Design Considerations for the Attenuation of Translational and Rotational Accelerations in American Football Helmets

Abstract

Participants in American football experience repetitive head impacts that induce negative changes in neurocognitive function over the course of a single season. This study aimed to quantify the transfer function connecting the force input to the measured output acceleration of the helmet system to provide a comparison of the impact attenuation of various modern American football helmets. Impact mitigation varied considerably between helmet models and with location for each helmet model. The current data indicate that helmet mass is a key variable driving force attenuation, however flexible helmet shells, helmet shell cutouts, and more compliant padding can improve energy absorption.

Introduction

The deleterious effects of repetitive head impacts in contact sports [1–7] and the difficulties associated with diagnosing the concomitant neurological abnormalities [8–16] have been chronicled extensively over the last two decades [17–20]. While progress has been made in identifying genetic markers that may make specific individuals more susceptible to head injuries [21–23], new tools for quantifying blood biomarkers have shown promise [24–29], and the potential benefits of some rule changes have been evaluated [3,30], the helmet still serves as the most fundamental piece of safety equipment and the “first line of defense” against mild traumatic brain injury.

Helmets were first required in college football in 1939. They were made of leather and were soon followed by plastic helmets with chin straps [31–35]. By the 1950s, facemasks had been developed and in 1973 the National Operating Committee on Standards for Athletic Equipment introduced a standard for helmet characterization that utilized a drop test-based criterion [36,37]. While drop test-based standards were an important development that eventually led to innovations in cushioning systems, these methods did not possess a well-defined input–output relationship between impact forces and the resulting translational and rotational acceleration profiles. This lack of a transfer function within the National Operating Committee on Standards for Athletic Equipment standard and its successors [38] made it difficult to directly evaluate the helmet design features that provide the best impact absorption. Recently, a set of dimensionless variables were introduced to quantify the dynamic force delivered to the helmet by a modally tuned impulse hammer and related it to the output acceleration profiles, making it possible to develop those transfer functions and characterize the impact mitigation of helmets [39].

A majority of the currently available football helmets utilize a polycarbonate shell although the Vicis Zero 1 (Vicis, Seattle, WA) employs a more flexible thermoplastic and the SG (SG Helmets, Brownsburg, IN) makes use of a composite weave [39]. For cushioning, there are several different padding systems including polymeric foams, structures that buckle at a fixed load, air bladders, and padding integrated with flexible shell elements. It is not uncommon for different padding thicknesses or different padding systems to be used at different locations within the same helmet. Because no single helmet has demonstrated superior reduction of clinical sequelae (e.g., concussion incidence) or consistent energy absorption at all the impact locations [39,40], it is important to catalog those design features that yield the best outcomes. Consequently, the goal of this study was to examine a wide range of modern football helmet brands and models in order to determine those design features that lead to improved impact mitigation. We hypothesized that, because it includes traditional padding elements as well as a flexible outer shell, the Vicis Zero 1 would exhibit the greatest energy absorption.

Theory

Several dimensionless groups were previously developed as a part of the study described by Cummiskey et al. to quantify the mitigative effects of helmets on accelerations experienced by a Hybrid III head form [39] when subjected to impacts from a modally tuned impulse hammer (PCB Piezotronics, Inc., Depew, NY). The acceleration terms, which consist of the peak translational acceleration, a_p, and the peak angular acceleration, ${\ddot{\theta }}_p$ were the primary output parameters in this dimensional analysis. The primary input parameter was the impulse delivered, $\int F\left(t\right)\hspace{0.17em}dt$, where F(t) was the time varying impact force. The normalization of these parameters was performed according to the method developed by Cummiskey et al. [39]. The dimensionless forms of the translational and angular accelerations were given by the variables ${\mathrm{\Pi }}_1$ and ${\mathrm{\Pi }}_2$, respectively,

$${\mathrm{\Pi }}_1=\frac{{a_p{t}^{\mathrm{*}}}^2}{W_n}$$ (1)

$${\mathrm{\Pi }}_2={{\ddot{\theta }}_p{t}^{\mathrm{*}}}^2$$ (2)

where t* is the difference between a reference time (100 ms) and the impact duration as measured by the impulse hammer, and W_n is the width of the neck (in this case the Hybrid III neck). The dimensionless input parameter relevant to this study is given by

$${\pi }_1=\frac{t^{\mathrm{*}}\int F\left(t\right)\hspace{0.17em}dt}{m_h{w}_n}$$ (3)

where m_h is the mass of the head form (not include the helmet). Details of the derivation can be found in Ref. [39].

Methods

Data Collection.

This study was designed to characterize the energy absorption and catalog the design features contributing to enhanced energy absorption of several adult size large helmet models designed for American football from Xenith (Xenith; Detroit, MI): the Xenith Epic+ and X2E+; three from Riddell (Riddell; Rosemont, IL): the Speed Flex, the Speed Icon, and the Speed Classic; one helmet from Vicis (Seattle; WA): the Zero 1; and three helmets from Schutt (Litchfield, IL): the Air XP Pro-VTD II, the Vengeance VTD II and the Vengeance Pro. For each helmet model, three separate helmets were tested for a total of 27 helmets.

The mass of each helmet was measured three times and averaged and then each helmet, in turn, was fit per manufacturer's instructions onto a 50th percentile Hybrid III head and neck assembly testing rig secured to a steel baseplate (Fig. 1) [41]. In order to develop a proper transfer function, both the input force and output accelerations were measured for each impact as described by Cummiskey et al. [39,42]. Impacts were administered to the bare head form using a modally tuned impulse hammer (Model PCB 086D20, PCB Piezotronics, Inc.; Depew, NY) [39,42], to record the applied force [39,42]. Impacts at each location were delivered by hand either normal or oblique. Oblique impacts were administered at an angle of approximately 45 deg to the surface of the helmet. For each helmet, 20 impacts were delivered at seven different locations: (1) front, (2) front-oblique, (3) front boss, (4) front boss-oblique, (5) side, (6) side-oblique, (7) rear boss, (8) rear boss-oblique, (9) rear, (10) rear-oblique, (11) top, (12) top-oblique, and the location of the Speed Flex Cut out, both (13) normal and (14) oblique. They were administered manually and equally divided into five distinct impulse ranges from 2–4 Ns, 5–7 Ns, 8–10 Ns, 11–13 Ns, 14 + Ns (Fig. 1).

Fig. 1.

A modally tuned impulse hammer and a Hybrid III head form (a) with a 3-2-2-2 accelerometer array (b) was used to deliver 14 different types of quantifiable impacts over a total impulse range of 2–16 Ns. The impact locations (c) and types were denoted by (1) front, (2) front-oblique, (3) front boss, (4) front boss-oblique, (5) side, (6) side-oblique, (7) rear boss, (8) rear boss-oblique, (9) rear, (10) rear-oblique, (11) top, (12) top-oblique, and the location of the Speed Flex Cut out, both (13) normal and (14) oblique. Oblique impacts were delivered at an angle of approximately 45 deg to the surface normal. The image in (c) was adapted from United States patents 20130180034A1 and 20120297525A1.

For each impact, model 9234 data acquisition modules (National Instruments, Austin, TX) were used to collect 200 ms time series with a 70 ms pretrigger, providing 130 ms of impact force measurements. All signals collected at 5120 Hz. To measure the resulting accelerations at the center of mass, a nine-accelerometer array in a 3-2-2-2 setup was used (triaxial accelerometer was model 356A03 and the biaxial accelerometers were model 352C22, PCB Piezotronics, Inc.; Depew, NY), as defined by Padgaonkar et al. [40,42,43]. The same 200 ms measurement window was used. Linear accelerations on each axis were measured, and resultant translational and angular accelerations were calculated.

After testing the bare head form, three helmets from each helmet model were fitted according to manufacturer specifications and tested in the same way for a total of 60 points at each location and 840 data points collected per helmet model to correct for any responses that may have been the result of a manufacturing defect in a single device.

Postprocessing.

The data were collected and processed using a custom matlab program that passed the acceleration traces through a low pass Butterworth filter with a cutoff frequency of 750 Hz to reduce noise [39]. Kinematic equations were used to calculate resultant translational and rotational acceleration at the center of mass [42,43]. Those values were output as dimensionless quantities to assess the response of the head form [39]. Peak translational acceleration and peak rotational acceleration were output as trace data for each recorded impact.

Statistical Analysis.

An intermediate asymptotics model was used to relate the output groups ${\mathrm{\Pi }}_1$ and ${\mathrm{\Pi }}_2$ to the input parameters in the form of the following equation [44]:

$${\mathrm{\Pi }}_i=B_i{\pi }_1^{{\beta }_{1i}}$$ (4)

For this analysis, as in the Cummiskey study, both ${\pi }_2$ and ${\pi }_3$ were removed from the final statistical considerations. A modified version of Grubb's method was used to remove outliers and a final curve fit was generated [39]. The final curve fits were generated using a log transformation given by

$${\ln (\mathrm{\Pi }}_i)={\ln (B}_i)+{\beta }_{1i}\mathrm{l}\mathrm{n}\left({\pi }_1\right)$$ (5)

allowing the model coefficients to be determined from a linear regression. The translational and rotational outputs, ${\mathrm{\Pi }}_1$ and ${\mathrm{\Pi }}_2$, for each helmet model were regressed against ${\pi }_1$ and an ANCOVA test with an α level of 0.05 was used to examine differences between the regression coefficients for each helmet. Statistical differences were determined using an ANCOVA test, a Tukey post hoc test with a Holm-Sidak p-value correction was utilized [39,40].

The final intermediate asymptotic curves generated by this process were then utilized to determine the impact mitigation for each helmet. The impact mitigation was simply defined as the area under the ${\mathrm{\Pi }}_1$ versus ${\pi }_1$ curve for each helmet subtracted from the area under the same relation obtained for the bare headform and that difference was normalized to the area under the curve for the bare headform. A total of 100 evenly spaced values spanning the distance between the minimum and maximum values for ${\pi }_1$ were used to calculate the area under each curve. This process was repeated for the rotational accelerations. With a maximum value of one, the higher the impact attenuation, the better the helmet performed.

Results

The mass of the helmets ranged from 1.72 kg (Schutt Vengeance Pro) to 2.10 kg (Vicis Zero 1) with only two others (Schutt Air XP Pro-VTD II and Riddell Speed Flex) exceeding 2 kg. (Table 1).

Table 1.

Mass of each helmet tested and description of as-tested material components

Helmet model	Mass (kg)	Shell material	Facemask material
Schutt Air XP Pro-VTD II	2.02	Polycarbonate	Titanium
Schutt Vengeance VTD II	1.94	Polycarbonate	Titanium
Schutt Vengeance Pro	1.72	Polycarbonate	Carbon steel
Xenith X2E+	1.94	Polycarbonate	Titanium
Xenith Epic+	1.96	Polycarbonate	Carbon steel
Riddell Speed Flex	2.02	Polycarbonate	HS4 steel
Riddell Speed Icon	1.83	Polycarbonate	HS4 steel
Riddell Speed Classic	1.81	Polycarbonate	HS4 steel
Vicis Zero 1	2.10	Thermoplastic elastomer	Titanium

During impact testing, each helmet significantly decreased the regression coefficients relating both the translational and angular acceleration to the impulse delivered using the modally tuned impulse hammer as compared to the bare head form. As expected, the intermediate asymptotic relations modeled the data well (Fig. 2), with R² values consistently above 0.85. One of the few exceptions was the Vicis Zero 1 helmet which appeared to exhibit a bilinear response for half of the impact types. It demonstrated excellent impact attenuation at locations 5–9, 11, and 12 for lower impulses, but markedly less attenuation beyond a dimensionless impulse of 1.

Fig. 2.

For each normal and oblique impact, the force was measured (a) by the modal impulse hammer, and the translational acceleration (b) and angular acceleration (c) were measured using the nine-accelerometer array within the Hybrid III head form

For ${\mathrm{\Pi }}_1$, the impact attenuation ranged from 0.528 (Riddell Speed Icon, location 13) to 0.832 (Vicis Zero 1, location 11) and varied considerably between helmet models and locations within a specific helmet model (Table 2). The Vicis Zero 1 demonstrated the highest impact attenuation at seven of the 14 locations (1–3, 11–14) as well as the highest average attenuation (0.769). The helmet with the greatest minimum translational impact attenuation across all impact locations was the Schutt Air XP Pro-VTD II (0.730).

Table 2.

Attenuation of translational acceleration by nine different helmets. Bold type indicates best performance at each location, highest average value across all helmet models, and highest minimum value across all helmet models.

Location code	Schutt Air XP Pro-VTD II	Schutt Vengeance VTD II	Schutt Vengeance Pro	Xenith X2E+	Xenith Epic+	Riddell Speed Classic	Riddell Speed Flex	Riddell Speed Icon	Vicis Zero 1
1	0.771	0.729	0.730	0.734	0.743	0.679	0.705	0.620	0.832
2	0.765	0.720	0.749	0.744	0.755	0.684	0.744	0.626	0.817
3	0.730	0.706	0.731	0.721	0.780	0.678	0.750	0.626	0.786
4	0.758	0.745	0.744	0.736	0.799	0.703	0.741	0.684	0.759
5	0.769	0.727	0.730	0.727	0.792	0.662	0.804	0.651	0.726
6	0.767	0.744	0.739	0.739	0.783	0.684	0.808	0.692	0.713
7	0.742	0.718	0.632	0.663	0.744	0.597	0.747	0.575	0.698
8	0.762	0.740	0.668	0.704	0.747	0.663	0.779	0.626	0.699
9	0.754	0.730	0.627	0.692	0.675	0.679	0.747	0.652	0.741
10	0.758	0.755	0.625	0.610	0.696	0.666	0.756	0.679	0.715
11	0.761	0.740	0.659	0.699	0.687	0.623	0.701	0.546	0.832
12	0.809	0.764	0.752	0.764	0.769	0.725	0.753	0.671	0.823
13	0.758	0.697	0.676	0.729	0.739	0.665	0.741	0.528	0.817
14	0.787	0.750	0.758	0.740	0.784	0.707	0.767	0.609	0.808
Average	0.764	0.733	0.701	0.714	0.749	0.672	0.753	0.627	0.769
Minimum	0.730	0.697	0.625	0.61	0.675	0.597	0.701	0.528	0.698

One of the primary differences between the Xenith X2E+ and the Epic+ was the addition of vent holes from the front boss to the rear boss locations. Interestingly, there was a notable increase in impact attenuation at these locations (Table 2) for both the normal and oblique impacts.

For the Riddell helmets, the Speed Flex demonstrated a notable improvement in translational impact attenuation over the Speed Classic and Speed Icon. In particular, the side impact locations exhibited more than 21% improvement for normal impacts to the side of the helmet and almost 17% improvement for oblique impacts. The most obvious difference in the construction of the Speed Flex helmet was the use of more compliant padding on the side than either of the other two Riddell helmets tested here.

Of the Schutt helmets, the oldest of the models tested here, the Air XP Pro-VTD II consistently demonstrated better results than either Vengeance model for attenuation of translational accelerations. Aside from the notable decrease in mass, the biggest difference in the helmets were changes in the shapes of the vent holes (from round to angular) and the addition of more complex geometric ridges in the Vengeance models. It is not clear whether those ridges stiffened the structure as a whole or whether the ridges caused a more complicated acceleration profile during impact.

For ${\mathrm{\Pi }}_2$, (Table 3) the impact attenuation ranged from 0.0967 (Vengeance Pro, location 9) to 0.8291 (Xenith Epic+, location 4) and also varied considerably between helmet models and locations within a specific helmet (Fig. 3). The Vicis Zero 1 and the Schutt Air XP Pro-VTD II both performed the best at five of the 14 locations. Of the remaining four locations, the Riddell Speed Flex performed best at two, and the Xenith Epic+ and the Schutt Vengeance VTD II performed best at one location each. The Schutt Air XP Pro-VTD II and the Vicis Zero 1 had the highest average rotational impact mitigation 0.704 and 0.703, respectively, but the helmet with the greatest minimum rotational impact attenuation across all impact locations was the Riddell Speed Flex (0.479).

Table 3.

Attenuation of angular acceleration by nine different helmets. Bold type indicates best performance at each location and highest average value across all helmet models, and highest minimum value across all helmet models.

Location code	Schutt Air XP Pro-VTD II	Schutt Vengeance VTD II	Schutt Vengeance Pro	Xenith X2E+	Xenith Epic+	Riddell Speed Classic	Riddell Speed Flex	Riddell Speed Icon	Vicis Zero 1
1	0.600	0.522	0.160	0.321	0.525	0.525	0.479	0.490	0.698
2	0.715	0.645	0.508	0.547	0.651	0.619	0.550	0.606	0.727
3	0.701	0.604	0.715	0.668	0.763	0.540	0.684	0.537	0.769
4	0.786	0.766	0.789	0.705	0.829	0.675	0.728	0.683	0.773
5	0.815	0.768	0.801	0.801	0.812	0.715	0.811	0.692	0.783
6	0.815	0.788	0.808	0.813	0.809	0.757	0.811	0.721	0.801
7	0.754	0.695	0.594	0.692	0.691	0.680	0.730	0.513	0.698
8	0.572	0.486	0.468	0.513	0.520	0.552	0.578	0.286	0.479
9	0.456	0.329	0.097	0.273	0.140	0.275	0.483	0.316	0.391
10	0.746	0.728	0.600	0.635	0.590	0.629	0.701	0.687	0.693
11	0.604	0.596	0.342	0.442	0.338	0.440	0.564	0.371	0.723
12	0.811	0.729	0.706	0.747	0.717	0.629	0.619	0.592	0.769
13	0.688	0.667	0.485	0.592	0.627	0.620	0.530	0.553	0.755
14	0.798	0.811	0.765	0.763	0.800	0.765	0.685	0.762	0.778
Average	0.704	0.652	0.560	0.608	0.629	0.601	0.639	0.558	0.703
Minimum	0.456	0.329	0.097	0.273	0.140	0.275	0.479	0.286	0.391

Fig. 3.

Representative dimensionless translational ( ${\mathrm{\Pi }}_1$) and rotational ( ${\mathrm{\Pi }}_2$) accelerations for Front impacts (a) and side impacts (b) for the bare head form (solid dots) and the nine helmets tested herein. All helmets substantially reduced ${\mathrm{\Pi }}_1$ at the front and the side locations, but the same was not true for ${\mathrm{\Pi }}_2$. There was large variability in the ability of the helmet to attenuate rotation for impacts to the front, but uniformly better results were obtained at the side location.

The top three helmets in terms of average ${\mathrm{\Pi }}_1$ attenuation were the three models with highest mass and a linear regression of average translational impact attenuation against helmet mass (Fig. 4) yielded an R² value of 0.62. Three of the top four helmets in terms of average rotational impact attenuation were the helmets with most mass tested and a linear regression of average rotational impact attenuation against helmet mass (Fig. 4) yielded an R² value of 0.78.

Fig. 4.

Average translational impact mitigation (filled circles) versus helmet mass and average angular impact attenuation (open circles) versus helmet mass. Both regressions were significant (p values <0.02 and <0.002, respectively).

Discussion

This set of experiments quantified the impulse delivered to the head form-helmet system and the resultant translational and rotational accelerations to provide a translational and rotational transfer function as previously described [40,42,43]. An intermediate asymptotic relationship was utilized to provide curve fitting parameters and generate statistical comparisons [44]. These dimensionless variables yielded measures of translational impact attenuation and rotational impact attenuation for each type of impact as well as average measures for each helmet model. Taken together, these data made it possible to document the impact mitigation for each helmet model and a variety of locations and correlate those data with design features that promote safety in American football.

There are two important strengths of this study. First is the use of a modally tuned impulse hammer to quantify the input load delivered to the helmet. The resulting transfer function provides a wealth of data that can be used to improve the overall design of helmets for contact sports. In this respect, it is similar to the pioneering work that developed the STAR rating system [38,45–47]. The second strength is the comparison to the bare head form, making it possible to define an impact attenuation metric for each impact type, minimizing the effects of directional differences in neck stiffness exhibited by many test head forms.

Adding helmets to the bare head form mitigated at least 52% of the area under the translational acceleration transfer function, ${\mathrm{\Pi }}_1,$ and up to 83% of that area depending on helmet and impact type. Additive to the effects of helmet mass, the flexible shell of the Vicis Zero 1 appeared to be a beneficial design feature as it performed best for seven different impact types. The addition of vents also improved impact attenuation, at least locally, as demonstrated by the results at the Front Boss location for the two Xenith helmets tested herein. Cracking was noted at the corners of the vents after the testing regimen was completed and the benefits clearly did not extend to other parts of the helmet. The Riddell Speed Flex demonstrated that the addition of more compliant padding on the sides of the helmet also had a beneficial effect, which was to be expected from the work-energy theorem. Outside of the Vicis Zero 1, the other helmet models exhibited low impact attenuation at the rear and top of the helmet. The Schutt Air XP Pro-VTD II performed the best at these locations, but each subsequent Schutt model exhibited a decrease in attenuation. Interestingly, they both included a ridge around the back of the helmet that may stiffen the overall structure or generate a more complicated acceleration profile based on the rapidly changing normal to the surface. It should also be noted that the Riddell Speed Flex did not exhibit the best ${\mathrm{\Pi }}_1$ impact attenuation at the Speed Flex location. Even though it incorporated a flexible shell element, it underperformed with respect to the Schutt Air XP Pro-VTD II and the Vicis Zero 1.

Similarly, the addition of helmets mitigated at least 9.7% of the area under the rotational acceleration transfer function, ${\mathrm{\Pi }}_2$, and up to 83% of that area depending on helmet and impact type. The data for the rotational attenuation was considerably more variable than that of the translational attenuation as the precise shape of the helmet and distribution of mass are considerably more important. Some studies suggest the rotational attenuation is the more important design variable [48–51] and should be a primary consideration for the design of helmets. While the Vicis Zero 1 and Schutt Air XP Pro-VTD II performed well in general, the Riddell Speed Flex was the helmet with the highest minimum attenuation of the ${\mathrm{\Pi }}_2$ transfer function across all impact types.

Interestingly, none of the helmets attenuated more than 48% of the rotational motion for impacts delivered to the rear of the helmet. From a design perspective, the rear of the helmet is structurally stiffer and none of the helmet models offset that stiffness with more compliant padding. It is possible that, because most head impacts occur to the front and side [2,4,52,53] that those locations have been the focus of the majority of design innovations or that the Hybrid III head form is not optimally designed to measure rear impacts. It should also be noted that most aggregate measures of helmet performance do not easily expose these particular types of shortcomings, despite the fact that severe impacts to the rear of the helmet are not unusual in quarterbacks and wide receivers. An effort to relate helmet performance to diagnosed concussions demonstrated a positive relationship between helmet performance score and the concussion rate per 10,000 player plays [54], but it should be noted that diagnosed concussions are not a complete characterization of head trauma [4,55,56].

The data obtained here corroborate the findings of Diekfuss et al. [57] who quantified the alteration of white matter microstructure following a season of American football at the high school level. They found that athletes wearing the Vicis Zero 1 demonstrated fewer changes in white matter structure than did athletes wearing the Riddell Speed Flex, the Riddell Speed Classic, and the Riddell Revolution. In addition to the fact that the Vicis Zero 1 is the most massive of all the helmet models considered herein, it exhibits the best impact mitigation across a wide range of locations. Notably, it does not perform as well as others in the rear impact location. It is possible that the participants in the Diekfuss et al. [57] study simply did not sustain many impacts to the rear location. It should be noted, however, that changes in white matter structure are likely multifactorial that may not be captured within this dataset. Future work should elaborate on the distribution of head impacts sustained in practices and games in order to better interpret these data. These results, combined with those of Svaldi et al. [16] and Bari et al. [11] which demonstrated that long-term brain changes correlate best with repetitive impacts greater than 50 g, provide useful design criteria, suggesting that future design efforts may result in a helmet that can mitigate the most deleterious effects of head impacts in sports.

The average translational and rotational attenuations both correlated well with helmet mass. While increased mass of helmets will certainly lead to lower accelerations, there must ultimately be a limit beyond which the athlete will have difficulty turning his or her head and may be unable to react quickly to the play on the field. Lighter helmets are also used in other contact sports such as lacrosse, but those helmets notoriously underperform from an impact attenuation perspective [40,58,59]. To our knowledge, the only football helmet to embrace a low mass design for American Football was the Simpson–Ganassi (SG) helmet with a mass of 1.11 kg [39]. It attempted to offset its lower mass by using a compliant composite shell and padding material that was substantially less stiff than that used in standard football helmets. Unfortunately, that combination did not lead to uniformly better performance and, during testing, considerable damage to the shell and liner were observed. Given that it was almost a kilogram less massive than the Vicis Zero 1, these data suggest that its performance was a noteworthy design. Consequently, a concerted effort to construct a flexible shell and more compliant padding in a more massive helmet, may provide a structure that can effectively eliminate the deleterious effects of head impacts in contact sports.

It is also possible that designers of modern football helmets are missing a unique opportunity. Specifically, a limitation of this study is also a current limitation of the industry. Without detailed measures of the forces delivered to the head in contact sports (as well as their concomitant accelerations), it is difficult to specify which range of ${\pi }_1$ is most relevant to a particular competition level (Pop-Warner, Middle School, High School, College, or Professional). Consequently, we utilized the entire range of experimental values to calculate the impact mitigation. Future work should examine the range of forces and accelerations incurred at each of the aforementioned competition levels in order to determine the possibility of designing helmets for both competition level and possibly position. The fact that the measured accelerations are similar across the competition levels studied so far [60–66] suggests that the forces are much different and there may be an opportunity to dramatically improve player safety, especially for youth athletes. Once the force ranges are better characterized, the shell and padding can be designed to maximize energy absorption and reduce the vast majority of head accelerations below the 50 g threshold. If we further characterize neck strength, it may also be possible to define an ideal mass to neck strength ratio for each competition level. Cumulatively, our data add significantly to the field to guide further development of more advanced concepts such as double shell designs or new energy absorbing materials. A further limitation with laboratory impact simulations such as the one conducted here is that the relationship between the impact force and the resulting acceleration depends on the response of the entire system, including the chosen head form, neck attachment, and base. To ameliorate those issues, the force profile of every impact were measured directly and each helmet was compared to the response of the head form without a helmet, focusing the mitigation value on the helmet design as much as possible.

It should also be noted that laboratory tests alone will not be able to determine what level of mitigation is sufficient protection for a particular athlete or group of athletes. Such quantification can only be achieved by a combination of laboratory tests like those described herein, field-based impact tracking [30,61,67], and longitudinal neurological assessment [4,9,10,13,14,16,24, 52,57,68–72].

Conclusion

The data obtained herein indicate that higher masses, flexible shells, and padding materials targeted to the range of forces at a given competition level may be sufficient to approach 80% impact mitigation for translational accelerations across all helmet locations, but mitigating angular accelerations to that level may require additional innovation. Future design work must incorporate more than a single aggregate measure to avoid helmets with low impact mitigation at specific impact locations and should consider translational and angular accelerations separately.