Cabin air quality – Quantitative comparison of volatile air contaminants at different flight phases during 177 commercial flights

Cabin air quality – Quantitative comparison of volatile air contaminants at different flight phases during 177 commercial flights

Accepted Manuscript Cabin air quality – Quantitative comparison of volatile air contaminants at different flight phases during 177 commercial flights ...

3MB Sizes 0 Downloads 22 Views

Accepted Manuscript Cabin air quality – Quantitative comparison of volatile air contaminants at different flight phases during 177 commercial flights Sven Schuchardt, Wolfgang Koch, Wolfgang Rosenberger PII:

S0360-1323(18)30723-6

DOI:

https://doi.org/10.1016/j.buildenv.2018.11.028

Reference:

BAE 5821

To appear in:

Building and Environment

Received Date: 28 August 2018 Revised Date:

22 November 2018

Accepted Date: 22 November 2018

Please cite this article as: Schuchardt S, Koch W, Rosenberger W, Cabin air quality – Quantitative comparison of volatile air contaminants at different flight phases during 177 commercial flights, Building and Environment (2018), doi: https://doi.org/10.1016/j.buildenv.2018.11.028. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1 1

Building and Environment

ACCEPTED MANUSCRIPT

Original Research Article

2

Title: Cabin air quality – Quantitative comparison of volatile air contaminants at differ-

4

ent flight phases during 177 commercial flights

RI PT

3

5

Authors & institutions

7

1

Sven Schuchardt, 1Wolfgang Koch, 2Wolfgang Rosenberger

1

Fraunhofer Institute for Toxicology and Environmental Medicine ITEM

8 9

M AN U

SC

6

Nikolai-Fuchs-Str. 1

11

30625 Hannover, Germany

12

Phone +49 511 5350-218

13

Fax +49 511 5350-1552

14

mailto:[email protected]

15

AC C

EP

TE D

10

16

2

17

Institute of Occupational Medicine

18

Carl-Neuberg-Str. 1

19

30625 Hannover, Germany

Hannover Medical School

2

Building and Environment

ACCEPTED MANUSCRIPT

Abstract

21

Quantitative assessment of human exposure to semi-volatile organic compounds (SVOC),

22

such as tricresyl phosphates (TCP) that may originate from engine oil contamination of the

23

cabin air, during air travel is challenging due to the technical complexity of the air supply in

24

commercial jet aircraft. Normal flight operations involve reduced air exchange before and

25

during takeoff, which results in increased concentrations of potential cabin air pollutants.

26

During cruise, normal ventilation rates (> 20 h- 1) are reestablished and thus lower pollutant

27

concentrations are also reestablished. This relationship between changes in ventilation rate

28

and associated changes in pollutant concentrations during the departure phase is first de-

29

scribed in the present study, although this effect was found by previous studies that investi-

30

gated distinct flight phases.

31

The perception of so-called "smell events" in cabin air does not necessarily indicate the

32

presence of harmful contaminants and TCP-containing oil mist must be clearly distinguished

33

thereof. Thus, aldehydes, VOCs, and organophosphates such as TCP were investigated. In

34

this paper, the occurrence of TCP contamination in a bleed air free Boeing 787 (B787) air-

35

craft is reported for the first time. The results presented here show that there are TCP

36

sources other than bleed air from leaky engines. Furthermore, exceptional release behavior

37

of TCP suggests that a more detailed classification for engine oil-triggered cabin air contami-

38

nation (CAC) events is necessary. This study evaluated measurement data from 177 flights

39

that were either commissioned by the EASA or conducted as part of studies with the support

40

of Lufthansa, Condor, and British Airways.

41 42

AC C

EP

TE D

M AN U

SC

RI PT

20

3

Building and Environment

ACCEPTED MANUSCRIPT

43

Highlights

44





47

ment of volatile organic compounds. •

49 50 51

“Smell events” classified as oil leakage with odor perception are mostly identified as false positives.



SC

48

Effect of different ventilation rates during different phases of flight on the measure-

VOC and TCP concentrations during normal flight conditions are considered no threat for human health.

M AN U

46

B787.

RI PT

45

Compilation of cabin air quality data from the most used aircraft types, including the

Keywords

53

Cabin air contamination (CAC), primary technical CAC , engine oil leakage, tricresyl phos-

54

phates, smell event, volatile organic compounds (VOC)

AC C

EP

TE D

52

4

Building and Environment

ACCEPTED MANUSCRIPT

1 Introduction

56

The occurrence of “smell events” associated with oil contamination in cabin air are often

57

alleged to have a negative impact on human health; however, this is controversial. Leaky

58

seals in the engine shafts are considered to be the main source of possible oil contamination

59

in bleed air operated aircraft. Cabin air quality (CAQ) investigations of different flight phases

60

are of particular interest, since it is widely assumed that flight specific maneuvers (e.g.,

61

thrust change during takeoff or ascent) may cause increased oil leakage [1,2].

62

The detection of tricresyl phosphate (TCP) in cabin air is still considered to be an indication

63

of oil contamination, although many other sources are well known. Consequently, isomers of

64

TCP have been the focus of previous and current research [3–5]; however, only the ortho

65

isomers (ToCP), which occur at trace levels (< 0.01%), are regarded as neurotoxic at suffi-

66

ciently high concentration [6–9]. However, most studies report that cabin air quality equals

67

or even exceeds the air quality in normal offices and homes [10–12]. Due to the rare occur-

68

rence of actual smell events, only a few measurements during flight operation have been

69

reported [13]. In addition, the identification of oil-related cabin air contamination (CAC)

70

events is assumed to not be possible by smell alone [1], because most reported in-flight

71

smell events have an origin (flight catering, deicing, lavatories, cleaning) other than oil con-

72

tamination. The analytical detection of TCP or other organophosphorus compounds (OPC),

73

such as tributyl phosphate and triphenyl phosphate, which originate from the hydraulic sys-

74

tem, is often considered as a more reliable indicator for oil contamination, although many

75

OPCs are likely present in cabin air as flame-retardants and plasticizers.

76

The present study aimed to clarify the suitability of TCP as an indicator for oil contamination

77

in aircraft cabins. In addition to TCP/OPC measurements, other parameters such as climate

78

data and concentrations of VOCs, aldehydes, carbon dioxide (CO2), carbon monoxide (CO),

AC C

EP

TE D

M AN U

SC

RI PT

55

5

Building and Environment

ACCEPTED MANUSCRIPT

and ozone (O3) were also measured in the aircraft cabin and cockpit. The wide range of air-

80

craft investigated combined with consideration of different ventilation rates during different

81

flight phases and investigation of several event flights, enabled new insights into the inter-

82

pretation of CAC events.

RI PT

79

83

2 Methods

85

The analytical methods used in the present study are briefly described in the next section

86

and have been described in detail in other studies [1,2].

87

2.1 Investigated flight phases and sampling procedure

88

Table 1 provides a detailed overview of the compounds sampled in the cockpit and cabin air

89

during different flight phases. The sampling units were placed on vacant seats in the cabin

90

and an auxiliary seat in the cockpit; the exact position of the auxiliary seat depended on the

91

type of aircraft being investigated. The exact position of the seats within the investigated

92

compartment did not affect the measured compound distribution, due to the good homoge-

93

nization of the cabin air [14]. Only long-term sampling was performed during cruise, since no

94

“smell events” were reported during this flight phase. The in-flight measurements were per-

95

formed between April 2013 to June 2016 and are comprised of 177 flights with ten different

96

types of aircraft/engine configurations. Nine different types of conventional bleed air sup-

97

plied aircraft and one non-bleed air supplied aircraft (electric compressed air) were investi-

98

gated. During the campaigns from 2013 to 2015, the concept of flight phase sampling was

99

not yet fully established; nevertheless, parts of this data were able to be incorporated in our

100

evaluation by taking this limitation into account. All other CAQ parameters, such as climate

101

data and the concentrations of CO2, CO, and O3, were continuously recorded.

AC C

EP

TE D

M AN U

SC

84

6

Building and Environment

ACCEPTED MANUSCRIPT

2.1.1 General sampling procedure

103

Two identical sets of measurement equipment were installed in the cockpit and cabin. Sam-

104

pling procedures were carried out using constant flow personal air samplers: GSA SG-10, GSA

105

SG-5100 (GSA, Ratingen, Germany), or Gilian 5000 (Sensidyne, St. Petersburg, USA) for or-

106

ganophosphates and GSA SG-350 (GSA, Ratingen, Germany) for aldehydes and VOCs. For

107

OPC analysis, the air flow was adjusted to 10, 3.5, or 1.0 L/min, depending on the sampling

108

period, in order to achieve maximal sensitivity for subsequent analysis. The flow rate of the

109

SG-350 pumps was adjusted to 0.3 (manual adjustment) or 0.333 L/min (pump internal ad-

110

justment) for aldehydes and 0.1 L/min for VOCs. Air flow control (± 5% maximal deviation)

111

was performed before and after sampling using a DC-Lite flow meter (Bios International Cor-

112

poration, Butler, New Jersey, USA 07405). All concentration results were corrected to stand-

113

ard atmospheric conditions (1013 hPa, 293 K). The LODs and further analytical quality data

114

for this sampling procedure have recently been published in detail [1,2,14].

115

Sampling procedures were generally started once the aircraft doors were closed. Long-term

116

sampling (whole flight) and the first short-term sampling (taxi-out) started simultaneously.

117

Upon reaching the runway, the second short-term sampling (takeoff/climb) was started by

118

replacing the short-term sampling tubes. The second short-term sampling was terminated

119

upon reaching cruising altitude or after 40 minutes, whichever occurred later. The third

120

short-term sampling started at the beginning of the descent and was terminated in parallel

121

with the long-term sampling after reaching the parking position. Samples were transported

122

and stored in cooling containers at 4 °C.

AC C

EP

TE D

M AN U

SC

RI PT

102

7

Building and Environment

ACCEPTED MANUSCRIPT

2.1.2 “Smell event” identification

124

For a smell event to be considered in this study, at least two, independent instances of oil-

125

related odor had to be reported by the crew or passengers. Additional VOC-sampling devices

126

(Tenax) were used during these events.

RI PT

123

127

129

2.2 Determination of TCP and organophosphate-based flame retardants and plasticizers

SC

128

We developed a method for cabin air measurements based on ISO 16000-31:2014 "Indoor

131

air -- Part 31: Measurement of flame retardants and plasticizers based on organophosphorus

132

compounds -- Phosphoric acid ester.” A more detailed description has been given in another

133

paper [1]. The filter absorbent combination used in this method allows simultaneous deter-

134

mination of gaseous and particulate OPCs. Sampling was carried out by drawing ambient air

135

through a quartz filter spiked with tributyl phosphate-d27 and triphenyl phosphate-d15 with

136

PUR foam connected downstream using a constant flow sampling pump. Depending on the

137

flight phase of sampling, the air flow was adjusted to 1, 3.5, or 10 L/min. After sampling had

138

been completed, the adsorbed OPCs were concentrated by Soxhlet extraction with di-

139

chloromethane. Toluene was added as a keeper and the sample was then evaporated to a

140

final sample volume of approximately 100 µL using a rotary evaporator and a nitrogen evap-

141

orator. The OPCs were determined by gas chromatography-mass spectrometry (GC-MS) and

142

quantified using external calibration.

143

2.3 Determination of VOCs and aldehydes in air using Tenax TATM tubes

144

Sampling was carried out by drawing ambient air through two Tenax TATM tubes using a

145

mass-flow-regulated pump to obtain an airflow of 0.1 L/min in each tube. The VOC analysis

AC C

EP

TE D

M AN U

130

8

Building and Environment

ACCEPTED MANUSCRIPT

was conducted according to the international standard methods for measuring organic com-

147

pounds in indoor air (ISO 16000-6:2011 “Indoor air -- Part 6: Determination of volatile organ-

148

ic compounds in indoor and test chamber air by active sampling on Tenax TATM sorbent,

149

thermal desorption and gas chromatography using MS or MS-FID”). The ISO 16000-6 method

150

includes a wide range of alkanes, aromatics, terpenes, aldehydes, etc., which are also known

151

to be representative of cabin air quality. Hitherto unknown VOCs were identified with a non-

152

targeted approach using the MS spectral libraries within the NIST database and subsequent

153

standard verification. When no standards for individual calibration were available, the quan-

154

tification was calculated as toluene equivalents.

155

2.4 Determination of aldehydes using DNPH cartridges and HPLC-UV absorption

156

This method was based on ISO 16000-3:2011 “Indoor air -- Part 3: Determination of formal-

157

dehyde and other carbonyl compounds in indoor air and test chamber air -- Active sampling

158

method.” Aldehydes were collected by drawing air through a cartridge containing a silica gel

159

substrate coated with DNPH (2,4-dinitrophenylhydrazine) reagent. Mass-flow-regulated

160

pumps were used to draw air and their flow rate was set to 0.3-0.333 L/min. Alde-

161

hyde/ketone hydrazones were eluted from DNPH cartridges with approximately 3.5 mL ace-

162

tonitrile, according to the manufacturer’s recommendations. The aldehyde hydrazones were

163

separated, identified, and quantified by HPLC with UV absorbance detection. Analyses were

164

performed using an Agilent 1100 HPLC System equipped with a UV/VIS detector set to 360

165

nm [2].

166

AC C

EP

TE D

M AN U

SC

RI PT

146

9

Building and Environment

ACCEPTED MANUSCRIPT

3 Results and discussion

168

3.1 VOCs and aldehydes (ALDs)

169

The mean VOC concentrations determined according to ISO 16000-6 (including middle-chain

170

aldehydes from hexanal to decanal) were 122 µg/m³ to 370 µg/m³ and the 95th percentiles

171

ranged from 188 – 887 µg/m³. In most cases, ethanol, n- and isopropanol, 1,2-propanediol,

172

acetic acid, benzoic acid, and acetonitrile were the main VOCs observed. The sources of

173

these emissions have been identified with great certainty: the monoalcohols are from fre-

174

quently used refreshment wipes, the propylene glycol from deicing procedures, the acetic

175

acid from services or similar activities, the benzoic acid from Tenax decomposition, and the

176

acetonitrile from the DNPH-cartridges used for aldehyde measurements. In addition, oxida-

177

tion products of VOC caused by the reaction with ozone cannot be fully excluded in cabin air

178

[15]. For most of the approximately 100 individual compounds that were quantified, the

179

concentrations were in the lower µg/m³ range. Similar results were obtained for the sum of

180

aldehydes, determined by the DNPH method (ISO 16000-3:2011). The mean concentrations

181

of short-chain aldehydes (formaldehyde, acetaldehyde, acrolein, propionaldehyde, cro-

182

tonaldehyde, butyraldehyde, benzaldehyde, isovaleraldehyde, valeraldehyde, o-, m- and p-

183

tolualdehyde, hexanal, and 2,5-dimethyl benzaldehyde) were 13 to 35 µg/m³ and the 95th

184

percentiles ranged from 15 to 70 µg/m³ (Table 6). The unsaturated aldehydes, acrolein and

185

crotonaldehyde, are difficult to quantify because air sampling on DNPH-coated silica gel car-

186

tridges is associated with formation of by-products, which add uncertainty to the results.

187

The formaldehyde measurements in this study did not reach any guideline values for indoor

188

air.

AC C

EP

TE D

M AN U

SC

RI PT

167

10

Building and Environment

ACCEPTED MANUSCRIPT

The measured cabin air concentrations of VOCs and aldehydes were generally comparable to

190

or often even lower than concentrations measured in common indoor environments (e.g.,

191

offices, kindergartens, private homes) [16–19] and are in good agreement with other aircraft

192

investigations [2,20–24]. Table 2 gives a comparative overview of the VOC classes present in

193

aircraft [14] versus public indoor environments [17,18]. Based on this comparison, the con-

194

centrations of VOCs and aldehydes in aircraft are very similar to those in public indoor envi-

195

ronments and should provide no cause for concern for human health [25]. This assessment is

196

additionally based on reference values provided by the German Committee on Indoor Guide

197

Values of the German Federal Environmental Agency (UBA), which are published as indoor

198

air guide values for individual substances or for compound classes (e.g., alkylbenzenes,

199

cresols, and terpenes [26–28]). The guide values (RW I and II) also take particularly sensitive

200

groups of people into account, such as children and the elderly, as well as long-term expo-

201

sure indoors (24-hour/day). Exposure to concentrations below the guide value (RW I), which

202

is the case for all concentrations measured here, is considered to be of no concern to health,

203

even during life-long exposure. In cases where guide value (RW II) is exceeded, appropriate

204

measures to reduce exposure must be taken immediately.

205

Differences observed between aircraft and indoor environments can be attributed to special

206

technical features in aircraft (e.g., electrical wiring, pneumatics) and to in-flight services (e.g.,

207

drinks and food service, cleaning) or external treatments (e.g., deicing). This is true for com-

208

pounds that are present at low concentrations, such as naphthalene (insect repellant), per-

209

fluorinated compounds (cooling liquid), isopropanol (disinfectant), and propylene glycol (de-

210

icing). Direct comparison of bleed air and non-bleed air aircraft (B787) reveals lower mean

211

concentrations of VOCs/aldehydes in the B787. This may be due to the use of combined

212

HEPA/charcoal filters in the B787, which enable the removal of volatile contaminants to a

AC C

EP

TE D

M AN U

SC

RI PT

189

11

Building and Environment

ACCEPTED MANUSCRIPT

limited extent. A similar reduction in volatile contaminants has recently been demonstrated

214

for Airbus 321 (A321) aircraft that use carbon filters [1]. The DNPH method was also used for

215

aldehyde sampling in all flights and confirmed the findings obtained with the Tenax method

216

(ISO 16000-6:2011) (Table 6).

217

More remarkable is a general trend of decreasing contaminant concentrations across the

218

investigated flight phases (see pictograms in Table 2). The contaminant concentrations dur-

219

ing taxi-out and takeoff are high because of reduced air exchange rates during these flight

220

phases. The significant reductions in contaminant concentrations observed during the de-

221

scent and landing phases (unpaired t-test: p values < 0.05, see Table 3) lead to smaller con-

222

centrations that are more difficult to interpret. The contaminant concentrations cannot be

223

connected to the air exchange in this case and thus are assumed be the result of interactions

224

between mechanisms that are currently unknown (e.g., gradual decrease of source

225

strength). Due to the change in ventilation during taxi-out, the pollutant concentrations ap-

226

pear to be highest during taxi-out for all of the investigated flights, including the B787. Non-

227

continuous or intermittent emissions sources such as in-flight services, human activities, and

228

aforementioned smell events interfere with this observed trend of concentration decrease;

229

thus, the differences in concentration between the investigated flight phases are not statisti-

230

cal relevant for all compound classes (Table 3). In the case of aldehydes, the differences in

231

concentration are particularly distinctive and it is assumed that the dominant aldehydes –

232

formaldehyde, acetaldehyde, as well as nonanal and decanal – do not show continuous

233

emissions behavior. The implications of this observation are discussed in Section 3.4.

234

3.2 Organophosphorus compounds (OPCs) including tricresyl phosphate (TCP)

235

Most OPCs have lower vapor pressures than the VOCs in cabin air and thus should have low-

236

er concentrations than VOCs in cabin air. This was confirmed by the measurements in this

AC C

EP

TE D

M AN U

SC

RI PT

213

12

Building and Environment

ACCEPTED MANUSCRIPT

study since the sum of OPC concentrations is approximately 100 times lower than the sum of

238

VOC concentrations (Table 6). Overall, OPC mean concentrations were in the range of 0.74

239

to 5.36 µg/m³ and the 95th percentiles ranged from 1.12 to 7.45 µg/m³, respectively. Tri-n-

240

butyl phosphate (TBP) was the most prominent OPC, which follows from its relatively high

241

vapor pressure. Tris(chloroisopropyl) phosphate, which is not attributed to engine oil or hy-

242

draulic fluids, exhibited exceptionally high values unique to the cockpit of the A321 aircraft

243

(data not shown), which indicates a specific source of emissions in this location. Only a small

244

OPC fraction (overall mean = 2-932 ng/m3; 95th percentile = 6 - 1.670 ng/m³) can be attribut-

245

ed to TCP, which is presumed to be an indicator of bleed air engine oil contamination. The

246

concentrations of TCP are in agreement with values reported by previous investigations

247

[4,29]. Organophosphorus compounds from other sources (e.g. hydraulic oil (TBP), fire re-

248

tardant, plasticizers, etc.) are common in aircraft environments, which may explain their

249

occurrence in the bleed air free B787 [30].

250

The wide distribution of TCP calls into question its viability as a unique marker for engine oil

251

contamination. No ortho isomers of TCP were detected in any of the measurements. In fact,

252

the observed OPC distribution (Table 4) and the sum of TCP m/p-isomers (26 ng/m3) in the

253

B787 are comparable to aircraft with bleed air (7-58 ng/m3), which again suggesting that

254

other sources of TCP are more likely than engine oil. For example, Yadav et al. were able to

255

detect comparable concentrations of TCP in residential buildings in Nepal [31] and studies

256

that have determined TCP contamination levels in household dust reported similar, or high-

257

er, concentrations of TCP in home environments in the US, Canada, China, and several Euro-

258

pean countries [24,32–37]. Thus, minor concentrations of TCP can be considered as normal

259

background contamination in aircraft cabins and common homes. The indoor concentrations

260

of other OPCs are usually higher than TCP because of their high usage and production. Due

AC C

EP

TE D

M AN U

SC

RI PT

237

13

Building and Environment

ACCEPTED MANUSCRIPT

to the ubiquity of OPCs, it is possible to carry out biomonitoring in public environments and

262

compare the results to the OPC concentrations in cabin air. The results of two studies that

263

used the same number of volunteers to investigate OPC metabolites in urine are given in

264

Table 5. The first study, conducted by Schindler et al. [38], analyzed urine samples from ex-

265

posed flight personnel after smell or fume incidents were reported. None of the 332 urine

266

samples contained o-cresyl-containing metabolites and meta and para TCP metabolites were

267

only detected in one sample, wherein the concentration was close to the limit of detection.

268

The second study from Fromme et al. [39] investigated 312 urine samples from children in

269

German day care centers and did not report significant differences in the OPC distribution

270

compared to Schindler et al. [37]. Table 5 compares the urine concentrations of tributyl

271

phosphate and triphosphyl phosphate from both environments, which are known to be ma-

272

jor constituents of hydraulic fluids, as well as tris(chloroisopropyl) phosphate (TCPP), which

273

is more common in indoor environments. The concentrations of metabolites from the hy-

274

draulic fluid constituents are found to be slightly higher in aircraft personnel, although the

275

increase is still within acceptable exposure limits (Table 5). The ubiquitous distribution of

276

TCP reported in common indoor environments worldwide and the available biomonitoring

277

results from exposed flight personnel may raise doubts about the health effects that are fre-

278

quently claimed to be caused by oil-triggered “smell events” in aircraft. However, it is not yet

279

possible to clearly distinguish TCP emissions from engine oil and other sources (e.g. by dif-

280

ferentiation of the specific isomer composition). Since TCP concentrations in identically fur-

281

nished B787 aircraft vary, an external source of TCP from oil contamination cannot be com-

282

pletely excluded.

283

According to Sagunski et al. [40], the guide value concept (RW I and II, discussed in the pre-

284

vious section) could be applied to the sum of all OPCs. Thus, nearly all of the mean values

AC C

EP

TE D

M AN U

SC

RI PT

261

14

Building and Environment

ACCEPTED MANUSCRIPT

and 95th percentiles of OPC concentrations recorded in this study are well below the guide

286

value (RW I = 5 μg/m³). The intervention guide value (RW II = 50 μg/m³) was not reached in

287

any situation.

288

It was observed that all OPCs, including TCP, exhibited the predicted decrease in concentra-

289

tion caused by the different ventilation rates during different phases of flight (Figure 1A,

290

Table 3); based on this, an internal source of contamination is more plausible. To our

291

knowledge, this is the first report investigating the presence of TCP in bleed air free B787

292

cabin air.

293 294 295

M AN U

SC

RI PT

285

3.3 Flight phase-dependent decrease in contaminant concentration caused by different ventilation rates

The observed decrease in contaminant concentration across the investigated flight phases

297

requires an explanation since it has also been observed in other cabin air studies. Figure 1A is

298

a pictogram comparing the concentrations of selected OPCs and VOCs detected during dis-

299

tinct flight phases with those from other studies. All studies reported comparable concentra-

300

tions and patterns for all detected compounds. Figure 1B illustrates how the flight phase

301

pictograms for toluene, used as an example, were generated using data from the Cranfield

302

study [21].

303

The general validity of the observed effect of different ventilation rates during the different

304

flight phases of aircraft was finally proven by evaluating the continuously monitored CO2

305

data [14]. Like the organic compounds, the same trend of concentration decrease across

306

flight phases was observed for the mean concentrations of CO2 (Figure 2). At constant CO2

307

exhalation (S = source strength; µg/s) and a constant air exchange rate (Λ = air exchange

AC C

EP

TE D

296

15

Building and Environment

ACCEPTED MANUSCRIPT

308

rate; 1/h,), an equilibrium CO2 concentration is established in any given aircraft cabin (V =

309

cabin volume; m3). The rate of change of the CO2 concentration at any time can be calculated

310

according of Equation 1:



At equilibrium, the concentration of CO2 (

=

313

+



Eq. 1

RI PT

312

= −



311

) is given by Equation 2:

Eq. 2



In order to determine the equilibrium concentration of CO2, the current ambient CO2 con-

315

centration (

316

erate an average of 0.44 g CO2 per minute (S), as given by Spengler et al. [41]. This approach

317

neglects minor fluctuations caused by CO2 generated from intermittent activities. If the air

318

exchange rate and number of passengers are based on a fully occupied cabin, the calculated

319

outcome is universally applicable for most aircraft types. A specific example of such an air-

320

craft simulation is given in the EASA study [14]. In any case, the calculation has a sharp drop

321

but exhibits a significant deviation from the high CO2 concentrations measured during the

322

taxi phase (Figure 2), which is normally attributed to the increased activity of occupants dur-

323

ing the boarding process. In fact, while the observable fluctuations in the CO2 curves can be

324

attributed to changes in the occupants' activities, the massive drop during takeoff certainly

325

cannot. The shape of the measured CO2 curve is congruent with the calculated drop in CO2

326

and thus can be attributed to the reduced air exchange during the taxi-out and take-

327

off/climb phases. According to the mass flow equations given above, any increase in air ex-

328

change (Λ>20) quickly results in lower equilibrium concentrations for all airborne com-

329

pounds, including particles and VOCs. Our assumptions about the changes in CO2 concentra-

SC

314

AC C

EP

TE D

M AN U

) was set to 388 ppm (710 mg/m³) and each passenger was assumed to gen-

16

Building and Environment

ACCEPTED MANUSCRIPT

tion were recently confirmed by investigations of 179 US domestic flights by Cao et al. [42]

331

and by information obtained from flight engineers.

332

While high carbon dioxide concentrations were observed and attributed to reduced air ex-

333

change effect during taxi-out and takeoff by much earlier studies [41,43–46], no previous

334

study has correctly interpreted the frequently measured decrease in airborne contaminant

335

concentrations [2,20–23]. Similar misinterpretations of particle distributions in cabin air

336

were recently published [47]. This study observed the effects of different ventilation rates

337

during different phases of flight on airborne contaminants, including the possibly particle-

338

bound TCP, and highlights the need for future evaluations of cabin air contaminants to con-

339

sider the flight phase-dependence of the ventilation rates. Based on our results, the basic

340

physical parameters of reduced air exchange rates during taxi-out and takeoff are also im-

341

plemented in B787 aircraft (Figure 1A, Table 3).

342

3.4 Detection of CAC events as deviations to the concentration decrease pattern

343

Intermittent emissions, such as short term contaminant sources or significant variations in

344

the strength of emissions sources, may reveal a different pattern than simple concentration

345

decrease. However, such intermittent contaminant emissions may not be easily detected

346

during the taxi-out phase. Figure 3 demonstrates that deviations to the pattern were ob-

347

served for several contaminants in cabin air. Most of the deviations can be attributed to in-

348

flight activities, such as food preparation or human activity, and are comparable to other

349

common indoor situations. Nearly 30% of the VOC concentrations in aircraft cabins can be

350

attributed to services and humans [24,48]. A particularly interesting example is the promi-

351

nent combination of nonanal and decanal in perfumes and correspondingly strong correla-

352

tion between them in cabin air [49]. This correlation lead Wang et al. to hypothesize that

AC C

EP

TE D

M AN U

SC

RI PT

330

17

Building and Environment

ACCEPTED MANUSCRIPT

their sources are related, which is confirmed by our investigation, as shown by the corre-

354

sponding pictograms in Figure 3A.

355

In addition, several of the deviations to the pattern that were observed might also originate

356

outside of the aircraft cabin. A prominent example is the isoalkane fraction (C14-C20), which

357

is often detected during takeoff and has not been fully characterized (Figure 3B). To our

358

knowledge, this is the first report of this contaminant, which seems to be a widely used lub-

359

ricant in aviation and is chemically different from the synthetic pentaerythritol or trime-

360

thylopropane esters used in engine oil. Even if the observed concentration of this lubricant

361

fraction (max. flight phase mean was 47 µg/m3) is below the critical toxicological threshold,

362

further investigations might be of interest.

363

3.5 Detection of oil-triggered CAC events using TCP as a possible indicator

364

The most striking deviations from the pattern related to different ventilation rates during

365

different flight phases are those of TCP. Most cabin air quality studies aim to detect these

366

technical cabin air contaminations (TCAC), which are attributed to oil entry from leaking en-

367

gine seals in individual flight phases. Theoretically, the occurrence of elevated TCP concen-

368

trations caused by engine leaking during the takeoff/climb and descent/landing phases can

369

be easily distinguished using the flight phase sampling strategy used in this study. In the cab-

370

in, the background level of TCP, which may originate from flame retardants, plasticizers, or

371

hitherto unknown sources, should exhibit the expected pattern related to the different ven-

372

tilation rates during different phases of flight (Figure 1, Table 6). Since primary TCAC events,

373

which are associated with aerosol formation, are rare and have a short duration, it is consid-

374

ered to be unlikely that such an event will occur during in-flight sampling [14,50–53].

AC C

EP

TE D

M AN U

SC

RI PT

353

18

Building and Environment

ACCEPTED MANUSCRIPT

Normally, oil contamination caused by a leaky engine (primary TCAC) should be detectable in

376

both measuring locations: cockpit and cabin. However, the A340-6 aircraft data given in Fig-

377

ure 3C indicate that oil leaks detected by increased TCP concentrations can occur in either

378

the cockpit or the cabin, which contradicts the above assumption of equally distributed con-

379

tamination. Therefore, we postulate that there may be another source of TCP contamination

380

that affects different parts of the aircraft. These contamination sources may include the slow

381

release of oil compounds deposited in the bleed air system, the environment control system

382

(ECS), and the ducts and under hitherto undefined circumstances, these oil reservoirs may

383

serve as secondary sources for oil/TCP contamination. High-boiling point compounds may be

384

deposited on surfaces when local oil vapor concentrations are high enough to favor conden-

385

sation by homogenous nucleation [14,54]; therefore, as the particles grow in size, they may

386

be deposited onto the surfaces of the highly branched ventilation system. The effects of such

387

deposition have already been observed on engine test stands, while investigating oil con-

388

tamination on HEPA filters [55]. The removal efficiency paired with this deposition mecha-

389

nism depends strongly on the particle size distribution and requires further investigation.

390

Material accumulated by deposition can be released at any time by desorption, which is con-

391

trolled by air temperature, air composition, humidity, vibrations, and airflow. We hypothe-

392

size the existence of so-called secondary TCAC events that may be caused by oil/TCP release

393

from deposits within the bleed air system, ECS, and ducts without detectable engine abnor-

394

malities [14]. Many observations and reports support an indirect contamination scenario. Oil

395

and dust deposits (black smear) that have been observed in aircraft ducts form the basis of

396

this hypothesis [47,54] and the existence of locally restricted TCAC events are further sup-

397

ported by the frequently reported local occurrences of smell events in either the cockpit or

398

cabin. Even the unusual TCP release in A340-6 aircraft discussed above is satisfied this local-

AC C

EP

TE D

M AN U

SC

RI PT

375

19

Building and Environment

ACCEPTED MANUSCRIPT

event concept. It should be noted that cabin air measurements cannot distinguish between

400

primary and secondary TCAC events and the absorption capacity and/or the reemission be-

401

havior of deposited layers in transfer ducts, composed of TCP and odorous substances and

402

considered to be the cause of secondary TCAC events, require further research.

403

3.6 Analytical differentiation between non-event, smell event, and TCP-event flights

404

Smell events are mostly undefined incidents and frequently the source of the smell cannot

405

be identified. Therefore, the link to bleed air contamination seems to be arbitrary in most

406

reported cases. Normally, the smell of “old socks” or “wet dog” is associated with oil con-

407

tamination, although the olfactory evidence has yet to be provided [1]. Most reported smell

408

events in aircraft are harmless incidents (e.g., food, cosmetics, disinfection, dirt, air pollution

409

at the airport, etc.) and primary TCAC events are only assumed to occur during a few hun-

410

dred takeoffs out of each million [51]. However, such rare oil leakage events are of particular

411

interest since engine oil contains up to 3% odorless TCP and traces (if any) of ToCP, which is

412

neurotoxic at high doses [5,24,56,57]. Nevertheless, detection of suspected oil contamina-

413

tion has been possible only by the smell of additional hitherto unknown odorous com-

414

pounds.

415

Eighteen cases of increased TCP concentrations, which is considered as indication for oil

416

leakage, were detected during the 177 flights in this study. None of these 18 events were

417

accompanied by complaints about smells in the aircraft cabin. Conversely, 17 flights with

418

reported smell events (alleged oil smell) were unable to be analytically distinguished from

419

the 142 non-event flights. A compilation of VOC, aldehyde, OPC, and TCP mean concentra-

420

tions are given in Table 6 along with the corresponding flight phase pictograms. The non-

421

event flights, the smell event flights, and the potential TCAC flights have been arranged from

422

top to bottom within the table. The 142 non-event flights also include eight flights of the

AC C

EP

TE D

M AN U

SC

RI PT

399

20

Building and Environment

ACCEPTED MANUSCRIPT

bleed air free B787 aircraft. The contaminant concentrations and distributions in the bleed

424

air free B787 are comparable to those in aircraft with bleed air supply, which is potentially

425

due to similar cabin furniture and ventilation conditions in both types of aircraft. The low

426

mean VOC concentration in the B787 (123 µg/m3) was attributed to the use of activated car-

427

bon filters (see section 3.1 and [1]) and the low mean concentration observed for TCP

428

(26 ng/m3) is believed to be from sources other than engine oil. Moreover, the pattern from

429

the effects of different ventilation rates during different flight phases, as illustrated by the

430

pictograms, is observed for almost all non-event flights, except for the TCP pictogram of the

431

B767 aircraft. The deviation in the descent and landing phases was caused by two B767 air-

432

craft (data not shown) and is considered to be insignificant because of the low magnitude of

433

the TCP concentrations (9 ng/m3). It should also be noted that the A380 study (LH-1) only

434

sampled two flight phases, in contrast to the set of three phases used in the EASA study [14];

435

however, the underlying pattern from the effects of different ventilation rates during differ-

436

ent phases of flight are still perceptible in the corresponding pictograms.

437

During the 17 flights with a reported smell event, the total concentrations of airborne con-

438

taminant remain mostly unchanged (see “Mean values of selected flights” in Table 6). This is

439

in good agreement with the biomonitoring results for OPC metabolites that were previously

440

discussed (Table 5) and consequently implies that odor is not detectable by the analytical

441

methods applied.

442

The situation is somewhat different when it comes to the analysis of used HEPA filters from

443

aircraft. Eckels et al. [55] compared HEPA filters after flights with putative smell events (77

444

filters) and normal flights (107 filters). The results from both groups were comparable, alt-

445

hough outliers with significantly increased concentrations of TCP were primarily related with

AC C

EP

TE D

M AN U

SC

RI PT

423

21

Building and Environment

ACCEPTED MANUSCRIPT

the filters from smell events. However, it can be concluded that uncertain event classifica-

447

tion by humans makes the interpretation of any “smell event” data challenging.

448

With the exception of the single A321 flight (LH-2 study), most of the air contaminant data

449

pictograms exhibit the pattern from the effects of different ventilation rates during different

450

phases of flight. Only the TCP concentration of the A321 flight (Table 6, first line in the smell

451

event block) has an unusual maximum in the descent/landing phase and the smell was re-

452

ported during the takeoff/climb phase. Even if the putative TCAC event is taken into consid-

453

eration (grey dot), this classification remains questionable since the mean TCP concentration

454

of 26 ng/m3 is the same as that of the bleed air free B787 aircraft. The uncertainty of smell

455

events becomes more pronounced when the 18 flights classified as putative TCAC events are

456

compared with this single, unusual A321 flight. The mean TCP concentration of the 95th per-

457

centiles from TCAC event flights was 750 ng/m3 (values in parentheses in Table 6), which is

458

significantly elevated compared with the 142 non-event flights (43 ng/m3) and 17 smell

459

event flights (41 ng/m3). Nevertheless, the set of smell event flights does not show the typi-

460

cal pattern in the flight phase pictograms, even though the mean values of the selected

461

flights are nearly unchanged compared with the non-event flights. The cause of this incon-

462

sistency remains unclear, but we currently assume that the total number of putative smell

463

event flights is too low for the typical pattern to be visible in the pictograms.

464

During the 18 classified TCAC flights, deviations from the pattern were less significant when

465

higher concentrations of OPCs were measured. With 95th percentile concentrations above

466

100 ng TCP/m3 during individual flight phases, the pictograms clearly display deviations from

467

the pattern of different ventilation rates during different flight phases, whereas the concen-

468

trations of VOCs and aldehydes are nearly unchanged (Table 6, Mean values of the selected

469

flights). This holds true even for increased TCP concentrations in the takeoff phase, as

AC C

EP

TE D

M AN U

SC

RI PT

446

22

Building and Environment

ACCEPTED MANUSCRIPT

demonstrated by two A321 flights (1670 and 558 ng TCP/m3) and one exceptional A380

471

flight (859 ng TCP/m3). Due to these elevated TCP concentrations, the pattern from different

472

ventilation rates during different phases of flight is clear, which can be interpreted as an in-

473

dication of a possible oil contamination during the taxi-out phase.

474

The TCP-related oil contamination hypothesis is supported by an additional observation. The

475

mean concentration of TBP (hydraulic oil) is elevated by a factor of two for the 18 TCAC

476

event flights, which is notable, although the mean 95th percentile concentration is the same

477

as that of all 177 investigated flights (1.6 µg/m3). There might be a correlation between TBP

478

and TCP if they are released in parallel. The measured maximum 95th percentile TBP values

479

for A320 (1.74 µg/m3) and A321 (5.83 µg/m3) aircraft appear to be significantly elevated,

480

which could be an indirect hint to the known physical connection between the hydraulic and

481

the bleed air system in the A320 series.

482

4 Conclusions

483

This study has demonstrated that quantitative interpretation of airborne contaminant data

484

from cabin air is particularly difficult because of the discontinuous ventilation conditions that

485

are normally applied (air exchange rates > 20 h-1). The quantitative comparison of different

486

flight phases highlights the need for a more careful evaluation, since the increase in air ex-

487

change during cruise causes a frequently misinterpreted reduction in contaminant concen-

488

trations. Furthermore, the uniquely high number of measurement flights presented here

489

(177 flights) made it possible to analytically compare non-event flights, smell event flights,

490

and TCAC event flights. In summary, the following conclusions can be drawn:

491

1) In-flight cabin air measurements during different flight phases need to consider the ef-

492

fects of different ventilation rates during different phases of flight, which explain why the

AC C

EP

TE D

M AN U

SC

RI PT

470

23

Building and Environment

ACCEPTED MANUSCRIPT

mean concentrations of airborne substances observed during cruise operation are lower

494

than those that are normally observed.

495

2) Tricresyl phosphate is still considered to be a useful indicator of engine oil contamination,

496

although a low background concentration was shown to be ubiquitous in all aircraft types.

497

The low concentration of TCP detected in the bleed air free B787 (95th percentile

498

< 100 ng/m3), reported here for the first time, indicates the TCP background contamination

499

that is commonly present in aircraft.

500

3) We recommend the use of primary and secondary TCAC event classifications in order to

501

address possible oil/TCP release from deposits in the bleed air system.

502

4) The analytical data presented herein suggest that most of the reported smell events can-

503

not be detected with commonly used analytical methods (including biomonitoring). More

504

frequently observed smell events that do not originate from bleed air should not be equated

505

with much rarer and better called “primary TCAC events”, which are caused by engine oil

506

contaminations. Because of the large number of odor sources in aircraft cabins, a meaningful

507

detection of oil or its thermal degradation products by olfactory sensing alone is not possi-

508

ble.

509

Furthermore, the presently available data on cabin air contaminants allow for remarks on a

510

risk assessment for human health. The TCP concentrations (para and meta isomers only)

511

detected on all investigated flights were well below the internationally established toxicolog-

512

ical thresholds for harm to human health. The maximum concentrations of TCP detected in

513

this study were less than 2 µg/m3 for the reported single events and less than 0.05 µg/m3 for

514

non-event flights, which is far below the occupational exposure limit (OEL) of 100 µg/m3 [58]

515

and the threshold limit value (TLV) of 20 µg/m3, which was most recently derived for the

AC C

EP

TE D

M AN U

SC

RI PT

493

24

Building and Environment

ACCEPTED MANUSCRIPT

more toxic ToCP by the American Conference of Governmental Industrial Hygienists (ACGIH,

517

2016). Furthermore, OEL values are considered safe for workers with 8-hour shifts. Neuro-

518

toxic ToCP as well as all other mono or di-ortho-isomers of TCP have not been detected in

519

any studies so far [1,4,14,20,38]. Even high TCP concentrations caused by major oil leakage

520

(primary TCAC event) are not considered a threat to human health because of the limited oil

521

amounts and exposure times reported [4,14,24,38,59,60]. Thus, the neurological impair-

522

ments that have occasionally been reported to be the result of smell events, as well as the

523

diffuse phenomenon of so-called “aerotoxic syndrome,” are not supported by the data pre-

524

sented in this study nor by the data in the literature [24].

525

The available data presented in this study is already considered conclusive; however, the

526

validity of the conclusions drawn here should be confirmed by further documentation of

527

case examples (e.g., dedicated measurements to identify the specific odorous

528

VOCs/particles).

SC

M AN U

TE D EP AC C

529

RI PT

516

25 530

Building and Environment

ACCEPTED MANUSCRIPT

Tables

531

RI PT

Table 1 Depiction of five distinguishable flight phases. Measured substances (VOC = volatile organic compounds; ALD = aldehydes; OPC = organophosphorus compounds) and flight phases investigated (grey background) in the evaluated studies. Note that the Lufthansa (LH) campaigns do not completely cover the flight phases established in the EASA study [14]. In campaign LH-1, taxi-out and takeoff/climb were combined for all measurement, whereas in campaign LH-2, the measurement of VOCs was identical to that in the EASA campaign. TVOC (total VOC), CO2, CO, O3, pressure (P), and temperature (T) were continuously monitored during all measurement flights (data not shown).

Campaign duration 2013-2015 2014-2015

EASA/69*

2015-2016

Taxi-out

Takeoff and climb

Cruise

VOC, OPC, ALD VOC** VOC VOC VOC* OPC, ALD VOC, OPC, VOC, VOC* ALD OPC, ALD

TE D

Campaign/ no. of flights LH-1/64 LH-2/44

M AN U

SC

532 533 534 535 536 537 538

Descent and landing*** VOC, OPC, ALD VOC, OPC, ALD VOC, OPC, ALD

540 541

AC C

539

EP

*including eight B787 flights; **optional VOC sampling in case of reported CAC events; ***combined decent and landing (4) and taxi-in (5) phase; EASA = preliminary cabin air quality (CAQ) campaign supported by Lufthansa, Condor and British Airways [14]; LH-1 and LH-2 = unpublished campaigns on A380 (LH-1) and A321 (LH-2) aircraft conducted by Lufthansa (LH).

26

Study 1A Compound Class Alkanes

3

Study 1B 3

[µg/m ]

[µg/m ]

Study 2 3

[µg/m ]

23 (60)

10 (73)

43 (392)

9 (23)

Alcohols

261 (2509)

33 (371)

28 (270)

77 (301)

Terpenes

83 (1412)

17 (124)

23 (351)

10 (22)

Esters

24 (182)

17 (187)

25 (411)

5 (13)

12 (38)

Aldehydes

49 (295)

3 (31)

106 (398)

11 (25)

27 (70)

475 (4739)

91 (982)

263 (2130)

133 (446)

273 (949)

560 561 562

TE D EP

559

AC C

558

[µg/m ]

21 (62)

553

557

3

T TO/C D/L

38 (308)

552

556

[µg/m ]

11 (196)

Sum

555

Bleed Air

3

35 (278)

20 (61) 20 (68)

SC

Aromatics

554

B787

RI PT

Table 2 Sum of 50th and (95th) percentiles from three representative indoor air studies (Study 1A, 1B, 2) compared to mean cabin air measurements from eight B787 flights and 61 bleed air supplied aircraft [14]. The pictograms on the right of the flight study columns (B787 and Bleed Air) show the relative distribution of the compound classes during the investigated flight phases, illustrating the effect of different ventilation rates during different phases (T = taxi-out; TO/C = takeoff and climb; D/L = descent and landing). Relative concentrations in the pictograms are given in a greyscale (high = dark grey, medium = grey, low = light grey). Study 1A = mean values of 49 indoor air measurements in new buildings and Study 1B = mean values of 285 indoor air measurements in school buildings and kindergartens [18]; Study 2 = mean values of 2,663 indoor air measurements in representative homes [17].

M AN U

542 543 544 545 546 547 548 549 550 551

Building and Environment

ACCEPTED MANUSCRIPT

169 (639) 25 (73)

T TO/C D/L

27

Table 3 Calculated p-values (unpaired t-test) for the observed trend of reduction in contaminant concentration between the investigated flight phases (T = taxi-out; TO/C = takeoff and climb; D/L = descent and landing) for bleed air (bleed) and B787 aircraft. The p-values that indicate the changes in concentration are not statistically relevant are given in bold.

Flight Phase

T → TO/C

TO/C →

T → D/L

p < 0.0001 p < 0.0001

p < 0.0001 p = 0.1111

p < 0.0001 p < 0.0001

p < 0.0001 p < 0.0001

p < 0.5926 p = 0.8652

p < 0.0001 p = 0.0005

p = 0.0087 p = 0.0609

p = 0.0209 p = 0.3255

p < 0.0001 p = 0.0122

p = 0.0295 p = 0.0627

p = 0.0008 p = 0.1374

p < 0.0001 p = 0.0018

p < 0.0001 p = 0.1432

p = 0.6451 p = 0.0376

567

p < 0.0001 p = 0.0122

AC C

EP

TE D

568

SC

VOC bleed B787 ALD bleed B787 OPC bleed B787 TnBP bleed B787 TCP bleed B787

RI PT

D/L

M AN U

563 564 565 566

Building and Environment

ACCEPTED MANUSCRIPT

28

Building and Environment

ACCEPTED MANUSCRIPT

569 Table 4 Mean measured airborne OPC concentrations (µg/m³) of all flight phases in the types of aircraft investigated by the involved studies (LH-1, LH-2, and EASA). Type of aircraft Samples n

A320

A321

A340

A380

B747

B767

B787

92

207

64

196

79

147

55

Concentration

µg/m³ 0.056

0.045

0.029

0.065

0.152

0.168

0.016

Tributyl phosphate (TnBP)

0.824

0.792

0.413

0.913

0.129

0.424

0.237

Tris(chloroethyl) phosphate

0.041

0.038

0.020

0.053

0.005

0.009

0.007

0.429

1.003

0.146

0.162

0.406

0.249

0.502

0.007

0.011

0.004

0.018

0.009

0.008

0.005

0.017

0.009

0.010

0.021

0.008

0.008

0.006

0.085

0.147

0.063

0.354

0.139

0.056

0.035

Diphenyl-2-ethylhexylphosphate

0.019

0.019

0.016

0.029

0.018

0.009

0.013

Tris(ethylhexyl)phosphate

0.004

0.005

0.009

0.027

0.008

0.003

< LOD

Tri-o-cresyl phosphate

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

Tri-omp-cresyl phosphate

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

Tri-oom-cresyl phosphate

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

Tri-oop/omm-cresyl phosphate

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

Tri-opp-cresyl phosphate

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

< LOD

0.003

0.009

0.015

0.012

0.002

0.002

0.007

Tri-mmp-cresyl phosphate

0.005

0.011

0.020

0.015

0.002

0.003

0.010

Tri-mpp-cresyl phosphate

0.003

0.008

0.015

0.011

0.002

0.002

0.006

Tri-p-cresyl phosphate

0.001

0.003

0.008

0.004

0.001

0.001

0.003

Trixylyl phosphate

< LOD

0.039*

< LOD

0.030*

< LOD

< LOD

< LOD

Sum of all TCP isomers

0.012

0.032

0.058

0.042

0.007

0.008

0.026

Sum of all compounds

1.431

2.076

0.715

1.475

0.818

0.923

0.820

M AN U

TE D

AC C

EP

Tri-m-cresyl phosphate

SC

Triisobytyl phosphate

Tris(chloroisopropyl) phosphate (TCPP) Tris(1,3-dichloroisopropyl) phosphate Triphenyl phosphate (TPP) Tris(butoxyethyl) phosphate

572 573 574 575 576 577 578 579

RI PT

570 571

*measured in two samples only; LOD = Limit of detection

29

Table 5 Distribution of OPC metabolites in human urine. Comparison of OPC derived metabolite concentrations in urine from affected flight personnel [38] and from normal children in German day care centers [39]. Tributyl phosphate (TnBP) ↓

Triphenyl phosphate (TPP) ↓

Measured OPC metabolite

Dibutyl phosphate (DnBP)

Diphenyl phosphate (DPP)

Metabolite concentration in urine

[µg/L]

Flight personnel [38]

[µg/L]

[µg/L]

Median

95th percentile

Max

Median

95th percentile

Max

Median

95th percentile

Max

0.28

1.38

9.72

1.1

6.25

302

0.16

1.22

6.87

0.2

0.9

6.6

0.8

4.0

0.5

8.4

(n = 332) Children [39]

Tris(chloroisopropyl) phosphate (TCPP) ↓ Bis(chloroisopropyl) phosphate (BCPP)

RI PT

Organophosphate (OPC) ↓

23.2

M AN U

(n = 312)

SC

580 581 582

Building and Environment

ACCEPTED MANUSCRIPT

AC C

EP

TE D

583

<0.2

ACCEPTED MANUSCRIPT

30 No. of

Study

Smell

TCAC

∑VOC

VOC

∑ALD

T TO/C L

ALD

∑OPC

T TO/C L

OPC T TO/C L

TBP

TBP T TO/C L

TCP

TCP T TO/C L

Table 6 Comparison of 142 non-event flights, 17 smell event flights, and 18 TCAC event flights according to the aircraft/engine combinations investigated. For the individual compound classes (e.g. VOCs), cumulative values (∑) of the mean concentrations are given. The presented flight phase pictograms have been explained in the text. T = taxi-out; TO/C = takeoff and climb; D/L = descent and landing; No. = number; = reported “smell events” during TO/C, *one flight during D/L and **prior to taxi-out; = classified as a TCAC event due to elevated TCP release; = putative TCAC event; VOC = volatile organic compound; a = main VOC: ethanol, propanol, 1,2-propanediol; b = main VOC: n-propanol; c = main VOC: ethanol, acetic acid; d = main VOC: ethanol, propanol, benzoic acid, acetonitrile, 1,2-propanediol; e = one sample only; ALD = aldehydes according to DIN ISO 16000-6; OPC = organophosphorus compounds; TBP = tributyl phosphate; TCP = tricresyl phosphate ; EASA = [14]; LH-1/2 = unpublished QAC campaigns on A380 (1) and A321 (2) aircraft conducted by Lufthansa (LH); values in parentheses = 95th percentile; n. a. = not analyzed.

AC C

584 585 586 587 588 589 590 591

EP

TE D

M AN U

SC

RI PT

Aircraft/engine type

Building and Environment

ACCEPTED MANUSCRIPT

Building and Environment

592

10 6 48 10 12 8 8 142 1 15 1 17 1 12 1 1 2 1 18

-

-

LH-2

-

-

EASA

-

-

EASA

-

-

LH-1

-

-

EASA

-

-

EASA

-

-

EASA

-

-

EASA

-

-

Mean values of selected flights LH-2 LH-1 EASA

* **

-

Mean values of selected flights EASA LH-2

-

LH-2

-

LH-2

-

EASA

-

LH-1

-

Mean values of selected flights

3

[ng/m ]

186a (450)

29 (68)

1.5 (2.8)

0.8 (1.7)

6 (26)

241d (565)

23 (43)

2.9 (7.5)

1.7 (5.8)

10 (35)

295a (530)

27 (63)

2.0 (4.8)

0.3 (0.7)

6 (16)

181a (292)

24 (59)

0.7 (1.3)

0.5 (0.9)

11 (54)

293b (887)

27 (70)

1.5 (3.7)

0.9 (2.3)

26 (137)

192a (393)

22 (46)

0.9 (1.8)

0.1 (0.2)

5 (15)

320a (501)

21 (52)

0.8 (2.1)

0.4 (1.2)

2 (6)

22 (47)

1.1 (2.7)

0.5 (1.3)

9 (30)

24 (61)

0.9 (1.9)

0.3 (0.7)

26 (68)

24 (57)

1.4 (3.2)

0.6 (1.6)

11 (43)

28 (30)

1.0 (1.1)

0.3 (0.3)

26 (41)

26 (54)

1.5 (3.9)

1.0 (2.6)

16 (64)

M AN U

No. of non-event flights A321 IAE V2500 A380 RR Trent 900 B757-330 RR RB211-535E4B No. of smell event flights A320 CFM56-5A1 A321 IAE V2500 A321 IAE V2500 A321 IAE V2500 A340-642 RR Trent 556-61 A380 RR Trent 900 No. of TCAC flights

29

EASA

3

[µg/m ]

215a (383) 123a (253) 227 (473) b

280 (333) 162b (331) n. a.

TE D

A321 IAE V2500 A321-211 CFM56-5B3/3 A340-642 RR Trent 556-61 A380 RR Trent 900 B747-830 GEnx-2B67 B767-31BER GE CF6-80C2B6F B767-330ER PW4062 B787-8 RR Trent 1000

11

3

[µg/m ]

19 (35)

1.1 (2.6)

0.6 (1.9 )

13 (17)

221 (332)

n. a.

24 (40)

1.2 (2.5)

0.6 (1.6)

18 (41)

370a (524)

25 (41)

1.8 (2.2)

1.4 (1.8)

60 (110)

284a (541)

22 (34)

1.8 (3.1)

0.9 (1.9)

50 (110)

122c (188)

15 (15)

2.9 (3.8)

1.0 (1.2)

932 (1670)

245a (418)

35 (43)

5.4 (5.4)

3.4 (3.5)

465 (558)

296a (623)

25 (58)

1.1 (1.9)

0.5 (1.0)

291 (1190)

13 (17)

1.2 (2.0)

0.4 (0.5)

344 (859)

23 (35)

2.4 (3.1)

1.3 (1.6)

357 (750)

EP

CFMI CFM56-5A1/5B4

3

[µg/m ]

AC C

A320-211/214

3

[µg/m ]

RI PT

flights

SC

31

342e

277 (439)

n. a.

32 593

Building and Environment

ACCEPTED MANUSCRIPT

Figures

594

M AN U

SC

RI PT

595

596

608

TE D

EP

607

Figure 1 A) Compilation of flight phase-dependent air contaminant pictograms (cumulative values) taken from studies by EASA [4,14,20], MHH [2,22,23], and Cranfield [21]. Note that the relative concentrations in the pictograms do not allow for quantitative comparison between different studies of the same compound classes. However, the concentrations of VOCs have comparable orders of magnitude (e.g., the mean values of toluene ranged from 2-16 µg/m3 for all studies). VOC = sum of all VOCs measured during the corresponding flight phase; PFC = perfluorinated compounds; aldehydes = aldehydes as determined with the DNPH method. B) Concentration profile of toluene (μg/m3) during the investigated flight phases, taken from the Cranfield study [21]. The arrows indicate how the flight phases of the Cranfield study were combined in order to match the flight phases investigated in the present study.

AC C

597 598 599 600 601 602 603 604 605 606

33

Building and Environment

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

609

610

TE D

EP

619

Figure 2 Validation of the effect of different ventilation rates during different phases of flight by comparison of mean CO2 concentrations (black line, taken from the EASA study) with calculated CO2 concentrations (dark grey line). The calculation is based on formulas and parameters given in the EASA study [14]. Especially during taxi-out (left box), the calculated decrease in CO2 for an air exchange rate of 20 h-1 (Λ) differs significantly from the measured value. During taxi-out and takeoff, there is commonly known to be a decrease in the air exchange rate. The mean CO2 concentrations measured during the sampled flight phases are illustrated by greyscale boxes from which the typical trend of decreasing concentration can be derived.

AC C

611 612 613 614 615 616 617 618

34

Building and Environment

ACCEPTED MANUSCRIPT

SC

RI PT

620

EP

TE D

Figure 3 Flight phase pictograms (T, TO/C, D/L) of VOCs measured by the EASA study [14] (mean values given in mg/m3), which reveal deviations in the trend of decreasing concentration related to the effect of different ventilation rates during different phases of flight A) Pictograms of VOCs from commonly used consumer products, which are used intermittently during flight (body care and food). B) Sum of all isoalkane fractions, which are mostly released during takeoff and are considered to be from an unknown lubricant used in aircraft. C) Deviations of increased m/p-TCP concentrations observed during two independent (*) A340-6 flights (either in the cockpit or cabin) and a single A320 flight (cabin).

AC C

622 623 624 625 626 627 628 629

M AN U

621

35

Building and Environment

ACCEPTED MANUSCRIPT

References

631

[1] Rosenberger W. Effect of charcoal equipped HEPA filters on cabin air quality in aircraft. A

632

case study including smell event related in-flight measurements. Building and Environ-

633

ment 2018.

634

[2] Rosenberger W, Beckmann B, Wrbitzky R. Airborne aldehydes in cabin-air of commercial aircraft:

636

dinitrophenylhydrazones.

637

2016;1019:117–27.

by J

HPLC

with

Chromatogr

UV B

absorbance

Analyt

detection

Technol

Biomed

of

2,4-

Life

Sci

[3] Denola G, Hanhela PJ, Mazurek W. Determination of tricresyl phosphate air contamina-

M AN U

639

Measurement

SC

635

638

RI PT

630

tion in aircraft. The Annals of occupational hygiene 2011;55(7):710–22. [4] Ree H de, van den Berg M, Brand T, Mulder GJ, Simons R, van Veldhuijzen Zanten B et al.

641

Health risk assessment of exposure to TriCresyl Phosphates (TCPs) in aircraft: a com-

642

mentary. Neurotoxicology 2014;45:209–15.

645 646 647 648 649 650

2002;89(2):146–64.

[6] Weiner ML, Jortner BS. Organophosphate-induced delayed neurotoxicity of tri-

EP

644

[5] Winder C, Balouet J-C. The Toxicity of Commercial Jet Oils. Environmental Research

arylphosphates. Neurotoxicology 1999;20(4):653–73.

AC C

643

TE D

640

[7] HENSCHLER D. Tricresylphosphate poisoning; experimental clarification of problems of etiology and pathogenesis. Klin Wochenschr 1958;36(14):663–74. [8] Furlong CE. Exposure to triaryl phosphates: Metabolism and biomarkers of exposure. Journal of Biological Physics and Chemistry 2011;11.

651

[9] Wang L, Liu Y-H, Xu Y, Yin W, Huang Y-G. Thirteen-year follow-up of patients with tri-

652

ortho-cresyl phosphate poisoning in northern suburbs of Xi'an in China. Neurotoxicology

653

2009;30(6):1084–7.

36

Building and Environment

ACCEPTED MANUSCRIPT

654

[10]Dumyahn TS, Spengler JD, Burge HA, Muilenburg M. Comparison of the environments of

655

transportation vehicles: Results of two surveys. In: Nagda NL, editor. Air Quality and

656

Comfort in Airliner Cabins; 2000, p. 3–23.

658

[11]Felten C, Metzdorf U, Weiß T, Hedtmann. Air quality in aircraft – Causes of odors and relationship with health complaints. Flug u Reisemed 2013;2013(20):19–23.

RI PT

657

[12]Guan J, Yang X, Li Z. Source contributions and control strategies of volatile organic com-

660

pounds (VOCS) in aircraft cabins. In: Indoor Air 2014 - 13th International Conference on

661

Indoor Air Quality and Climate; 2014, p. 727–734.

663 664 665

[13]Shehadi M, Jones B, Hosni M. Characterization of the frequency and nature of bleed air

M AN U

662

SC

659

contamination events in commercial aircraft. Indoor Air 2016;26(3):478–88. [14]Schuchardt S, Bitsch A, Koch W, Rosenberger W. CAQ - Preliminary cabin air quality measurement campaign. EASA_REP_RESEA_2014_4; 2017.

[15]Beko G, Allen JG, Weschler CJ, Vallarino J, Spengler JD. Impact of cabin ozone concentra-

667

tions on passenger reported symptoms in commercial aircraft. PLoS One

668

2015;10(5):e0128454.

TE D

666

[16]Jiang C, Zhang P. Indoor carbonyl compounds in an academic building in Beijing, China:

670

concentrations and influencing factors. Frontiers of Environmental Science &

671

Engineering 2012;6(2):184–94.

AC C

EP

669

672

[17]Hofmann H, Plieninger P. Bereitstellung einer Datenbank zum Vorkommen von

673

flüchtigen organischen Verbindungen in der Raumluft. ISSN 1862-4340. Dessau-Rosslau;

674

2008.

675 676

[18]Birger H, Ostendorp G. Raumluftuntersuchungen in öffentlichen Gebäuden in SchleswigHolstein: Teil 4: Neubauten für Schulen und Kindergärten. ISSN 0935-4379. Kiel; 2013.

37 677 678

Building and Environment

ACCEPTED MANUSCRIPT

[19]World Health Organization. WHO Guidelines for Indoor Air Quality: Selected Pollutants. Geneva; 2010. [20]Guan J, Gao K, Wang C, Yang X, Lin C-H, Lu C et al. Measurements of volatile organic

680

compounds in aircraft cabins. Part I: Methodology and detected VOC species in 107

681

commercial flights. Building and Environment 2014;72(0):154–61.

682 683

RI PT

679

[21]Crump D, Harrison P, Walton C. Aircraft Cabin Air Sampling Study; Part 1 of the Final Report: Cranfield University; 2011.

[22]Rosenberger W, Wrbitzky R, Elend M, Schuchardt S. Determination of organic com-

685

pounds in cabin air after aircraft de-icing of commercial aircraft. Gefahrstoffe

686

Reinhaltung Der Luft 2014;74(11-12):467–75.

689 690 691

M AN U

688

[23]Schmitz-Spanke S, Nesseler T, Letzel S. Umweltmedizin: Neue Erkenntnisse aus Wissenschaft und Praxis; 2017; pp. 163-177.

[24]Wolkoff P, Crump DR, Harrison PTC. Pollutant exposures and health symptoms in aircrew

TE D

687

SC

684

and office workers: Is there a link? Environment International 2016;87:74–84. [25]Umweltbundesamt. Beurteilung von Innenraumluftkontaminationen Beurteilung von Innenraumluftkontaminationen

693

Bundesgesundheitsbl - Gesundheitsforsch - Gesundheitsschutz. DOI:10.1007/s00103-

694

007-0290-y; 2007.

696 697 698

AC C

695

mittels

Referenz-

und

Richtwerten:

EP

692

[26]Indoor

air

guide

values

for

C9-C15

alkylbenzenes.

Bundesgesundheitsblatt

Gesundheitsforschung Gesundheitsschutz 2012;55(9):1201–14. [27]Indoor air guide values for cresols. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2012;55(8):1061–8.

38 699

Building and Environment

ACCEPTED MANUSCRIPT

[28]Indoor

air

guide

values

for

monocyclic

monoterpenes

(limonene).

700

Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2010;53(11):1206–

701

15. [29]Solbu K, Daae HL, Olsen R, Thorud S, Ellingsen DG, Lindgren T et al. Organophosphates in

703

aircraft cabin and cockpit air--method development and measurements of contami-

704

nants. Journal of environmental monitoring JEM 2011;13(5):1393–403.

RI PT

702

[30]Lamb J, McGonagle C, Cowie H, Cherrie JW. Cabin Air - surface residue study; 2012.

706

[31]Yadav IC, Devi NL, Zhong G, Li J, Zhang G, Covaci A. Occurrence and fate of organophos-

707

phate ester flame retardants and plasticizers in indoor air and dust of Nepal: Implication

708

for human exposure. Environ Pollut 2017;229:668–78.

710

M AN U

709

SC

705

[32]Marklund A, Andersson B, Haglund P. Screening of organophosphorus compounds and their distribution in various indoor environments. Chemosphere 2003;53(9):1137–46. [33]Vykoukalova M, Venier M, Vojta S, Melymuk L, Becanova J, Romanak K et al. Organo-

712

phosphate esters flame retardants in the indoor environment. Environment interna-

713

tional 2017;106:97–104.

TE D

711

[34]Wu M, Yu G, Cao Z, Wu D, Liu K, Deng S et al. Characterization and human exposure as-

715

sessment of organophosphate flame retardants in indoor dust from several microenvi-

716

ronments of Beijing, China. Chemosphere 2016;150:465–71.

AC C

EP

714

717

[35]Brandsma SH, Boer J de, van Velzen MJM, Leonards PEG. Organophosphorus flame re-

718

tardants (PFRs) and plasticizers in house and car dust and the influence of electronic

719

equipment. Chemosphere 2014;116:3–9.

720

[36]Zhou L, Hiltscher M, Puttmann W. Occurrence and human exposure assessment of or-

721

ganophosphate flame retardants in indoor dust from various microenvironments of the

722

Rhine/Main region, Germany. Indoor Air 2017;27(6):1113–27.

39 723 724

Building and Environment

ACCEPTED MANUSCRIPT

[37]Luongo G, Ostman C. Organophosphate and phthalate esters in settled dust from apartment buildings in Stockholm. Indoor Air 2016;26(3):414–25. [38]Schindler BK, Weiss T, Schutze A, Koslitz S, Broding HC, Bunger J et al. Occupational ex-

726

posure of air crews to tricresyl phosphate isomers and organophosphate flame retard-

727

ants after fume events. Archives of toxicology 2013;87(4):645–8.

RI PT

725

[39]Fromme H, Lahrz T, Kraft M, Fembacher L, Mach C, Dietrich S et al. Organophosphate

729

flame retardants and plasticizers in the air and dust in German daycare centers and hu-

730

man biomonitoring in visiting children (LUPE 3). Environment international

731

2014;71:158–63.

734 735

M AN U

733

[40]Sagunski H, Rosskamp E. Not Available. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2002;45(3):300–6.

[41]Spengler JD, Vallarino J, McNeely E, Estephan H. In-Flight/Onboard Monitoring: ACER’s Component for ASHRAE 1262, Part 2; 2012.

TE D

732

SC

728

[42]Cao X, Zevitas CD, Spengler JD, Coull B, McNeely E, Jones B et al. The on-board carbon

737

dioxide concentrations and ventilation performance in passenger cabins of US domestic

738

flights. Indoor and Built Environment 2018;124:1420326X1879399.

740 741 742 743 744 745 746

[43]Lee SC, Poon CS, Li XD, Luk F. Indoor air quality investigation on commercial aircraft. In-

AC C

739

EP

736

door Air 1999;9(3):180–7. [44]Lindgren T, Norback D. Cabin air quality: Indoor pollutants and climate during intercontinental flights with and without tobacco smoking. Indoor Air 2002;12(4):263–72. [45]Guan J, Li Z, Yang X. Net in-cabin emission rates of VOCs and contributions from outside and inside the aircraft cabin. Atmospheric Environment 2015;111:1–9. [46]Spengler JD, Wilson DG. Air quality in aircraft. Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering 2003;217(4):323–36.

40

Building and Environment

ACCEPTED MANUSCRIPT

747

[47]Cao Q, Xu Q, Liu W, Lin C-H, Wei D, Baughcum S et al. In-flight monitoring of particle

748

deposition in the environmental control systems of commercial airliners in China. At-

749

mospheric Environment 2017;154:118–28.

751

[48]Wang C, Yang X, Guan J, Li Z, Gao K. Source apportionment of volatile organic compounds (VOCs) in aircraft cabins. Building and Environment 2014;81:1–6.

RI PT

750

[49]Wang C, Yang X, Guan J, Gao K, Li Z. Volatile organic compounds in aircraft cabin: Meas-

753

urements and correlations between compounds. Building and Environment 2014;78:89–

754

94.

756 757 758

[50]German Federal Bureau of Aircraft Accident Investigation. Study of reported occurrences

M AN U

755

SC

752

in conjunction with cabin air quality in transport aircraft.

[51]Shehadi M, Jones B, Hosni M. Characterization of the Frequency and Nature of Bleed Air Contamination Events in Commercial Aircraft. Indoor air 2015. [52]Murawski JTL, Supplee DS. An attempt to characterize the frequency, health impact, and

760

operational costs of oil in the cabin and flight deck supply air on U.S. commercial air-

761

craft. Journal of ASTM International 2008;5(5).

TE D

759

[53]Overfelt RA, Jones BW, Loo SM, Haney RL, Neer AJ, Andress JR, Yang X, Zitova A, Prorok

763

BC, Fergus JW, Simonian AL, Kiepert J, Pook M and Anderson M. Sensors and Prognos-

764

tics to Mitigate Bleed Air Contamination Events. Progress Report 2012; 2012.

AC C

EP

762

765

[54]Safety Regulation Group. Cabin Air Quality; 2004.

766

[55]Eckels SJ, Jones B, Mann G, Mohan KR, Weisel CP. Aircraft Recirculation Filter for Air-

767 768 769

Quality and Incident Assessment. Journal of aircraft 2014;51(1):320–6. [56]Inoue N, Fujishiro K, Mori K, Matsuoka M. Triorthocresyl phosphate poisoning--a review of human cases. Journal of UOEH 1988;10(4):433–42.

41

Building and Environment

ACCEPTED MANUSCRIPT

770

[57]Classen W, Gretener P, Rauch M, Weber E, Krinke GJ. Susceptibility of various areas of

771

the nervous system of hens to TOCP-induced delayed neuropathy. Neurotoxicology

772

1996;17(3-4):597–604. [58]Health and Safety Executive: Workplace exposure limits. EH40/2005. 2011st ed.

774

[59]Bagshaw M. Health Effects of Contaminants in Aircraft Cabin Air. Aerospace Medical

775

RI PT

773

Association; 2013.

[60]Weiss T, Schindler BK, Schutze A, Koslitz S, Broding HC, Bunger J et al. Reply to the letter

777

of Anderson J entitled "Comment on Schindler, BK; Weiss, T; Schutze, A; et al. Occupa-

778

tional exposure of air crews to tricresyl phosphate isomers and organophosphate flame

779

retardants after fume events, Arch Toxicol (2013) 87: 645-648". Archives of toxicology

780

2015;89(2):263–4.

M AN U

SC

776

AC C

EP

TE D

781