fairly constant. 461

A simplified mathematical model can be used to assess the effective ventilation rate. The model assumes a 464 uniform CO 2 concentration in the room with volume V, in other words: perfect mixing of the generated CO 2 465 and of the supplied ventilation air with the CO 2 . It also assumes a steady release of CO 2 by the subjects. With 466 these assumptions, the mass balance for CO 2 can be written as:

With c the CO 2 concentration (ppm), G the CO 2 emission rate (ppm.m³/h), Q V the ventilation rate (m³/h) and c 0 471 the CO 2 concentration (ppm) in the supplied ventilation air. For scenario 1 (physical exercise and ventilation), 472 the solution of this first-order ordinary differential equation is:

For scenario 2 (only ventilation), the solution is: 477 478 Least squares fitting of Eq. (5) to the data of CO 2 in Figure 7 yields Q V = 995 m³/h. The CO 2 production by the 487 subjects in scenarios 1, 3 and 5 is obtained by fitting Eqs. (4), (6) and (7) to the data in Figure 7 , yielding G between these two estimates. The value of Q V implies that the effective ventilation rate is only 51% of the 496 actual supply ventilation flow rate of 1948.6 m³/h. This is attributed to fact that the simplified mathematical 497 model assumes a uniform CO 2 source, a uniform concentration distribution and a uniform effect of the 498 ventilation system. In reality, the CO 2 generation occurred at test person height and the measurements were 499 conducted close to the CO 2 source, while both the ventilation supply openings and the exhaust openings were 500 positioned near the ceiling. Q V = 995 m³/h could therefore be considered as the "effective" or local ventilation 501 flow rate for the zone in the lower part of the room, in which the test persons were present. This "effective" 502 ventilation rate is more than twice that required by the Dutch Building Code (i.e. 433 m³/h), however, 503 evaluation of ventilation systems with regard to building codes generally occurs based on the total supply 504 ventilation flow rate. The variability in CO 2 emission could be attributed to subjects having performed more or 505 less intensive exercise from one session to another, subject fatigue, subjects switching from cardio to weight 506 machines and some inter-subject variability in CO 2 emission under similar physical exercise. 507 508

6. Simplified mathematical model for aerosol particle concentrations 509 510

6.1. Aerosol particle production, deposition, ventilation and air cleaning 511 512

We consider the five size fractions also used in Figure 6 : 10-2.5 µm, 2.5-1 µm, 1-0.5 µm, 0.5-0.25 µm and 513 below 0.25 µm. Let G denote the aerosol particle production rate by physical exercise, which is the sum of the 514 production rates by respiration, resuspension, machine component friction, clothing friction and the like. Q V 515 denotes the ventilation flow rate, Q AC the total AC flow rate, η AC the AC efficiency, K N the natural deposition 516 loss rate under calm indoor airflow conditions (ventilation and ACs off), K V the deposition loss rate in the 517 ventilation flow regime (ventilation on, ACs off), K AC the deposition loss rate in the AC flow regime 518

(ventilation off, ACs on), V the room volume and c 0 the concentration in the incoming ventilation air. 519

Assuming well-mixed conditions and a steady emission of aerosol particles in all five size fractions by the 35 520 subjects, the mass balances for the six scenarios in Figure 6 for a given size fraction are: Considering the left-hand sides of these equations as known by the data in Figure 6 , these six equations have 549 seven unknowns: G 1 , G 3 , G 5 , K V , K N , η AC Q AC and K AC . Note that η AC Q AC cannot be considered known from 550 the CADR tests in the small test room of 24.2 m³, as reported in subsection 4.1, because Noh and Oh [97] 551 showed that for the same AC device, the experimental CADR decreased as the size of the test chamber 552

increased. To solve the system of equations, the sum η AC Q AC +K AC V is taken as a single variable. Least squares 553 fitting of Eqs. (15), (17) and (19) to the data in Figure 6 yields the values of Q V + K V V, K N V and η AC Q AC + 554 K AC V for every size fraction. Using these values into Eqs. (14), (16) and (18) yields the values of G 1 , G 3 and 555 G 5 . K V V is calculated based on Q V = 995 m³/h (see section 5). Table 5 holds the results. It also shows the 556 deposition loss rates K N and K V based on V = 886 m³. The last row shows the measured c 0 concentration 557 values. The following observations are made: 558

• The low measured c 0 values are representative of a large degree of air filtering in the mechanical ventilation 559 system. 560

• Aerosol particle removal due to deposition in scenarios 2 (K V ) and 4 (K N ) rapidly decreases with decreasing 561 size fraction. This is expected given the lower mass and associated smaller settling velocities of the smaller 562 aerosol particles. Table 5 are situated in these ranges. The larger values in the two smallest size 569 fractions could be attributed to the large number of surfaces in the gym room. 570

• Aerosol particle removal due to deposition in scenario 4 (K N; ventilation off, ACs off) is much less 571 pronounced than in scenario 2 (K V ; ventilation on, ACs off), which is attributed to the indoor airflow 572 pattern in the latter scenario generated by the ventilation system. Indeed, Friedlander and Johnstone [108] 573 demonstrated the strong increase in deposition from turbulent gas streams with increase in the flow 574

Reynolds number, attributed to the larger eddies and larger inertial forces favoring deposition. For the size 575 fraction 10-2.5 µm, the deposition rate is 2.2 times larger in scenario 2 than in 4, while for the size fraction 576 0.5-0.25 µm, it is 4.8 times larger. 577

• Earlier, it was shown by comparing Figure 6b with 6f and rows 2 and 6 in Table 4 , that the aerosol particle 578 concentration reductions by ventilation versus ACs were quite similar, with the ACs appearing a bit more 579 effective than ventilation in several size fractions. This in spite the fact that the ventilation ACH was 2.20 h -580 1 while the AC ACH was 1.39 h -1 , which is a 58% difference. This is confirmed by the fact that the sum Q V 581 + K V V is larger than the sum η AC Q AC + K N,AC V. This is attributed to the lower effectiveness of the 582 ventilation system which is attributed to two reasons: (1) the presence of the ventilation inlet and outlets  583 near the ceiling and (2) the fact that the incoming ventilation air also contained -albeit fairly low -584

concentrations of aerosol particles. It is also attributed to the fact that the AC units were positioned in the 585 region where the aerosol particles were generated, which can explain their relatively larger effectiveness. 586

• The aerosol particle production rates are very different among scenarios 1, 3 and 5, with the differences 587 also differing per size fraction. This could be attributed to inter-subject variability in aerosol particle 588 emission under similar physical exercise regimes but also by subjects having performed more or less 589

intensive exercise from one session to another, subject fatigue and subjects switching from cardio to weight 590 machines. It could also be attributed, at least partly, due to the use of only two measurement points for 591 aerosol particle concentrations. 592

• In terms of the magnitude of aerosol particle production, You et al.

[109] measured the short-term emission 593 rates of particles by persons with different clothing and activity intensities in a sealed chamber. The 594 activities did not involve gym machines but included walking, upper body and arm movements. Based on 595 their data for a cotton suit and for slight to strong activity intensity and assuming a particle density of 1000 596 kg/m³, the following ranges can be derived for 35 persons: 5635 -39238 µg/h for the size fraction 10-2.5 597 µm, 2004 -2227 µg/h for 2.5-1 µm, ≈2738 µg/h for 1-0.5 µm and finally 760 -844 µg/h for the size 598 fraction below 0.5 µm. Taking into account that the 35 persons in the gym performed moderate rather than 599 strong activity and that the numbers by You et al.

[109] do not include particles generated by the friction 600 between components of the cardio and weight machines, the values of G 1 , G 3 and G 5 in Table 5 A scenario that was not considered in the experimental campaign was the combination of ventilation and air 605

cleaning. Therefore, the simplified model is applied to investigate this additional scenario with ventilation and 606 air cleaning combined and using the aerosol particle production rate from the first scenario (G 7 = G 1 ): 607 608 It is assumed that the combination of ventilation (with supply and exhaust near the ceiling) and air cleaning 612 (near ground level) also combines the deposition rates by both technologies. The simplified model was also 613 used for other scenarios, as shown in Figure 8 that presents the results of six scenarios for a 60-minute period, 614 all with the same aerosol generation rate G 1 . Figure 8a is the calculated result for scenario 1. Figures 8b and c  615 present scenarios 3' and 5' that are identical to scenarios 3 and 5 but now with G 3' = G 5' = G 1 . Figure 8d  616 presents scenario 7 (ventilation and air cleaning combined). Figure 8e and f present two additional scenarios in 617 which the number of ACs is raised from 2 to 4 and 6 units, respectively. All figures show that the 618 concentrations tend towards an asymptote over time, as dictated by the exponential functions in the above-619 mentioned equations. Figure 9 shows the asymptotic values as reached in every scenario at t = ∞, and Table 6  620 lists the corresponding values. The following observations are made: 621

• Figure 8 shows that the duration during which concentrations keep rising significantly is largest for 622 scenario 3' (Fig. 8b ; no ventilation, no ACs, only natural deposition in calm indoor airflow conditions). 623

Evidently this is also the scenario in which the highest concentrations are obtained. Figure 9 and Table 6  624 indicate that these concentrations go up to 27. • Figure 8 also shows that the duration at which near-equilibrium conditions are obtained is shortest for 629 scenario 9 in which most intensive air cleaning is engaged. Evidently this is also the scenario in which the 630 lowest concentrations are obtained. Figure 9 and Table 6 • For all other scenarios, the duration towards near-equilibrium and the near-final concentrations are situated 635 between those of scenarios 3' and 9. In scenario 1, ventilation alone reduces the final concentrations of 636 scenario 3' (no ventilation, no ACs) by factors 2.8, 6.6, 7.7, 9.8 and 8.1 for the size fractions 10-2.5 µm, 637 2.5-1 µm, 1-0.5 µm, 0.5-0.25 µm and below 0.25 µm, respectively. In scenario 5', air cleaning alone (with 638 2 ACs) yields slightly lower reduction factors: 2.5, 5.8, 6.3, 6.4 and 6.3, in relation to scenario 3'. 639

Combining Therefore, concerns about high aerosol particle concentrations in indoor sports centers, fitness centers and 656 gyms are justified. Regardless, more research is needed to assess aerosol particle emissions by persons 657

performing physical exercise at different levels of intensity and heart rate. 658

To the best of our knowledge, there was no study in the scientific literature that specifically focused on 659 respiratory aerosol production in fitness centers. There was even relatively little published research about air 660 quality in fitness centers in general, as opposed to residential buildings and other types of public spaces such as 661 schools and offices [62] [63] [64] . The few studies that were available in the scientific literature had measured PM 662

concentrations without a focus on saliva aerosol particles and without an attempt to discriminate between 663 endogenous and exogenous particles. To provide some first preliminary insights in the proportions between 664 endogenous versus exogenous particles, in the present study, a small test was performed. While some clear 665 trends could be discerned, especially for the size fractions below 2.5 µm, especially the large inter-subject 666 variability was noted. This however was in line with earlier studies that also indicated very large inter-subject 667

variability [13, 14, 20, 31] . Much more research is needed on endogenous versus exogenous aerosol particle 668 emission as the lack of knowledge about their relative proportions will continue to complicate advice 669 concerning saliva aerosol emission and reduction in view of limiting infection risk. Due to the inability to 670 discriminate between endogenous and exogenous particles in the gym study in the present paper, the focus in 671 the paper is on the combination of these two types. 672

The scenarios considered in the present study are neither a worst-case nor a most beneficial scenario. 673

On the one hand, in actual gym settings, many persons performing physical exercise will apply long breaks 674 between exercises, either to rest or to talk to other people. In that regard, the present study considered fairly 675 vigorous and continuous exercise, in an attempt to obtain a steady release of both endogenous and exogenous 676 aerosol particles. On the other hand, in intensive cycling sessions such as spinning, the intensity and the heart 677 rates are higher than those in the present study that included a combination of cardio and weight machines. The 678 measured CO 2 emission rates confirm that the present scenarios are in between vigorous and moderate exercise. 679

A wide variety of gyms and ventilation systems exist. Nevertheless, most gyms are characterized by a 680 large height and most gym have mixing ventilation systems with supply and exhaust openings near the ceiling. 681

In order to generalize the results on ventilation effectiveness from the present study, a number of additional 682 gyms will need to be investigated, after which potentially a common denominator could be defined and some 683 general advices could be established in terms of required ventilation and/or air cleaning flow rates. 684

Similarly, a wide variety of ACs exist. As mentioned in this paper, ACs need to have both a sufficiently 685 high efficiency and a proper capacity (flow rate) in order to be effective. ACs have a sometimes less good 686 reputation because of the presence of some very deficient and even harmful types on the market. However, also 687 high-quality ACs have been developed and are commercially available. High-quality certification and 688

international standardization are imperative. 689

Aerosol particle deposition is an important factor. The present study suggests that the engagement of 690 ventilation or air cleaning, by inducing an overall more turbulent airflow pattern in the room, substantially 691 enhances the deposition, in line with a previous study [108] . 692

In view of the COVID-19 pandemic and potential future pandemics, ventilation of indoor environments, 693

gyms included, will need to be reconsidered. At the same time, energy efficiency should be upheld to the 694 largest degree possible, in view of limiting climate change. Suggestions by politicians, scientists and opinion 695 makers that ventilation has to be massively incremented to avoid potential aerosol SARS-CoV-2 infection, 696

would unavoidably give rise to large investment costs to upgrade ventilation systems and large energy 697 consumption and losses (if heat recovery is not massively deployed) and the associated costs. Therefore, we 698 suggest to not engage in expensive upgrades of existing mechanical ventilation systems, on condition that they 699 already enable -pandemics aside -a healthy and comfortable indoor environment, using ventilation rates that 700 are above the minima required by building codes. Instead, these expensive and already available systems can 701 be supplemented with lower-cost mobile professional AC units. The present study has shown that the 702 effectiveness of high-quality AC units can be similar to that of a mechanical ventilation system (with aerosol  703 filtering) with a 60% higher flow rate. AC units do not require the air to be heated, cooled or (de)humidified, as 704 it is indoor air being handled and exhausted back into the room. Ventilation air coming from outside will often 705 need extra energy for heating, cooling and (de)humidifying, even if heat recovery is applied. However, it 706

should be stressed that ventilation at the minimum flow rates as required by building codes remains imperative, 707 because many ACs do not remove gasses, such as CO 2 . 708

A gym is rather complex indoor environment in the sense that it has a large height, the sources are 709 present near the floor while generally the ventilation supply and exhaust openings are present near the ceiling. 710

Therefore, future work should consider measuring aerosol particle concentrations not only at two positions at 711 similar height as in the present study, but also measuring concentration gradients along the height of the room. 712

Given the large height, vertical concentration gradients could be present, irrespective of the type of mixing 713

ventilation system or ACs that are present. Future work will include CFD simulations to provide more inside 714

into the vertical gradients and the related effectiveness of ventilation and AC units. 715

The objects. There is mounting evidence that it can also be transmitted by inhalation of infected saliva aerosol 732

particles. These particles are generated when breathing, talking, laughing, coughing or sneezing. It can be 733 assumed that aerosol particle concentrations indoors should be kept low in order to minimize the potential risk 734 of airborne virus transmission. This paper presents measurements of aerosol particle concentrations in a gym, 735

where saliva aerosol production is pronounced. 35 test persons performed physical exercise and aerosol 736 particle concentrations, CO 2 concentrations, air temperature and relative humidity were obtained in the room of 737 886 m³. A separate test was used to provide some information on the amount of human endogenous versus 738 exogenous aerosol particles. This test showed large inter-subject variability, with one person emitting much 739 more exogenous than endogenous particles, while another emitted similar amounts of both types. to the lower effectiveness of the ventilation system due to two reasons: (1) the presence of the ventilation inlet 745 and outlets near the ceiling and (2) the fact that the incoming ventilation air also contained -albeit fairly low -746

concentrations of aerosol particles. It was also attributed to the fact that the AC units were positioned in the 747 region where the aerosol particles were generated, which can explain the relatively larger effectiveness of ACs. 748

Simplified mathematical models were engaged to provide further insight into ventilation, air cleaning and 749

deposition. It was shown that combining ventilation and intensive air cleaning with up to six AC units with a 750 total ACH of 4.17 h -1 -as recommended in the scientific literature -can reduce the concentrations by factors 751 of 2.3 up to 3.7 depending on aerosol size, compared to ventilation alone, and by factors of 10.3 up to 23.9 752 depending on aerosol size, compared to a situation without ventilation and AC units. It is suggested that if 753 aerosol particle concentrations need to be reduced in view of the COVID-19 pandemic, this should not 754 necessarily be done by an expensive upgrade of the existing mechanical ventilation system. Instead, it could 755 also be achieved by supplementing this system with mobile professional high-quality AC units. When the AC 756 units are installed near ground-level in gyms with large height (e.g. 5 m), they can have a higher effectiveness 757 FIGURE AND J o u r n a l P r e -p r o o f 

This lowers investment and operational costs because 759 AC units do not require the air to be heated, cooled or (de)humidified, as it is indoor air being handled and 760 exhausted back into the room. Ventilation air coming from outside will often need extra energy for heating, 761 cooling and (de)humidifying, even if heat recovery is applied. However, it should be stressed that ventilation at 762 (at least) the minimum flow rates required by building codes remains imperative

Conflict of interest and funding 766 Part of this work was funded by the Topteam Sports in the Netherlands. The other part was self-funded. None 767 of the authors had any involvement in the anonymous peer-review process. The review process

The authors acknowledge the many organizations that directly or indirectly enabled this research project. The 772

Topteam Sports of the Netherlands is acknowledged for having funded part of this project

Merit Cloquet (Sports Innovation Offer at Sportinnovator), Paul 774 van der Kolk (Programme Manager at Sportinnovator) and Iris Nijland

Sportinnnovator) are acknowledged for having brought the many partners in this project together, in record 776 time, around the central goal of investigating options to allow gyms to stay open safely in times of COVID-19

PlasmaMade) are 778 acknowledged for having provided the air cleaning units, the 110 AQS2020PRO sensors, their overall support 779 during the set-up of the measurement campaign and many valuable discussions on the design, set-up and 780 results of the campaign. Jan Hazelhof is acknowledged for his assistance with the regular medical disinfection 781 of the measurement equipment on the measurement day

Technology are acknowledged for having allowed and 783 facilitated the measurements in their building. Medical doctors Dr. Jelle Oosterhof and Dr. Wouter Bisseling 784 are acknowledged for having attended the test day in case medical assistance would have been required

Jorritsma, the Mayor of the city of Eindhoven and Stijn Steenbakkers

Jaap van Dissel, Director of the Centre for Infectious Disease Control 788 (Cib) of the Netherlands National Institute for Public Health and the Environment (RIVM) is acknowledged for 789 his advice concerning the required COVID-19 testing procedure as part of the safety protocol. All test persons, 790 most of them students of Eindhoven University of Technology, are acknowledged for their enthusiastic 791 collaboration -one student even flew over on

The authors are also grateful to the anonymous peer 794 reviewers whose valuable comments have improved this manuscript

European Centre for Disease Prevention and Control

A familial cluster of pneumonia associated with the 2019 novel 802 coronavirus indicating person-to-person transmission: a study of a family cluster

Early transmission dynamics in Wuhan, China, of novel coronavirus-infected 805 pneumonia

Clinical features of patients infected with 2019 novel coronavirus in 807 Wuhan

Active monitoring of persons exposed to patients with 809 confirmed COVID-19 -United States

Geneva: World Health Organization; 2020a

Modes of transmission of virus causing COVID-19: implications for IPC 816 precaution recommendations. Scientific Brief

Aerodynamic analysis of SARS-CoV-2 in two Wuhan hospitals

Aerosol and surface stability of SARS-CoV-2 823 as compared with SARS-CoV-1

Airborne transmission of SARS-CoV-2: the world should face the reality

A letter about the airborne transmission of 827 SARS-CoV-2 based on the current evidence

The coronavirus pandemic and aerosols: Does 829 COVID-19 transmit via expiratory particles?

The mechanism of breath aerosol formation

Effect of airway opening on production of exhaled 833 particles

Effect of voicing and articulation manner on aerosol 835 particle emission during human speech

The flow physics of COVID-19

Characterizations of particle size distribution of the droplets 839 exhaled by sneeze

Aerosol emission and superemission during human speech 841 increase with voice loudness

Paricle concentration in exhaled breath -Summary Report

The size distribution of droplets in the exhaled breath of healthy human 845 subjects

Inhaling to mitigate exhaled bioaerosols

Airborne spread of expiratory droplet nuclei between the occupants of indoor 849 environments: A review

On air-borne infection. Study II. Droplets and droplet nuclei

Transport characteristics of expiratory droplets and droplet nuclei in indoor 852 environments with different ventilation airflow patterns

How far droplets can move in indoor environments -revisiting 854 the Wells evaporation-falling curve

Aerosol Technology

Guidelines for Isolation Precautions in Hospitals

Transmission of infectious diseases during commercial air travel

Possible SARS coronavirus transmission during 862 cardiopulmonary resuscitation

Toward understanding the risk of secondary airborne infection: 864 emission of respirable pathogens

Size distribution and sites of origin of droplets 866 expelled from the human respiratory tract during expiratory activities

Relation between humidity and size of exhaled 868 particles

Violent expiratory events: on coughing and 870 sneezing

Turbulent gas clouds and respiratory pathogen emissions: potential implications for 872 reducing transmission of COVID-19

Flow in air conditioned rooms. English translation and revision of PhD Thesis

Velocity-pressure field of cross-876 ventilation with open windows analyzed by wind tunnel and numerical simulation

Prediction of room air motion by Reynolds-stress models

Comparing turbulence models for buoyant plume and displacement ventilation 881 simulation

Computational fluid dynamics and room air movement

Coupled urban wind flow and indoor natural ventilation modelling on a 885 high-resolution grid: a case study for the Amsterdam ArenA stadium

Airflow assessment in cross-ventilated buildings with 888 operable facade elements

CFD simulation of cross-ventilation for a generic isolated building: impact 890 of computational parameters

Computational Fluid Dynamics for Urban Physics: Importance, scales, possibilities, 892 limitations and ten tips and tricks towards accurate and reliable simulations

A computational and experimental investigation of the human thermal 895 plume

Role of ventilation in airborne transmission of infectious agents 897 in the built environment -a multidisciplinary systematic review

SARS-CoV-2 & Potential Airborne Transmission

European Centre for Disease Prevention and Control. 2020. Heating, ventilation and air-conditioning 902 systems in the context of COVID-19

How to operate HVAC and other building service systems to prevent the spread of the 908 coronavirus (SARS-CoV-2) disease (COVID-19) in workplaces. REHVA COVID-19 guidance 909 document

Exercise and outdoor ambient air pollution

Infection risk in gyms during physical exercise

Mobility network models of COVID-19 explain inequities and 915 inform reopening

Cluster of coronavirus disease associated with fitness dance classes

Emerging Infectious Diseases

How the epidemic got derailed in Antwerp (in Dutch: Hoe the epidemie 919 ontspoorde in Antwerpen)

Delivery of infection from asymptomatic carriers of COVID-19 in a 923 familial cluster

Transmission of 2019-ncov infection from an 925 asymptomatic contact in Germany

Exercise and fitness in the age of social distancing during 927 the COVID-19 Pandemic

Coronavirus disease (COVID-19): The need to maintain regular 930 physical activity while taking precautions

The psychological effects of quarantining a city

Fitnessbranche in Nederland 2018: kerncijfers vraag-en aanbodzijde

Indoor air quality in two university sports facilities

Indoor air quality of environments used for physical exercise and sports 940 practice: systematic review

Human exposure to air contaminants in sports 942 environments

Estimating the inhaled dose of pollutants during 944 indoor physical activity

Indoor school environment: easy and low cost to assess 946 inorganic pollutants

Indoor air quality in primary schools

Characterization of coarse particulate matter in school gym

Exposure to indoor air pollutants during physical 952 activity in fitness centers

Indoor particulate pollution in fitness centres with 954 emphasis on ultrafine particles

Building ventilation: theory and measurements

Air Infiltration and Ventilation Centre

Ventilation of Buildings

Ventilation performance prediction for buildings: A method overview and recent 961 applications

50 years of Computational Wind Engineering: Past, present and future

Building Code. Bouwbesluit Online (in 965 Dutch)

Pandemic COVID-19 and Airborne Transmission. Environmental Health Committee 972 (EHC) Emerging Issue Brief

ASHRAE position document on filtration and air cleaning. American Society of Heating, 976 Refrigerating and Air-Conditioning Engineers

Performance and costs of particle air filtration technologies

Efficiency of a portable indoor air cleaner in removing pollens and 981 fungal spores

The effect of air filtration on airborne dog allergen

Measurement and simulation of the IAQ impact of particle air cleaners 985 in a single-zone building

Method for measuring performance of portable household electric room air cleaners

Approved American National Standard -Association of Home Appliance Manufacturers

Aerosolforschung on understanding the role of aerosol particles in SARS-CoV-2 infection. Association 992 for Aerosol Research

Können mobile Raumluftreiniger eine indirekte SARS-CoV-2 994

Infektionsgefahr durch Aerosole wirk-sam reduzieren?

Ultrafine particle removal and generation by portable air cleaners

Can commonly-used fan-driven air cleaning technologies improve 999 indoor air quality? A literature review

Submicrometer particle removal indoors by a novel electrostatic 1001 precipitator with high clean air delivery rate, low ozone emissions, and carbon fiber ionizer

Portable photocatalytic air cleaners: efficiencies and 1004 by-product generation

Urban aerosol evolution and particle formation 1006 during winter time temperature inversions

Portable aerosol spectrometer, model 11-D. Grimm Aerosol Technik 1008

Product sheet Cloud Garden climate sensor CG.TRHCV.D. Cloud-Garden-1011

Klimaatsensor-Nederlands-incl-technische-specs

Variation of clean air delivery rate and effective air cleaning ratio of room air 1013 cleaning devices

Atlanta: American Society of Heating, Refrigerating and Air-1015 Conditioning Engineerings, Inc

In-flight/onboard monitoring: ACER's 1017 component for ASHEAE 1262, Part 2

Carbon dioxide generation rates for building occupants

A new clean air delivery rate test applied to five 1022 portable indoor air cleaners

Using time-and size-resolved particulate data to quantify 1024 indoor penetration and deposition behavior

Penetration coefficient and deposition rate as a function of 1026 particle size in non-smoking naturally ventilated residences

Indoor particles: a review

What is an effective portable air cleaning device: A review

Effects of room furnishings and air speed on 1031 particle deposition rates indoors

Deposition of tobacco smoke particles in a low ventilation 1033 room

Deposition of suspended particles from turbulent gas streams

Measuring the short-term emission rates of particles in the 1037 "Personal Cloud" with different clothes and activity intensities in a sealed chamber

Can indoor sports centers be allowed to re-open 1042 during the COVID-19 pandemic based on a certificate of equivalence? Build. Environ. 180: Article nr

Calculated asymptotic values of aerosol particle concentrations (µg/m³) for six scenarios: 1, 3', 5

Aerosol particle concentration measurements in a gym with 35 exercising persons. • Assessment of deposition, ventilation and air cleaning (AC) for aerosol reduction

 1089  1090  1091  1092  1093  1094  1095  1096  1097  1098  Table 5 . Flow rates associated with aerosol particle production, deposition, ventilation and air cleaning, for five size fractions. Deposition loss rates and the concentrations in the incoming ventilation air are also given.

2.5-1 µm 1-0.5 µm 0.5-0.25 µm < 0.25 µm 

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:J o u r n a l P r e -p r o o f