Professional Documents
Culture Documents
Age Does Not Affect The Material Properties of Expanded Polystyrene Liners in Field-Used Bicycle Helmets
Age Does Not Affect The Material Properties of Expanded Polystyrene Liners in Field-Used Bicycle Helmets
Kroeker
MEA Forensic Engineers & Scientists,
11-11151 Horseshoe Way,
Richmond, BC V7A 4S5, Canada
Age Does Not Affect the Material
Stephanie J. Bonin Properties of Expanded
MEA Forensic Engineers & Scientists,
23281 Vista Grande Drive,
Laguna Hills, CA 92653
Polystyrene Liners in Field-Used
Alyssa L. DeMarco Bicycle Helmets
MEA Forensic Engineers & Scientists,
11-11151 Horseshoe Way, Bicycle helmet foam liners absorb energy during impacts. Our goal was to determine if
Richmond, BC V7A 4S5, Canada the impact attenuation properties of expanded polystyrene (EPS) foam used in bicycle
helmets change with age. Foam cores were extracted from 63 used and unused bicycle
Craig A. Good helmets from ten different models spanning an age range of 2–20 yrs. All cores were
Collision Analysis, impact tested at a bulk strain rate of 195 s1. Six dependent variables were determined
43 Skyline Crescent NE, from the stress–strain curve derived from each impact (yield strain, yield stress, elastic
Calgary, AB T2K 5X2, Canada; modulus, plateau slope, energy at 65% compression, and stress at 65% compression),
Schulich School of Engineering, and a general linear model was used to assess the effect of age on each dependent vari-
University of Calgary, able with density as a covariate. Age did not affect any of the dependent variables; how-
2500 University Drive NW, ever, greater foam density, which varied from 58 to 100 kg/m3, generated significant
Calgary, AB T2N 1N4, Canada increases in all of the dependent variables except for yield strain. Higher density foam
cores also exhibited lower strains at which densification began to occur, tended to stay
Gunter P. Siegmund1 within the plateau region of the stress–strain curve, and were not compressed as much
MEA Forensic Engineers & Scientists, compared with the lower density cores. Based on these data, the impact attenuation
11-11151 Horseshoe Way, properties of EPS foam in field-used bicycle helmets do not degrade with the age.
Richmond, BC V7A 4S5, Canada; [DOI: 10.1115/1.4032804]
School of Kinesiology,
Keywords: bicycle helmet, expanded polystyrene foam, impact, age, compression,
University of British Columbia,
material properties
210-6081 University Boulevard,
Vancouver, BC V6T 1Z1, Canada
e-mail: gunter.siegmund@meaforensic.com
1 Introduction densification phase [8–10]. During the linear elastic phase, cell
walls bend or stretch and the stress increases at a near constant
Bicycle helmet foam liners absorb energy during impacts. They
rate with strain. In the plateau phase, the cell walls collapse and
also distribute the impact force over a larger area of the head and
stress in the foam remains relatively constant while the foam con-
reduce head accelerations compared to a bare head impact. While
tinues to compress. During the densification phase, opposing cell
new helmets offered for sale to the public are required to meet
walls come into contact and the stress rises quickly with relatively
minimum impact attenuation standards [1–3], there are no regula-
small increases in strain. The majority of the impact attenuation
tions governing their impact performance over time. Helmet man-
benefits provided by helmets is in the plateau region where the
ufacturers recommend replacing a helmet after an impact or if
stress, and hence the force applied to the head, remains relatively
there is visible damage (e.g., crushed or cracked liner, or a
constant while the foam collapses.
cracked outer shell), but some manufacturers also provide specific
The impact properties of closed cell foams have been exten-
recommendations in their product manuals to replace a helmet
sively studied, including the effects of strain rate and density. For
every 2–10 yrs regardless of impacts or visible damage (e.g.,
increasing strain rates, the yield stress increases, the elastic modu-
Bern—2 yrs; Giro, CCM, Lazer, Supercycle—3 yrs; Specialized,
lus increases, the plateau region becomes narrower, and the strain
Alpina—5 yrs; and Nutcase—10 yrs). The Snell Memorial Foun-
at which densification begins to occur decreases [10–13]. A simi-
dation also recommends that helmets be replaced after 5 yrs of
lar change in behavior occurs as density is increased [8,9,11,13].
use, or less if recommended by the manufacturer [4]. The scien-
The material properties of foam are also affected by high heat and
tific basis for this replacement period is unclear, and it remains
high humidity [5]. Understanding how a variety of conditions and
unknown if helmet impact performance changes with age.
material property changes affect the impact response of foam pro-
Foam liners in bicycle helmets are generally made of EPS, a
vides an opportunity to optimize the foam properties and improve
closed cell polymeric foam [5–7]. In compression, polymeric
the benefit provided by a helmet foam liner.
foams exhibit a typical behavior that can be described by three de-
When used in helmets, EPS foam liners are exposed to heat,
formation phases: a linear elastic phase, a plateau phase, and a
humidity, sweat, oxygen, ultraviolet (UV) light, and other factors
that can degrade their performance over time [14–16]. Degrada-
1
Corresponding author.
tion from one or more of these factors can decrease the yield stress
Manuscript received November 8, 2015; final manuscript received February 4, and yield strain, and increase the elastic modulus of polymeric
2016; published online March 3, 2016. Assoc. Editor: Barclay Morrison. foams [17]. Few studies have focused on the effect age has on the
Year of manufacture
Make (models) Sizes Cert 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Traditional
Giro (Torrent, Venus) 1S, 4M 1 x x x x x
Bell (Adrenaline, Blade, 1S, 5M 1 x x x x x x
Belisima)
Oro, CCM, XTRAZ 3S, 2M, 1, 2 x x x x x x
(Ice, unknown) 1L
Bell (Avalanche, 4M, 1L 1, 2 x x x x x
Oasis Pro),
unknown (Tango II)
Bell Amigo 1XS, 4S 1 x x x x x
Kids Visor
LG (Oro, Terrible X, 4XS, 6S 1 x x x x x x x x x x
unknown)
LG (Terrible, unknown) 3XS, 3S 1 x x x x x x
Giro (Rodeo, ME2) 2XS, 5S 1 x x x x x x x
BMX
Bell (Barbie, Rumor, 2XS, 3S, 1 x x x x x x
Psycho, Headspace) 1M
Bell Faction 1S, 6M 1 x x x x x x x
Fig. 1 Helmet types tested in this study (left to right: Traditional, Kids Visor, and BMX)
impact attenuation properties of EPS bicycle helmet foam liners. following criteria: (i) legible labels specifying their date of manu-
Age is briefly mentioned in Ref. [9], where they reported that the facture, (ii) no evidence of impact-related damage, and (iii) certi-
value of the elastic modulus “…is rarely known with precision for fied to a helmet standard. Helmets were then sorted into matched
polymer foams because it depends on the degree of polymer-chain sets, which were defined as helmets of essentially identical design
alignment, on chemical changes brought about by the foaming and construction, though of potentially different makes, models,
agent and on the gradual ageing and oxidation of the polymer.” colors, and sizes. Only matched sets that contained 5 different
Pearsall et al. [18,19] studied age effects directly and observed manufacturing years were selected for testing here. This selection
increased peak headform accelerations in 10-yr-old hockey hel- process resulted in 63 helmets from ten matched sets with helmets
mets, but not in 6 - or 2-yr-old helmets; however, they tested manufactured between 1994 and 2012 (Table 1). Five matched
multi-impact helmets constructed of expanded polypropylene and sets were traditional-style bike helmets, three matched sets were
vinyl nitrile foam that is different from the EPS used for single- kids helmets with built-in EPS visors, and two matched sets were
impact bicycle helmets. Based on these limited data, it remains BMX-style helmets (not full-face) (Fig. 1). All helmet liners
unclear whether age affects the impact attenuation properties of were composed of EPS and no cores appeared to be made of dual-
bicycle helmets. density foam. Average polystyrene bead sizes were about
The goal of this study was to determine if the impact attenua- 1.5–2 mm, which resulted in about 9–12 intact beads across the
tion properties of EPS foam in bicycle helmets change with age. diameter of each core sample (Fig. 2).
Based on manufacturer recommendations that bicycle helmets be All helmets had been previously drop tested as part of a larger
replaced every 2–10 yrs, we hypothesized that the old helmet study using a 5 kg trolley/headform mass onto a flat anvil at
foam would exhibit poorer impact attenuation properties com- both the left and right front regions, at 3.0 m/s on one side and at
pared with new helmet foam. 6.2 m/s on the opposite side. Damage from these impact tests
typically included cracked shells and/or cracked/compressed foam
liners and was greatest near the location of impact. All foam cores
2 Materials and Methods tested here were taken from the undamaged occipital region of the
Cylindrical foam cores (19 mm diameter 19 mm height) were helmets.
extracted from 63 new and used helmets drawn from a pool Core size and extraction location were limited by helmet design
of >1500 helmets donated by the public. Potential test helmets and vent location. All foam cores from helmets within a matched
were first selected from the donated pool of helmets using the set were taken from the same location, but this location varied
Impact onset was defined as the first digitized sample where the 0.05 (a)
force data were one standard deviation greater than the no-load
condition and remained above that level. Accelerometer data were yij ¼ b00 þ b10 densityij þ b20 ageij
double integrated from initial contact to obtain compression data. X
N
This method was validated against high-speed video acquired at þ ðb0j þ b2j ageij Þ þ eij (2)
100,000 Hz, and the peak strain was found to be within j¼1
2.2 6 0.1% of the video-based strain measurements. Engineering
stress (r) and strain (e) were calculated according to the following
equations:
3 Results
F DL The pre-impact density of the foam cores ranged from 58 to
r¼ ; e¼ (1)
A Li 100 kg/m3. The Bell BMX-style helmets (e.g., Barbie, Rumor,
Psycho, and Headspace) had the lowest average density
where F is the force, A is the initial contact area, Li is the initial (62.7 6 2.6 kg/m3), and the Giro traditional-style helmets (e.g.,
core height, and DL is the change in core height. All data were Torrent and Venus) had the highest average density
(90.7 6 10.2 kg/m3). The range in densities for the BMX helmets 4 Discussion
was 58–82 kg/m3, for the kids visor helmets was 64–88 kg/m3,
and for the traditional helmets was 60–100 kg/m3. Our goal was to determine if the impact attenuation properties
The foam cores were compressed in about 6 ms with a rapid of bicycle helmet foam changed with age. Based on our tests of
increase in force and acceleration at maximum compression foam cores extracted from ten matched sets of 63 helmets
(Fig. 4). The bulk strain rate at impact was 195.2 6 0.2 s1 between 2 and 20 yrs old, age does not affect the dynamic me-
(impact speed was 3.73 6 0.04 m/s). Maximum foam core com- chanical properties of the EPS foam liners used in bicycle hel-
pression ranged from 73% to 96% of the initial height. All foam mets. As a result, we reject our hypothesis that age affects the
cores exhibited the typical stress–strain behavior for foam (Fig. 5 impact attenuation properties of the EPS liners used in bicycle
and Table 2). No significant associations were observed between helmets. Based solely on impact attenuation properties, we find
helmet age and any of the dependent variables within matched no support for replacing a bicycle helmet every 2 or more years
sets or among the whole population (Fig. 6 and Table 3). Age was as suggested by some manufacturers. Despite these findings, we
also not related to density; however, density had a significant recognize that other helmet components, i.e., straps, buckles,
effect on all of the dependent variables (p < 0.001), except for shell, etc., also affect a helmet’s overall performance and we
yield strain (p ¼ 0.29) (Fig. 7 and Table 3). When ranked by foam did not test these components to assess how age affects their
density, the stress–strain curves revealed that the lower density performance. Nevertheless, our findings indicate that consumers
foam cores displayed higher degrees of densification and wider should not discard or avoid wearing an old helmet simply
plateaus (Fig. 8). because it is old.
Foam density had a large effect on the mechanical properties of we saw that higher density foams generated higher initial acceler-
the foam we tested and similar behavior with density has been ations but lower peak accelerations, whereas lower density foams
reported previously [8,9,11,13]. Increased density resulted in an generated slightly lower initial accelerations but much higher
increased elastic modulus, increased yield stress, and decreased peak accelerations (see thin gray lines in Fig. 9). This relationship
strain at the onset of densification. The range of our foam densities between density and acceleration highlights the danger of letting
and the mechanical properties we observed during our dynamic the foam reach densification and also suggests that helmet design-
tests were similar to those reported previously by others (Table 4), ers might be able to use foam density to tune a helmet’s response.
and therefore suggests that our findings are not likely an artifact For instance, using low-density foam could potentially minimize
of our sample or our sampling method. the peak head acceleration for impacts well below the level used
The density-related differences we saw in the foam’s mechani- in helmet certification standards at the price of potentially rapid
cal properties manifested as very different peak accelerations densification in more severe impacts. Alternatively, use of high-
during our tests. For the 9.03 6 0.18 J of energy we applied to density foam could potentially avoid densification in impacts
each foam core, higher density foams generated lower peak accel- above certification levels, but expose users to slightly higher peak
erations (see thick solid line at the top of Fig. 9). When we head accelerations in lower severity impacts. Helmet design, how-
extracted the instantaneous acceleration after specific increments ever, is complicated by foam thickness and foam area (e.g., vents
of absorbed energy (e.g., 1, 3, 5, 6, 7, and 8 J) during the impact, and coverage), which affect the amount of foam recruited and
thus the amount of energy absorbed during different kinds of
impacts.
We extracted and tested only one foam core sample from each
helmet and assumed that it reflected the overall density and
Table 2 Material properties of the foam core samples response of the helmet. To evaluate this assumption we extracted
ten cores from a Giro Rodeo helmet (included in our sample) and
Dependent variable Units Mean 6 SD 16 cores from a Supercycle helmet (not in our sample) and found
3 that foam density varied by 6 5.7 and 6 4.3 kg/m3, respectively
Density (p) kg/m 72.7 6 9.1
Elastic modulus (E) MPa 14.0 6 3.5 [26]. Based on this finding, the density obtained from a single
Yield stress (ry) MPa 0.8 6 0.2 sample may not accurately represent the density throughout the
Yield strain (ey) 0.08 6 0.01 entire helmet, and therefore, several samples from the same
Plateau slope (mp) MPa 1.7 6 0.3 helmet might better describe the overall response of a particular
At 65% compression helmet’s foam. Since here we measured the density of the sample
Energy absorbed (E65) J 4.3 6 0.9 we tested, we do not expect that this within-helmet variation in
Stress (r65) MPa 1.9 6 0.4 density would change our findings with respect to the absence of
age effects.
Table 3 Statistical results (p-values) from the general linear model for each material property’s relationship to density (coefficient
b10) and age for the entire population (coefficient b20) and within the matched sets (coefficient b2j). The p-values represent whether
the coefficient was significantly different from 0, i.e., no effect.
To obtain foam cores with similar heights, the outermost layers eliminated by our foam core preparation methods. If present, we
of polystyrene beads were trimmed from each end of the cylindri- expect that the influence of these surface effects would be small
cal cores. The surface of the foam helmet liner contains cells that given that we compressed the foam cores by 73–90% of their orig-
are flattened during the molding process and are more exposed to inal length.
the environment than the inner cells we tested. If material prop- We chose to test field-used helmets rather than new helmets
erty changes related to age are concentrated on the inner and outer exposed to accelerated aging (e.g., ISO 2440 or ASTM
surfaces of the liner, then these age effects may have been D3045). With field-used helmets, we knew that the aging
processes (e.g., heat, humidity, salt, oxygen, UV light, and scission that may lead to an embrittled surface and a variation in
time) experienced by each helmet represented a real combina- the material properties along the depth of the foam [15,16].
tion of these processes, although we did not know what combi- Changes in the material properties may include a decrease in the
nation of these processes a particular helmet had experienced yield strength and yield strain, and an increase in the elastic mod-
and thus could not quantify the variability in these processes ulus [17]. These exposure effects vary with the levels and time of
within our sample. With accelerated aging of new helmets, we exposure and can in some cases be mitigated. For example, stabil-
would have been able to better control each process and reduce izers in the foam can help reduce the effects of UV radiation and
the variability between helmets, but different protocols would temperature [16]. In addition to not knowing the exposure history
be needed for different aging processes, and we could not be for each helmet, we do not know what design or manufacturing
certain that the sequence and combination of aging protocols changes occurred over the range of manufacturing years we
we selected captured the spectrum of these effects that existed tested. We attempted to minimize design changes by grouping
in the field. The larger variability due to these aging processes helmets into similar matched sets and specifically controlling for
in our field sample could have masked an age effect that might changes in foam density in our analysis. Nevertheless, the manu-
be detectable in a sample exposed to controlled accelerated facturing history of our sample remains unknown.
aging; however, the dominant effect of density on our mechan- In summary, we tested EPS foam cores extracted from 63 hel-
ical property measures suggests that any age related effects are mets grouped into ten matched sets spanning 20 yrs and found
likely small by comparison. that the age of the sample did not influence the mechanical proper-
Environmental and use factors that may be expected to influ- ties related to its impact attenuation performance. Instead, we
ence the material properties of polymers include UV radiation, found that the density of the foam core significantly affected all of
exposure to fluids and chemicals, changes in temperature, and a the mechanical properties except the yield strain. Based on these
history of impacts. While we excluded helmets that had evidence data, old helmets should not be discarded based on an incorrect
of impact-related damage, exposure to these other factors remains assumption that their impact performance has deteriorated with
unknown, but are potential causes of cross-linking or chain time.
Fig. 9 The peak acceleration (1) during impact decreased as the density of the foam core
increased (shown by the exponential curve fit—thick solid line). As the foam core absorbed
increasing amounts of energy, the more dense specimens had smaller increases in accelera-
tion. This is depicted by the instantaneous acceleration corresponding to different levels of
energy absorption: 1J (w), 3J (䊊), 5J (䉫), 6J (䉬), 7J ($), and 8J (N).