Modeling Particle Emissions from Three ... - ACS Publications

Subscriber access provided by UNIV OF SOUTHERN INDIANA

Environmental Modeling

Modeling Particle Emissions from 3D printing with ABS Polymer Filament Tracy Zontek, Scott Hollenbeck, Burton R Ogle, and John T Jankovic Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.9b02818 • Publication Date (Web): 26 Jul 2019 Downloaded from pubs.acs.org on July 28, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 23

Environmental Science & Technology

RUNNING HEAD: Modeling Particle Emissions

1

Modeling Particle Emissions from 3D printing with

2

ABS Polymer Filament

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Tracy L. Zontek, PhD, CIH, CSP* University of Tampa 401 W. Kennedy Blvd. Tampa, FL 33606 [email protected] Scott Hollenbeck, MPH, CIH Oak Ridge National Laboratory, Center for Nanophase Materials Sciences PO Box 2008 MS-6487 Oak Ridge, TN 37831-6487 Burton R. Ogle, PhD, CIH, CSP Western Carolina University 4121 Little Savannah Road Cullowhee, NC 28723 John Jankovic, MSPH, CIH Oak Ridge National Laboratory, Center for Nanophase Materials Sciences PO Box 2008 MS-6487 Oak Ridge, TN 37831-6487

1 ACS Paragon Plus Environment


RUNNING HEAD: Modeling Particle Emissions 24

Abstract

25

An eddy diffusion model using data from a desktop 3D printer was developed

26

under laboratory conditions and then coupled with Monte Carlo analysis to

27

estimate the potential range of particulate concentrations in and around various

28

industrial-size 3D printers. In this case large additive manufacturing processes

29

using acrylonitrile-butadiene-styrene. polymer feedstock. The model employed

30

mass emission estimates determined from thermal gravimetric analysis and

31

printer enclosure particle loss rates. Other model inputs included ranging terms

32

for extrusion rate. temperature, print time, source-to-receiver distance, printer

33

positions, particle size fraction, and environmental diffusivity estimates based

34

on air changes per hour Monte Carlo analysis bracketed measured

35

environmental particulate concentrations associated with large scale additive manufacturing processes

36

(3D printing). Statistically there was no difference between the average near field particle concentrations

37

measured and that of the model-derived average. However, the model began to vary higher statistically, if

38

not practically, from air monitoring results in the far field. Diffusivity and extrusion rate emerged as the

39

two most important variables in predicting environmental concentrations. This model can be used to

40

estimate air concentrations over a range of varying conditions, such as one might employ in a “what if”

41

type of evaluation to estimate employee exposure. For example, as a compliance effort with OSHA

42

standard 29 CFR Part 1910.132 requireing a formal hazard assessment for work environments as a

43

“before exposure” effort to determine if respiratory protection is needed.

44 45


Page 2 of 23

Page 3 of 23



Introduction

47

There has been a substantial increase in use of additive manufacturing and associated materials;

48

projections of the future market range from $20 billion by 2020 and $100-250 billion by 2025 (1) (2) (3).

49

The main industries utilizing some form of additive manufacturing include aerospace and defense,

50

automotive, medical, and consumer-goods industries (2). Additive manufacturing has resulted in new

51

anthropogenic sources of nanoscale or ultrafine particulate (particles < 0.1 um or 100 nm, UFP). Early

52

research has identified that nanoscale materials can reach all areas of the respiratory system, including the

53

alveoli and thus have the potential to move throughout the body (4). Specific research in the field of

54

additive manufacturing has identified an asthma case study (5), asthma and allergic rhinitis in those

55

working with acrylonitrile butadiene styrene (ABS) filament more than 40 hours per week (6), and

56

inhalational studies of rats documented acute hypertension (7). Emerging research of additive

57

manufacturing has reported particle emission rates of 108 to 1011 particles per minute in size ranges

58

classified as UFP during additive manufacturing (3D printing) with desktop size printers (8). It has been

59

reported in clinical studies that inhalation of ultrafine carbon particles causes subtle alterations in vascular

60

function in healthy non-exercising subjects at about 10µg/m3 (9) (10). Further, an increase in outdoor

61

UFP concentrations of approximately 10,000 #/cc has been associated with an increased risk of mortality

62

by 3% (11).

63 64

Regulatory limits for ultrafine particles include the Occupational Safety and Health Administration

65

(OSHA) Particulate Not Otherwise Classified, Respirable (< 4 µm, 5 mg/m3) measured as an eight-hour

66

time weighted average. The Environmental Protection Agency (EPA) has a National Ambient Air Quality

67

Standard for PM2.5 (35 µg/m of particulate matter less than 2.5 µm, measured as a 24-hour average) (12).

68

Additive manufacturing particulate emissions, mostly in the UFP range have no specific exposure limit.

69

Aerosol mass concentration is low, even at high particle concentrations, making traditional industrial

70

hygiene filter monitoring an aerosol that is composed of carbonaceous materials purportedly capable of

71

impacting respiratory health, highly problematic (13). 3 ACS Paragon Plus Environment


RUNNING HEAD: Modeling Particle Emissions 72 73

Thermal oxidative degradation of various polymers used as filament materials during heating and

74

extrusion are known to release a variety of irritants and systemic toxins in low amounts (14).

75

Manufacturing applications using 3D printing are burgeoning, and the availability of low-cost desk-top

76

size printers suggest their use in non-industrial settings, where poor ventilation may exist, will become

77

more prevalent as well (13). Azimi, Fazli, and Stephens (15) used particle emissions and volatile organic

78

compound concentrations from published studies to estimate the impact of different ventilation scenarios;

79

the conclusions drawn mimic traditional industrial hygiene controls, specifically the importance of

80

controlling the contaminant at the source. They found that high flow/high capture efficiency spot

81

ventilation systems, followed by using a 3D printer with an enclosure that filters emissions to be the most

82

effective (15).

83 84

The expanding use of additive manufacturing, limited data in the literature, a lack of exposure limits, as

85

well as the absence of standardized air monitoring protocols and methods of data analysis, make a strong

86

case for using readily available models coupled with uncertainty analysis to elucidate aerosol

87

concentration and dispersion during 3D printing operations (16) (17). Modeling can provide an estimate

88

of aerosol concentration that can be used to conduct a risk assessment and determine appropriate controls.

89

This study uses particle emissions from a desktop 3D printer to develop and validate a modeling

90

approach; then applies the approach to a large scale area additive manufacturing process.

91 92


Page 4 of 23

Page 5 of 23



Materials and Methods

94

A Da Vinci XYZ model 1.0 3D desktop printer (DTP) was used in poorly ventilated room (~1.8 air

95

changes per hour) in order to obtain emission estimates and air concentration maps for model inputs.

96

Time series measurements were collected inside and outside the DTP enclosure. The printer enclosure

97

itself was not designed to contain printing emissions as there were numerous openings allowing free air

98

movement from and into the environment. Speculatively, the enclosure was meant to prevent contact with

99

the heated/ moving elements of the printer. These openings served as sampling ports inside the enclosure;

100

temperature data inside the DTP enclosure was not monitored since it was open to free air flow. The

101

monitoring position outside the enclosure was located at a point representative of the breathing zone of a

102

person sitting at the table on which the printer was located (Figure 1). Background particulate

103

concentrations were determined prior to filament heating and extrusion.

104 105

Figure 1. Enclosed DTP and breathing zone/ambient air monitoring positions.

106 107

Room concentration maps were constructed for background, heating, early printing, and late stage

108

printing using 3DField mapping freeware (18) and a handheld TSI® model 3007 Condensation Particle

109

Counter (CPC) with size detection from 10nm to > 1µm size at concentrations up to 105 particles/cm3).

110

Mapping was used to compare model results with air sampling results. Particle size and mass

111

concentration was determined using a combination of scanning mobility particle analyzers (SMPS)

112

including a TSI® SMPS 3080 equipped with either a long differential mobility analyzer (DMA) or nano

113

DMA (TSI® 3085). The long DMA and nano DMA were alternated in order to obtain particle sizes from

114

2 nm to 300 nm. A TSI® optical particle sizer (OPS) was simultaneously used to measure particle size




information from 300 nm to 10 µm. TSI’s MIM merging software (19), was used to combine the small

116

fraction mobility diameters with the large fraction optical diameters into one normalized particle size

117

distribution as either ΔN/Δlog dp or ΔM/Δlog dp. By definition, both electrical mobility and optical

118

diameters refer to equivalent spherical particle behavior in an electric field or in a scattered light beam

119

respectively. These instruments provided corresponding number and mass concentrations using TSI®

120

algorithms provided in their respective AIM® software. The concentration measurements from the DTP

121

trials were used in comparison with model predictions in order to refine input variables. The conversion

122

from count to mass on the basis of sphericity is justified on the basis that the instruments used to measure

123

mass were also based on light scattering and the assumption of sphericity. Additionally, the density of

124

ABS plastic is about 1g/cc.

125 126

The DTP used in the test room was operated at a printing temperature of 213°C using an ABS filament.

127

These are fixed parameters for this particular printer. The resolution was set at best quality (fine 0.1 mm)

128

resulting in the longest printing time. Filament feed rate is a variable function of the part being printed, as

129

is the amount of filament used. The test printer did not provide the precise amount of filament used.

130

Therefore, the same part was printed for all of the trial measurements thereby keeping quantity and print

131

time constant. The final part mass was not measured; if the printing build had any issues it was re-started.

132 133

The near-field eddy diffusion model was selected in favor of a one or two-compartment box model

134

because it does not require an assumption of a “well mixed” environment as do the others. (20)

135

Further, this model accounts for varying distances, more characteristic of 3D printing operations where

136

printer operators and ancillary personnel are not stationary. When accompanied with concentration

137

mapping, the model can provide useful information for time and motion type exposure assessment. Either

138

the number or mass emission rate of the process is necessary to estimate concentration with models that

139

account for removal by general ventilation (21), or more complex models such as the eddy diffusion

140

model for predicting concentration at various distances and times (22). Since personnel may move about 6 ACS Paragon Plus Environment

Page 6 of 23

Page 7 of 23



in the printing area, the eddy diffusion model was used to estimate concentrations at various points in the

142

near exposure field. The number emission rate (E) was calculated from CPC measured particle

143

concentration decay after extrusion was completed (Eq. 1) (23). The loss rate factor and emission rate

144

contained all of the particle sinks such as removal by ventilation, wall losses, and agglomeration (Eq 1,2)

145

(8).

146

147

Eq.(1) 𝐿 =

(𝐶𝑠𝑡𝑎𝑟𝑡 ― 𝐶𝑏𝑔𝑠𝑠) 𝑙𝑛 (𝐶 ― 𝐶 𝑒𝑛𝑑 𝑏𝑔𝑠𝑠) ∆𝑡

, Where

148

L is total particle loss rate by all particle removal mechanisms (min -1)

149

Cstart is particle concentration at end of printing (#/cm3)

150

Cend is particle concentration at end of decay period (#/cm3)

151

Cbgss is background particle concentration before printing (#/cm3)

152

Δt is elapsed time between Cstart and Cend

153

Note: background is a measured quantity either taken at a time just preceding the printing operation or it

154

was derived from historical monitoring during extended periods when printing is not

155

taking place.

156

Eq. (2) 𝐸 = 𝐿 ∙ 𝐶𝑝𝑠𝑠 ∙ 𝑉, where

157

E is particle emission rate (#/min)

158

Cpss is printer enclosure or room concentration at steady state minus background

159 160

(#/cm3) V is volume of printer or room enclosure (cm3)

161 162

It was necessary to convert CPC measured particle concentrations to mass concentrations for comparison

163

with mass-based exposure data both by us and others such as that by Frampton et al (10). The relationship

164

between number concentration and mass concentration, as functions of geometric mean particle diameter

165

and geometric standard deviation, was approximated from SMPS data. The algorithm used to convert #/cc 7 ACS Paragon Plus Environment



to µg/m3 used the mean diameter from the distribution of diameters and assumed particle sphericity. The

167

justification for this approach was discussed earlier. Mass emission rates for ABS filaments used in

168

printing at various temperatures were determined experimentally by thermal gravimetric analysis (TGA)

169

using a TGA Q5000; this study focused on particle emission rates as printing temperatures were not high

170

enough for much vapor formation. A Monte Carlo analysis (Analytica 2.0) using the eddy diffusion

171

model was performed to evaluate the distribution of possible aerosol concentrations as extrusion rate,

172

extrusion temperature, printing time, distance of observer from source, and diffusivity variables that were

173

allowed to vary (Figures 2 and 3) (24).

174 175

Eq. (3)

𝐺

C𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑎𝑡 𝑝𝑜𝑖𝑛𝑡 r = 2𝜋𝐷𝑟𝑒𝑟𝑓𝑐

[

𝑟 (4𝐷 ∙ ∆𝑡)

], where

176 177

D

= eddy diffusivity

178

r

= sample point distance from source

179

Δt

= elapsed time since start of release

180

G

= continuous emission rate

181

erfc = 1- error function

182 183

Figure 2. Model variables and outcomes

184


Page 8 of 23

Page 9 of 23



Variable importance, the relative contribution to uncertainty in the outcome, was used to indicate which

186

variables were most critical to the exposure estimates; and, where possible, additional efforts were made

187

to refine those variables. Filament extrusion rate is governed by operator choices such as build quality

188

and build rate. Sophisticated printers (Figure 3) may provide accurate information on printing time and

189

polymer consumption thus providing an accurate average extrusion rate. Less expensive printers provide

190

less accurate or no information from which to determine average extrusion rate. The test printer provided

191

filament usage accurate to the nearest meter. This study measured printing time (hh/mm/ss) rather than

192

using the printer estimates; the extrusion rate in length per minute was converted to weight per minute

193

based on the density of the filament type.

194 195

Eddy diffusion is a process by which a substance mixes in ambient air due to air currents. We initially

196

used diffusivities reported for various industrial processes other than 3D printing, which were later

197

replaced with more current information (25) relating diffusivity to ventilation in terms of ACH (Eq. 4),

198

experimentally derived by Cheng (26).

199 200

Eq (4)

Diffusivity (m2/min) = L2((0.6∙ACH)+0.25)/60, where

201

L

202

ACH = mixing air changes in the room volume (min-1)

= cube root of the room volume (m)

203 204

The DTP derived model was used to estimate aerosol concentrations from two Big Area Additive

205

Manufacturing (BAAM) operations involving 3D printing changing only the relevant operational

206

parameters used by the printers. The BAAMs used were a prototype system (7’x7’x3’ build volume,

207

10lb/hr deposition rate) and a Cincinnati 100 ALPHA size 2 (8’x2’x‘6’ build volume, 100 lb/hr

208

deposition rate). As with the DTP, the aerosol concentrations were monitored (CPC, SMPS, OPS) at the

209

printer location in the large open-bay operation (Figure 3); the equipment cart was placed between the

210

Cincinnati and gantry printers, as determined by space and electrical resources, while not interfering with 9 ACS Paragon Plus Environment



egress or normal working operations. An additional CPC was located approximately 20 feet away from

212

the cart with the CPC, SMPS, and OPC; again, location was based on space conditions. A third point at

213

some distance and on a level above the printer floor was also monitored (mezzanine level). Plate flow

214

around the extruder was assumed to be laminar as the head movement is on the order of a few feet per

215

minute. Thus, eddy currents were not expected to result from the printer itself.

216 217

Figure 3. Large scale open-bay gantry 3D printer

218 219 220 221 222 223

Results The particles emitted by the DTP printer, presented as a cumulative frequency distribution in Figure 4a,

224

are almost entirely within the ultra-fine particulate classification (23). Particles collected near the printer

225

in the theoretical breathing zone of someone monitoring the operation are also presented as a normalized

226

number distribution. The aerosol has a bimodal distribution with modes at about 7 and 15 nm (Figure 4b).

227

This was typical of BZ values for most printing runs in the early stages of extruder heating and printing.

228

As time of printing progressed, often the particles grew in size to between 20 and 30 nm (not depicted).

229

Sixty-eight percent of the particle mass was present as UFP (Figure 4c). The normalized size distribution 10 ACS Paragon Plus Environment

Page 10 of 23

Page 11 of 23



again presented as bimodal with the largest mode at about 90 nm. A second, smaller mode was centered at

231

about 230 nm (Figure 4d). Particle size and concentration numerical data is presented in Table 2. Figure 5

232

depicts size variation over a 60-minute print job.

233 234

Fig 4. Cumulative number/mass frequencies (a,c) and representative aerosol number/mass particle size

235

(b,d) in BZ from SMPS (7 to 294 nm size instrument resolution).

236 237 238 239 240 241 242




number mobility dia. (nm)

number mobility dia. (nm)

mass mobility dia. (nm)

mass mobility dia. (nm)

243

Figure 5. Inside printer enclosure; variability in size and cumulative frequency distributions from SMPS

244

with Nano DMA (2 to 100 nm size resoultion).

245 246 247

Eddy diffusion modeling results for the test room containing the small enclosed 3D printer are presented

248

in Figure 6, modeling results for the large scale printing process are presented in Figures 7 and 8. The

249

measured values in Figure 7 are determined from a Monte Carlo analysis using the mean and standard

250

deviation (SD) from the large printing facility. The predicted values are derived from the eddy

251

diffusion/Monte Carlo model. As expected, allowing for variability in the inputs resulted in a wider data 12 ACS Paragon Plus Environment

Page 12 of 23

Page 13 of 23



spread; however, the model average was in good agreement with the sampling location averages (-1% to

253

+20%). Modeling results for the large printing facility (Figure 8) suggest about a 5% chance that an

254

exposure could exceed 10 µg/m3 (alterations in vascular function in healthy adults (9) (10)). Statistically

255

there was no difference in mean values obtained from the model and ambient measurements for the two

256

large-scale printers tested or the mezzanine. P values for a one-way ANOVA followed with a Post Hoc

257

Tukey HSD test are presented in Table 1.

258

Table 1. P values for model comparison P value Gantry with model prediction

0.344

Cincinnati with model prediction

0.899

Mezzanine with model prediction

0.155

259 260

Although the averages are in good agreement; individual point predictions vary widely but do bracket the

261

range of measured concentrations (Figure 7). The model’s utility then is in providing a probability

262

distribution from which one can consider the likelihood of some established limit being exceeded.

263

Autocorrelation in time series data is frequently an issue when normality and randomness are required in

264

order to apply normal statistics. Measured concentration data was tested for autocorrelation and was

265

found to be random (white noise) with p values >> 0.05. Tests were performed using an autocorrelation

266

function (ACF) in RealStat-2010 for Microsoft Excel. The ACF score was 0.08 where a value

267

approaching 1 is considered highly correlated.

268

Figure 6. Mapped mass concentration (a) and eddy diffusion model exposure prediction (b) for the test

269

printer




a P of exceeding 10 µg/m3 ≈ 0

270 271 272 273 274 275

Figure 7. Modeling predictions bound the monitoring values by location in the large printing facility


Page 14 of 23

Page 15 of 23



Gantry

Cincinnati

Mezzanine

Model

Concentration (µg/m3)

1E+05

1E+04

1E+03 0

20

40 60 Iteration (Run)

80

100

276 277 278 279 280 281

Figure 8. PM0.1 mass exposure prediction values for the large printing process.

282



Page 16 of 23


Table 2. Aerosol parameters measured inside and outside DTP enclosure during printing (ABS 213°C)

284

(13). Inside printer enclosure

Outside printer enclosure in breathing zone

SMPS

CPC

Calculated

(Avg, ~ 1

(Avg, ~ 1

hour)

hour)

SMPS

Printer Parameter function GM

Mode

GS

#/cc

µg/m3

GM (nm)

Mode

GSD

(nm)

(nm)

D

Mean

24.2

16.4

20.7

522,000*

83.3

14.7

9.3

1.79

SD

3.5

2.4

0.34

582,258*

70.2

2.9

2.5

0.56

Mean

30.5

34.1

1.95

71,450

7.6

16.3

9.4

1.71

SD

4.6

13.2

0.12

45,068

6.2

0.5

2.1

0.01

Mean

16.3

14.3

1.50

66,308

0.4

N = 10 for both sample sets.

SD

1.1

1.0

0.05

38,234

0.2

Door

Mean

25.6

19.2

2.17

834

0.2

open

SD

3.3

4.7

0.21

33

0.1

(nm)

Heating

Printing

Cooling

285

*Value underestimates actual concentration and requires dilution or coincidence correction not performed

286

here.

287 288


Page 17 of 23



TGA emission rates were determined for the range of extrusion temperatures expected in the printing

290

processes (Table 3). Temperatures varied during warm-up, filament changing/loading, and cool-down;

291

this variation was included in the modeling process using Equation 5 in the temperature variable.

292 293

294 295

Table 3. TGA derived mass emission rates for 1.75 mm diameter ABS filament. T (°C)

200

213

215

230

E (µg/mg)

9.2

10.9

11.7

14.8

Note: data represents the mass fraction estimates of the emissions only.

Eq. (5).

𝜇𝑔 𝑚𝑔

= 0.3733𝑒0.016𝑇, where

296 297

µg/mg is µg emitted/mg of filament extruded

298

T is extrusion temperature (°C)

299 300

Discussion

301

The eddy diffusion model as applied here is not novel. The utility of the approach here is to introduce the

302

use of modeling data in a visualized probabilistic format, i.e., model results plus a Monte Carlo analysis

303

to incorporate uncertainty and predicted concentration mapping to aid in visualization of possible

304

outcomes. Other researchers have implemented the eddy diffusion model for use by field industrial

305

hygienists (16) (22) (24) and this provides guidance to determine particle emissions from the emerging

306

field of additive manufacturing where traditional mass concentration measurements and regulatory limits

307

may not adequately describe the potential health risk. This technique, in contrast to a 3D numerical

308

simulation by Computational Fluid Dynamics, is more readily available to field industrial hygienists as it

309

can run in a spreadsheet and does not require additional software/training. Further, it only requires the use

310

of a CPC; the SMPS and OPC data were included to provide the basis for mass conversions to compare to

311

health effects literature, as developed in the Introduction.



RUNNING HEAD: Modeling Particle Emissions 312 313

Keeping in mind that the aim is not to provide a precise estimate of a point exposure, we never the less

314

determined the normalized root mean square error (NRMSE) between the model and each of the

315

printers/areas in order to evaluate the model’s accuracy. If the model’s prediction is too large the

316

NRMSE would be greater than 1. All NRMSEs were below 1 (0.39 for the Gantry, 0.44 for the

317

Cincinnati, and 0.41 for the mezzanine).

318 319

As another means of assessing uncertainty we performed a type of sensitivity analysis in the form of

320

variable importance. A function in Analytica™, the software used to run the model and the Monte Carlo

321

trials. The two most important variables (Figure 9) were filament extrusion rate and the eddy diffusion

322

coefficient.

323 324

Figure 9. Model Variable Importance

325

326 327

Modeling provides insight into the range of possible air concentrations pre-exposure, the implications of

328

which are obvious. Using a CPC, concentration maps, generation rates, and simple modeling techniques

329

are useful in hazard assessments for purposes of specifying personal protective equipment a priori,

330

requiring ventilation recommendations, and establishing suitable printer locations with respect to

331

occupied locations. Creating an exposure distribution allows one to view the probability of an exposure 18 ACS Paragon Plus Environment

Page 18 of 23

Page 19 of 23



exceeding a pre-determined value, i.e., an exposure limit when the conditions resulting in the exposure

333

are variable providing advanced information upon which to inform hazard assessment. Additional

334

printing trials at different extrusion temperatures, feed rates, filament sizes and types would aid in

335

providing better bounding for both mass and particle generation rates for future modeling.

336

Statistically there was no difference in the averages of the model prediction and the various printers;

337

however, the statistical agreement between averages fails to tell the whole story. The model by itself

338

would fail to accurately predict a single point determination at some distance from the source in real time

339

with any statistical degree of accuracy. This is because the model, as applied here, uses distributions for

340

the input parameters rather than discrete values. The output from the model was designed to be used as a

341

simulation from which to develop a probability density function (use of the Monte Carlo routine). The

342

utility of this approach is its probabilistic output from which one can develop the likelihood of exceeding

343

some predetermined concentration as part of a hazard assessment.

344

It should also be noted that the model is based on mixing gases. In the case of 3D printing the emitted

345

particulate is well down in the nanoscale region of the aerosol transition regime (molecular mean free

346

path ≈ particle diameter) with Knudsen numbers approaching 10. Thus, for nanoscale particles it is

347

reasonable to treat their movement similar to other molecules forming a field (27).

348

Conclusion

349

Using process filament extrusion rates and temperature specific mass emission rates, the eddy diffusion

350

model coupled with Monte Carlo analysis reasonably estimates ranges of mass and number

351

concentrations encompassing the actual process ambient air concentrations. The SMPS data was used to

352

provide diameters and subsequent mass conversions as health effects literature is primary mass-based.

353

Model input data is easily obtainable, requiring only a means of obtaining the process parameters, a

354

sampler such as a CPC to verify the model performance, and both model and analysis routines are

355

available with existing freeware. Modeling, once developed, can be used to predict air concentrations 19 ACS Paragon Plus Environment



over a range of varying conditions, such as one might employ in a “what if” type of evaluation. Some

357

OSHA standards require formal hazard assessments for work environments where PPE may be needed

358

(29 CFR Part 1910.132). Modeling data may be useful in meeting such a requirement.

359 360

Conflict of Interest

361

There is no conflict of interest to disclose in the study.

362

This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with

363

the U.S. Department of Energy. The United States Government retains and the publisher, by accepting

364

the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-

365

up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or

366

allow others to do so, for United States Government purposes. The Department of Energy will provide

367

public access to these results of federally sponsored research in accordance with the DOE Public Access

368

Plan (http://energy.gov/downloads/doe-public-access-plan).

369

Acknowledgements

370

This research was conducted at the Center for Nanophase Materials Sciences, which is sponsored at Oak

371

Ridge National Laboratory (ORNL) by the Scientific User Facilities Division, Office of Basic Energy

372

Sciences, U.S. Department of Energy. Additionally, this research was supported by the U.S. Department

373

of Energy (DOE) Higher Education Research Experience (HERE) and was administered by the Oak

374

Ridge Institute for Science and Education (ORISE).

375 376

References

377

1. Eigel-Miller, N.; Kline Jr, S. World machine-tool output and consumption survey. [Online] February

378

27, 2014. https://www.gardnerintelligence.com/report/world-machine-tool.


Page 20 of 23

Page 21 of 23



2. McKinsey & Co. Additive manufacturing: A long-term game changer for manufacturers. [Online]

380

McKinsey & Co., September 2017. https://www.mckinsey.com/business-functions/operations/our-

381

insights/additive-manufacturing-a-long-term-game-changer-for-manufacturers.

382

3. Flynn, H.; Hwang, D.; Holman, M. Nanotechnology update: Corporations up their spending as

383

revenues for nano-enabled products increase. UN Environment Depository, 2013.

384

http://hdl.handle.net/20.500.11822/20163

385

4. Oberdorster, G.; Oberdoster, E.; Oberdorster, A. Nanotoxicology: An emerging discipline evolving

386

from studies of ultrafine particles. Environ Health Perspect. 2005, 113 (7), 823-39.

387

5. House, R.; Rajaram, N.; Tarlo, S.M. Case report of asthma associated with 3D printing. Occup Med.

388

2017, 67, 652-54.

389

6. Chan, F.L.; House, R.; Kudla, I.; Rajaram, N.; Tarlo, S.M. Health survey of employees regularly using

390

3D printers. Occ Med. 2018, 68 (3), 211-214.

391

7. Stefaniak, A.B.; LeBouf, R.F.; Duling, M.G.; Yi, J.; Abukabda, A.B.; McBride C.R.; Nurkiewicz,

392

T.R..Inhalation exposure to three-dimensional printer emissions stimulates acute hypertension and

393

microvascular dysfunction. Toxicol Appl. Pharmacol. 2017, 335, 1-5..

394

8. Stephens, P.; Azimi, P.; El Orch, Z.; Ramos,, T. Ultrafine particle emissions from desktop 3D printers.

395

Atmos.Environ. 2013, 79, 334-39.

396

9. Alessandrini, F.; Schutz, H.; Takenaka, S.; Lentner, B.; Karg, E.; Behrendt, E.; Jakob, T. Effects of

397

ultrafine carbon particle inhalation on allergic inflammation of the lung. Allergy Clin. Immunol. 2007, 117

398

(4), 824-30.

399

10. Frampton, M. W.; Utell, M. J.; Wojciech, Z.; Oberdorster, G.; Cox, C.; Huang, L. Effects of exposure

400

to ultrafine carbon particles in healthy subjects and subjects with asthma. Res. Rep. Health Eff. Int. 2004,

401

126, 49-63.

402

11. Stolzel, M.; Breitner, S.; Cyrys, J.; Pitz, M.; Wolke, G.; Kreyling, W.; Heinrich, J.; Wichmann, H-E.;

403

Peters, A. Daily mortality and particulate matter in different size classes in Erfurt, Germany. Journal of

404

Exp. Sci. Environ. Epi., 2007, 17, 458-67. 21 ACS Paragon Plus Environment



12. U.S. Environmental Protection Agency. Particulate matter. 2016. https://www.epa.gov/pm-pollution

406

13. Zontek, T. L.; Ogle, B. R.; Jankovic, J. T.; Hollenbeck, S. M. An exposure assessment of desktop 3D

407

printing. J. Chem. Health Saf., 2017, March/April, 15-25.

408

14. Rutkowski, J. V.; Levint, B. Acrylonitrile-butadiene-styrene copolymers (ABS): Pyrolysis and

409

combustion products and their toxicity - a review of the literature. Fire and Mats. 1986, 10 (3-4), 93-105.

410

15. Azimi, P.; Fazli, T.; Stephens, B. Predicting concentrations of ultrafine particles and volatile organic

411

compounds resulting from desktop 3D printer operation and the impact of potential control strategies. J.

412

Ind.l Eco. 2017, 21 (1), S107-19.

413

16. Keil, C. Patterns of exposure: The power and utility of mathematical models. Synergist. 2017,

414

January.

415

17. Hewett, P. Easy modeling: Forecasting occupational exposures using "well mixed room" models.

416

Synergist, 2017, April.

417

18. Galouchko, V. 3D Field. [Online] 2012. http://3dfmaps.com.

418

19. TSI, Inc. Users guide: Multi-instrument manager (MIM) software for SMPS spectrometers and OPS

419

sizers. TSI, Inc. 2014. www.tsi.com

420

20. Jaycock, M. A major advance in exposure modeling tools. Human health risk assessment to

421

chemicals: The eddy diffusion model is now usable. [Online] December 2013. https://jayjock-

422

associates.blogspot.com/2013/12/the-eddy-diffusion-near-field-model-is.html

423

21. American Conference of Governmental Industrial Hygienists (ACGIH). Industrial ventilation: A

424

manual of recommended practice for design. Cincinnati, OH. 1995.

425

22. Donovan, B. L.; Sahmel, J.; Scott, P. K.; Pauenbach, D. J. Evaluation of bystander exposures to

426

asbestos in occupational settings: A review of the literature and application of a simple eddy diffusion

427

model. Crit. Rev. Toxicol. 2011, 41 (1), 52-74.

428

23. Wallace, L. A.; Emmerich, S. J.; Howard-Reed, C. Source strengths of ultrafine and fine particles due

429

to cooking with a gas stove. Env. Sci. Technol. 2004, 38 (8), 2304-11.


Page 22 of 23

Page 23 of 23



24. Shade, W. D.; Jaycock, M. A. Use of Monte Carlo uncertainty analysis of a diffusion model for the

431

assessment of halogen gas exposure during dosing of brominators. Amer. Ind. Hyg. Assoc. J. 1997, 58 (6),

432

418-24.

433

25. Jaycock, M. A major advance in exposure modeling tools. Human health risk assessment to

434

chemicals: IH Mod 2.0 – A major advance in exposure modeling tools. [Online] September 2018.

435

https://jayjock-associates.blogspot.com/2018/09/ih-mod-20-major-advance-in-exposure.html

436

26. Cheng, K-C.; Acevedo-Bolton, V.; Jiang, R-T.; Klepeis, N. E.; Ott, W. R.; Fringer, O. B.; Hildemann,

437

L. M. Modeling exposure close to air pollution sources in naturally ventilated residences: Association of

438

turbulent diffusion coefficient with air change rate. Environ. Sci. Technol. 2011, 45 (9), 4016-22.

439

27. Madler, L.; Friedlander, S.K. Transport of nanoparticles in gases: Overview and recent advances. Aer.

440

and Air Qual. Res. 2007, 7 (3), 304-342.

441 442


Modeling Particle Emissions from Three ... - ACS Publications

Recommend Documents