Jim Sundahl's Letter to CPSC
on its Bicycle
Summary: A 1998 letter from Bell engineer Jim Sundahl pleading for testing with lighter headforms for child sizes. The ASTM F1892 standard adopted his solution, but CPSC did not, and child helmets are still tested with headforms that are too heavy to be realistic. The result is stiffer helmets than necessary for toddlers.
6350 San Ignacio
San Jose, CA 95119
U.S. Consumer Product Safety Commission
c/o Scott Heh
Directorate for Engineering Sciences
Washington, DC 20207
Dear Sirs and Madams
We are very happy that the "Safety Standard for Bicycle Helmets"
is ready to be finalized and approved. It has been a long and
difficult process. Scott Heh and his staff should be congratulated for
the effort they applied along the way. We also thank the commission for
providing a process that involved industry and consumer advocate groups.
Moreover, the ASTM bicycle helmet task group, which I chair, has been
allowed to be influential in this process. The ASTM group is comprised of
industry, independent test lab people, medical people, consumer advocates, a lawyer
and, of course, Scott Heh. We hope that this new standard is implemented
as quickly as possible.
However, there is one change in this last draft standard that
we strongly oppose: The change of test headform mass to 5 kg for
infants/toddlers. I have been a strong advocate of lower headform
mass for years and feel that I have more information than is indicated
in tab D of the briefing package. Moreover, the ASTM standard
for infants and toddlers that I drafted would have been in effect
at least one year ago if not for administrative over sight at ASTM. It
is now approved and going forward with a mass of 3.2 kg for the A size
headform and 4.0 kg for the E size headform. We have no field experience
with helmets designed to these weights under the ASTM standard because
of this delay.
We do have other field experience. The Department of Transportation,
DOT, safety standard for motorcycle helmets, F.M.V.S.S. 218, has
used 3.5 kg for the small headform for adults or children, for
many years. Every study of the effectiveness of this standard has
corroborated it. In no instance has evidence been raised that
the small headform causes helmets that are too soft. The liners
in motorcycle helmets are typically in the 2.4 to 3.0 pound per
cubic feet density range whereas liners for bicycle helmets, adult
or infant, are typically 5.0 and up.
Another source of field experience is our experience with damaged
helmets returned to customer service. We pioneered infant/toddler
bicycle helmets beginning in the early '80's. We developed the
first Lil Bell Shell in the absence of bicycle helmet standards.
We followed our intuition, experience and test data. We pushed
ourselves up to 4# density just to make the helmets sturdier and
more dent resistant in handling. We didn't think that was too
high. Since then we have sold hundreds of thousands of infant/toddler
helmets. At times standards and design details have
forced us as high as 5.75#. We now run at 5# for all infant model
helmets. In all this time, with all these models, we have never
seen an infant toddler helmet that was anywhere near bottoming
out. Moreover, I collected damaged infant/toddler helmets for
several months in 1995. Not only did I not see bottomed out helmets,
I didn't see any helmet showing signs of crushing on the inside.
This poses the question of whether the helmets are stiffer
than infant heads or do infants just not hit that hard. The evidence
is that most of the time infants don't hit all that hard. But
the evidence also indicates that bottoming out is not a risk for
Now I want to offer some common sense and basic physics. First,
energy management is often discussed regarding helmet standards.
This is a false concept. No helmet standard in the world even
measures energy management or absorption
nor have a pass/fail criteria for energy
management. A helmet can absorb zero energy and still pass any
helmet standard in the world. Energy absorption is a function
of input velocity minus rebound velocity. No standard requires
a laboratory to even measure rebound velocity never mind dictating
that the coefficient of restitution be less than 0.5 or something.
A helmet can rebound with the full input velocity and pass quite
well. Moreover, it can be imagined that any number of liner materials
could absorb energy better than contemporary helmet liners but
in fact produce a very poor helmet. A couple of good energy managers
are soft lead sheet and modeling clay. Impacting either of these
produces negligible rebound velocity. In other words, they absorb
virtually all of the impact energy. None of us are advocating
these materials for helmet liners because energy absorption is
not very important for helmets. I think that any discussion of
helmet test criteria that includes the word "energy" is suspect
and might be misleading.
Acceleration management is what helmets are about. All helmet
standards measure acceleration and enforce a pass/fail criteria
that includes a maximum acceleration rate. Some standards measure
other aspects of the acceleration/time event. This acceleration/time
event is caused by an initial velocity between a head/helmet and
an anvil. The higher the initial velocity the more distance, thickness
of liner, is required to control the acceleration/time curve to a given
set of parameters. The mass of a test headform has no effect upon this thickness.
The mass of the headform does determine the stiffness, usually
the density, of the helmet liner in accordance with f=ma. In the
case of the small A size headform weighing the same as the medium
adult headform, the helmet liner will need to be 30% stiffer in
the infant helmet simply because the contact area of the small
headform is only 77% of the area of the medium headform.[Footnote 1] Thus
with all else equal this makes an infant helmet liner stiffer
than an adult helmet liner.
The average newborn baby weighs about 7 lb.s and cannot have an
11 lb. head. It is obvious that small baby heads weigh less than
their heads will weigh as adults. So let's suppose that A size
infant heads actually weigh the 3.2 kg that I recommend. Now let's
impact this head mass with the helmet designed for the 5 kg A
size headform which is already 30% stiffer than an adult helmet.
Substituting 3.2 for 5 in f=ma, with all else equal, indicates
that the observed acceleration with the real infant headform weight
is 56% higher than in the case with the falsely heavy 5 kg headform.
The 5 kg headform that produces say 250 g's in a laboratory test
would produce nearly 400 g's in an identical impact in the real
world given the weight of real baby's heads. Substituting in f=ma
a liner resistance in kN:
12.3 kN = 5.0 kg * 250 g * 9.80665 m/s^-2/g
12.3 kN = 3.2 kg * 391 g * 9.80665 m/s^-2/g
7.8 kN = 3.2 kg * 250 g * 9.80665 m/s^-2/g
Clearly the helmet developed around a 3.2 kg headform will produce
lower acceleration rates for real world accidents. Any valid
argument in favor of 5 kg headforms would be even more valid for
10 or 20 kg headforms. If real infants have 3 kg heads but we
should test with 5 kg headforms, we should test adult helmets for
5 kg adult heads with 8 kg headforms. In fact, a 50 kg headform for testing would
lead helmet designers to develop helmets that could "absorb
far more energy before bottoming out."
123 kN = 50 kg * 250 g * 9.80665 m/s^-2/g
123 kN = 3.2 kg * 3906 g * 9.80665 m/s^-2/g
This gross over simplification ignores the fact that the light
headform would not crush the liner to a point that far into the
spring rate. But it is obvious that such a helmet would provide
unsuitable acceleration rates for real children. Actual tests that
we have done and math models that we and other have tried show a
small and reasonable change over the small and reasonable ranges that we tested.
I propose that test headforms should be as close as possible to
the average weight of real human heads so that we can properly
control and estimate acceleration rates in the real world and not just in the
We thank you for your consideration of this matter.
Jim G. Sundahl
p.s.: Paragraph 1203.5, Construction Requirements - projections.
The last sentence mentions "fixture" an undefined term.
Please clarify this in the final draft.
The A,E,J,M & O test headforms are "photographically" scaled. Their relative
geometry is as follows:
Size A Size E Size J Size M Size O
Circumference 50 54 57 60 62
rel. area 0.77 0.9 1 1.11 1.18
rel. volume 0.68 0.85 1 1.17 1.29