Use of Thrust Study: Protective Equipment Standards Revisited (Part 1)

In a series of earlier posts, I talked a bit about equipment safety standards and how the homologation rules used by the FIE could be useful in constructing a set of HES-specific equipment safety guidelines. However, much of that was very high level as I had no data to back up some of my claims and could not work through some comparisons with the FIE standards. In my posts on effective bending stiffness and blade buckling, I used FIE weapons rules to provide some comparisons with HES trainers but did not discuss their relevance to safety equipment in much detail.

So now that we have some data on thrusts, and that is the major focus of FIE homologation, let's use the data to see what we can learn! In this part, I'll focus on neck injuries and some mask constraints based on the thrust data.

Standards Revisited

Because modern sport fencing masks are so widely used, let's revisit the homologation rules used by the FIE (CEN EN13567, level 2: Protective clothing - Hand, arm, chest, abdomen, leg, genital and face protectors for fencers - Requirements and test methods). There are four important points to bear in mind:

• Masks, protective garments and weapons can be sold which are not FIE homologated, and therefore cannot be guaranteed to meet the FIE standard. There is a formal minimum in the EU, EN13567 level 1, which requires much lower force levels and impact energies than the level 2 used by the FIE.
• USA Fencing does not require fencing masks, protective garments or weapons to be FIE homologated for domestic competition or day-to-day training. In fact, short of the 12 kg punch test for masks, it's hard to find what they do require of the garments.
• Homologated masks are tested in an un-modified manner. The FIE does not have formal standards for back of the head protection or coach's mask protectors, as these are intended to be additional devices for non-competition situations.
• Unless there's a published standard with a legal backing and corresponding labeling requirements, you're taking the manufacturer's word about any safety claims.

So what does an FIE label means on a garment (Jacket, knickers/pants and plastrons)?

• High speed penetration resistance > 800 N, as measured by a force transducer attached to a 3.0 mm square cross-section penetrator with a point having an included angle of 120\(^o\), and a speed at impact of 6-8 m/s. For drop-rig systems, the falling mass must be greater than 5 kg.
• Seam burst strength > 300 kPa, as measured via hydraulic or pneumatic textile burst tester. This involves a  diaphragm whose pressure can be controlled, pressing against a specimen whose edges are clamped down until the seam or fabric is ruptured.

The above values correspond to EN13567 level 2. For level 1, the high speed penetration force requirement is 350 N and the seam burst strength is unchanged.

For masks, there are quite a few more relevant requirements:

• Mechanical testing of mesh (size of openings: < 2 mm square in flat regions and < 2.1 mm square in curved regions) & materials characterization (bending and tensile testing, microstructural characterization)
• Vertical pull test to ensure that mask cannot be pulled off of a human head model with a 20 N vertical load
• Horizontal push test to ensure that the mask face cannot be brought into contact with the wearer's face with a 50 N horizontal load delivered at eye-level (approximately the center of gravity of the head) on the front of the mask.
• Bib attachment testing (tension on the bib center, pull down and toward the back from the `chin' of the mask at 45 degrees to the horizontal) with a ramped load of 0-800 N over 20-60 seconds, holding at 800 N for 5 seconds.
• High velocity penetration resistance for the bib >  1600 N, as measured by a force transducer attached to a 3.0 mm square cross-section penetrator with a point having an included angle of 120\(^o\), and a speed at impact of 6-8 m/s. For drop-rig systems, the falling mass must be greater than 5 kg.
• Low-velocity (quasi-static) penetration test for the hard-shell (the rigid protection), performed in at least 6 specific locations (more if the hardshell includes a visor of a separate material). Penetration resistance > 1000 N, permanent deformation < 10 mm
• Impact test for the hard-shell, that uses a guided mass of 1.5 kg (using the same penetrator as earlier) falling from a sufficient height to deliver 8.5 J at impact. The test is performed at a number of specific locations, similar to the low-velocity test. Penetration resistance > 1000 N.

The above values are for EN13567 level 2. For level 1 the main differences are the following:

• Bib high velocity penetration test requiring only 350 N
• Hard shell low velocity penetration requires only 600 N and impact test only requires 5.5 J.
• Lateral protection impact test requires only 4.75 J.

The penetrator size specified for the penetration and impact tests is a cross-section that is actually not the same as any of the FIE weapons. For all of the FIE weapons (foil, sabre and epee), the minimum dimensions of the blade occur just behind the point. For sabres and foils, the blade cross-section in this region (as in the majority of the blade) is rectangular, with minimum dimensions of 1.4 mm x 3.6 mm for foils and 1.2 mm x 4.0 mm for sabres. However, about 1/3 of the distance from the point (noted in online forums as being the common breaking point for foils), foils get closer to 2.0 x 4.0 mm. Epees have a hollow-ground triangular form or a Y-shape with a minimum width of 2.4 mm.

As an aside, I tried a few months back to contact the US representative that sits on the FIE's SEMI commission (the folks responsible for equipment homologation) to see if I can get an explanation for the selection of the penetrator and the particular loads. Unfortunately, I never received a response. Maybe writing back as the author of this blog will help, or someone with the right contacts will hook me up (if so: hint, hint). I'll update this whole discussion if I learn more. I suspect that the 3 mm square penetrator was chosen because it is an easy to make approximation to an average weapon cross-section so that a single penetrator could be used. That or it was readily available during early testing, and they kept it. I'll get more into the importance of the penetrator in the next part of this series.

Applying Thrust Calculations and Measurements

Moving on, if we assume the above as the basic design requirements, then we still need quantitative data to see how these line up with some characteristic values for HES. Luckily, I already had some tests in the oven, so let's go ahead and use that information to see what we can find out!

For ease of comparison, I've reproduced the buckling load tables from my previous post below:

In my thrust experiments (from the same post), I found peaks that are above both the pinned/pinned and guided/pinned critical loads, though the majority are between these two loads. This is likely due to the following things:

• The real end conditions are likely somewhere between a pinned/pinned and guided/pinned support
• The critical load is the onset of transverse deformation, but it is possible that higher loads can be momentarily supported during dynamic loading
• The variations in motion during the thrust may lead to transverse loading on the blade, lowering the axial load required to buckle the blade.

However, as the low-force plateau region is typically above the guided/pinned critical load, it is less likely that this deformation mode was dominant mode during the test. So we find that the pinned/pinned condition is probably the best simple model for the onset of blade flex, with the caveat that due to dynamics and variation the significantly higher load predicted by the clamped/pinned end conditions is probably better when a conservative estimate is needed. For example: in equipment design, as we're considering here.

Face Contact Loads

To start, lets look at the FIE requirements for face-contact. This is one of the less stringent requirements on FIE homologated masks, and is the load that can be supported without bringing the front portion of the mask mesh into contact with the face. The EN13567 level 2 requirement is that a 50 N force "gently applied" horizontally at approximately eye-level on the front of the mask hard shell does not result in any part of the mesh contacting the wearer's face. For this test, it is noted that the subject's head is supported to prevent movement. Presumably the force is applied with a device which spreads the load across some area of the mesh, but it is not stipulated exactly how the load should be applied.

If we assume that the mask hard-shell remains undeformed under the 50 N load and the head does not move, we can consider the situation from a purely statics perspective and do a direct comparison of applied forces. An interesting finding based on my analysis of the buckling loads of FIE homologated weapons is that for the most constraining end condition (clamped/pinned), the buckling loads are two to four times higher than the 50 N requirement. Only for the least constraining end condition (guided/pinned) are the loads less than the face contact load requirement, with the exception of the lower end of stiffnesses for FIE foils, which approach the face contact load under the pinned/pinned end condition.

Moving to the HES trainers, we can see from the table in the last section that even under the least constraining end conditions (guided/pinned), the majority of the HES trainers have critical buckling loads well above the 50 N requirement. The exceptions to this trend are the Del Tin rapier blade and the Darkwood-Hanwei rondel trainer, whose lowest critical buckling loads fall below the face contact load requirement. From a comparative standpoint, if the HES community were to want an equivalent face contact load based on the trainers studied here, the face contact force requirement would need to be closer to 200 N (discounting the Swordcrafts longsword), or roughly double the average buckling load for the guided/pinned case. So while it would be possible for the mask to be brought into contact with the face under the more extreme conditions (clamped/pinned), the lower bound of buckling loads would be well below the face contact requirement.

Head Injury Criteria-based Analysis

Another interesting question to ask is: if we increase the face-contact load due to the stiffness of HEMA trainers, how does the injury risk compare to FIE weapons? Thanks to relatively recent work in the biomechanics arena, correlations of head/neck injuries (concussion,  fractures, etc.) to impact are easily applied to our current topic. Given the importance sports such as boxing and football have placed on understanding the occurrence of concussions and head/neck injury it may also make for an interesting motivation for a more rigorous study in historical fencing.

Schematic of the skull and neck to provide reference points for linear acceleration of the head, as well as flexion and extension moments for the neck. Skull and spine image courtesy of Patrick J. Lynch, medical illustrator; C. Carl Jaffe, MD, cardiologist via Wikimedia Commons.

If we assume the load is applied as a pulse of some known duration and profile, and that the load will simply cause the head to move while the body remains stationary (and the mask doesn't absorb energy in any way), we can make use of the Head Injury Criterion (HIC). This criterion correlates the linear acceleration of the head (see figure above) over a given period of time to the probability of injury. It is used in vehicle crash test standards (such as FMVSS-208), as well as sports injury research. The typical form of this model is as follows:
\[HIC = \left[\frac{1}{t_2-t_1}\int_{t_1}^{t_2}{ a(t)\;dt}\right]^{2.5} (t_2-t_1)\]
Where \(t\) is time in seconds, and \(a\) is the acceleration in units of gravitational acceleration ('g').

With the above equation, we can perform a fairly simple calculation to get an estimate for the HIC for different applied loads. This means we can ask if the increased face contact load for HES will result in a significant change in head injury probability over the 50 N EN13567 Level 2 value!

Let's assume a dynamic force application which has a total duration of 15 ms (based on the boxing example in [1]), a peak force of 50 N (the EN13567 Level 2 face contact load) and a triangular pulse form (rises from zero to peak and down again at a constant rate), and only moves the head straight back. The head and neck assembly for a Hybrid III dummy (meant to mimic a 50th percentile adult male) weighs approximately 4.38 kg [1] and the center of gravity is approximately at eye-level (roughly where the face contact load is applied in testing). We can neglect the mass of the mask to give a higher estimate of acceleration.

If we assume that the 50 N load is applied at the center of gravity, then the force results in only the linear acceleration in a triangle pulse of 0-1.16 g. The resulting value for the HIC is approximately 0.004: corresponding to essentially a zero chance (order of \(10^{-25} \%\), for those wondering) of an Abbreviated Injury Scale level 2 or greater (AIS2+) skull fracture [2] and well below the acute concussion threshold of 250 proposed in [1]. Meanwhile, if we were to perform the same analysis using the 200 N HES face contact load, we'd arrive at a triangle acceleration pulse of 0-4.66 g. This corresponds to an HIC of 0.12, significantly larger but with essentially the same probability of a skull fracture (nearly zero) and still well below the concussion threshold proposed in [1].

But how does that assumed triangle pulse line up with the reality of a thrust, and what effect does that have on the HIC?

In my earlier post on buckling, I presented some thrust load data I measured using a load cell arrangement. With that setup I was able to resolve an initial high-load peak with a subsequent plateau at a lower load value. Due to the time resolution of my setup (~15 ms), the peak was generally only 3 points, or about 30 ms from just before contact through the first point in the post-impact plateau. Schematically, this is what the triangle pulse described above would look like compared to the qualitative form of my impact measurements (time-aligning on the peak):

Schematic of measured thrust pulse vs assumed thrust pulse.

Of course, I noted that the time resolution of my thrust measurements is not sufficient to reliably capture the peak impact force, but it does indicate that there's a definite asymmetry to the loading pulse. The loading pulse also has a larger area, which means the integral for the HIC will be larger as well.

If we create a curve like the one above, but scaled to the 200 N face contact force (so the pulse goes linearly from 2.33 to 4.66 g in 7.5 ms then down to 3.15 g in another 7.5 ms), we arrive at an HIC of approximately 0.40, quite a bit larger than the value derived from the simple triangle pulse, but still well below the acute concussion or skull injury thresholds. If we instead take the maximum buckling load for an Albion Lichtenauer (1023 N) and use that as the maximum force in the asymmetrical triangle pulse, the HIC rises to approximately 24.26, still well below the acute concussion threshold in [1], whereas the maximum buckling load for an FIE homologated weapon (226 N, for the clamped/pinned sabre) results in an HIC of about 0.56.

Whatta pain in the neck...

There is a major limit to the HIC, however: it only considers linear acceleration of the head. The rotational acceleration and compression of the neck are also important, and the NHTSA uses a normalized neck injury index (\(N_{ij}\)) to quantify this aspect [2,3]. For vehicle crash safety design, the NHTSA recommends limits on the axial loading of a 50th percentile male's neck of 4000 N in compression and 4170 N in tension, with smaller values for the 50th percentile female [3].  For flexion and extension (see Figure above), the moments are limited to 310 N*m and 122 N*m respectively (lower for the 50th percentile female). The normalized neck injury index is calculated as follows:
\[N_{ij} = \frac{F_z}{F_{int}} + \frac{M_y}{M_{int}}\]
Where \(N_{ij}\) is the normalized neck injury index for a given mechanism combination (i.e Compression or Tension and Flexion or Extension), \(F_z\) is the axial force on the neck, \(F_{int}\) is the appropriate normalization value (i.e. the compression or tension limit), \(M_y\) is the flexion/extension bending moment calculated at the occipital condyles (roughly where the spine meets the skull) and \(M_{int}\) is the corresponding normalization limit.

The NHTSA recommends a limit of \(N_{ij} = 1.0\) for automotive collision design purposes, corresponding to a roughly 22 percent chance of serious neck injury (Abbreviated Injury Scale level 3 or AIS-3). For reference, an AIS-2 (moderate) severity neck injury could include injuries such a dislocation or minor fracture of cervical vertebrae without impinging on the spinal cord, whereas an AIS-3 (serious) severity neck injury could include multiple fractures or minor spinal cord impingement. Neither are something I'd like to deal with myself - while clinically they may be classified as moderate and severe, I'd classify them both as severe as far as injuries in a fencing bout go.

If we assume that we apply the estimated HES face-contact load of 200 N horizontally at eye level, approximately 3 cm above the occipital condyles, we end up with an net flexion moment of 6 N*m and no axial component, therefore \(N_{ij}= 0.019\). The correlations used by the NHTSA would predict a roughly 11.6% chance of a moderate (AIS-2) head injury. However, at low \(N_{ij}\) values, the NHTSA correlations predict unrealistically high injury probabilities. For example, for the no loading case \(N_{ij} = 0\) but the NHTSA provided correlations predict an 11.4 percent chance of moderate neck injury and a 3.8 percent chance of serious neck injury (AIS-3). This failure of the NHTSA model is entirely due to the lack of data at very low loads and moments. More recent work by researchers affiliated with the US Air Force has led to a new correlation with better agreement at low loads [4] (Lt. Col. Parr was nice enough to share it with me). With this newer correlation, the prediction of an AIS-2 injury at \(N_{ij} = 0\) is a much more reasonable 0.56%, while for AIS-3 type injuries the prediction is 0.49%.  For our estimated \(N_{ij}\) value of 0.019, the model of Parr et al. predicts a probability of 0.562% for an AIS-2 injury and a 0.522% probability of an AIS-3 injury. The FIE face-contact load of 50 N*m would result in a flexion moment of 1.5 N*m and \(N_{ij}=.005\), which corresponds to injury probabilities that are small fractions of a percent lower than that of the HES estimated face contact load.

But what about the buckling loads? If we use the maximum buckling load for an Albion Lichtenauer (1023 N) we obtain a flexion moment of approximately 31 N*m and \(N_{ij}= 0.1\): close to 0.781% chance of an AIS-2 neck injury, only slightly greater than that of the 200 N load. Meanwhile, for the maximum buckling load for an FIE homologated weapon (226 N, for the clamped/pinned sabre) the flexion moment is 6.78 N*m and the resulting \(N_{ij}=0.02\) and injury probabilities are nearly identical to that obtained based on the estimated HES face contact load. In all of these cases, the \(N_{ij}\) values are below that found for straight punches by Olympic boxers to a Hybrid III dummy's jaw (\(N_{ij}=0.27\pm0.07\))[1] by at least half, though I do not have enough information at this time to determine how the values found for Olympic boxers would compare to the maximum transmitted loads, velocity and accelerations during an actual thrust. This is definitely something worth taking a more detailed look at.

Pull-off Load/ Twist Load

One area where a big chunk of HES differs significantly from modern sport fencing is that there may be legal actions that result in loads on the mask that could lead to it being removed or rotated on the head: grapples and throws, for example. Because of some of the factors at play, it gets tricky to decide on how to treat the pull-off load: 20 N is not terribly large (4.5 lb-f), and there are reasons to prefer a mask get pulled off than staying on during a grapple (they often make far better handles than heads do). However, a mask which can be launched off by a rising thrust or cut is not necessarily desirable either. But these concerns also play heavily with how a bout is being managed: loss of equipment should be noticed by someone and appropriate actions taken (whatever that may be). In my opinion, making a recommendation beyond the EN13567 pull-off values would require significantly more study.

But there is one area that the EN13567 standard is lacking that should be addressed for HES purposes: the rotational loading that can be supported without bringing the mask into contact with the wearer's face. Here, I think it makes sense to apply the same 200 N load as the face-contact load, but in a different direction. In the rotational case, the load could still be applied at eye-level, but now it would be applied along a line traveling from the wearer's right to left. For practical purposes, the load would need to be centered roughly on the planform area that extended past the face (see image below), to generate the desired moment. The drawback is that I do not have injury correlation data for this loading type.

So what does this mean, Sparky?

In order to obtain a similar factor of safety relative to the buckling loads of the characteristic weapons as the 50 N value specified by EN13567 (and thus the FIE), we'd need a value of about 200 N for HES applications. Further, based upon my simple analysis here, the change in probability of serious neck/head injury resulting from the change in face-contact load from 50 N to 200 N would be negligible. In other words, by expecting the mask to not significantly deform under that type of loading, we would not expect a significantly higher probability of severe acute head or neck trauma. It would be silly to have equipment that could save your face, but break your neck. Further, I'd add in an explicit side-loading (twist) requirement that is similar to the frontal load face contact requirement.

For things like bib attachment, I see no reason to modify the EN13567 specification for HES applications. An argument could be made to increase the vertical pull-off force, because some aspects of HES involve grapples, but that will require further consideration. I think that could more easily, and more robustly, be handled by those managing the bouts rather than equipment changes.

So to summarize, I think the following is a reasonable face-contact recommendation for masks intended for HES applications:

No part of the mask mesh or equivalent rigid protective surface should contact the face of the wearer (i.e. nose, cheeks or chin) when a static load of 200 N (45 lbs-f) is applied parallel to the ground at approximately eye-level, oriented in the following directions:

1. Pushing the front of the mask towards the wearer's face (fore to aft alignment)
2. Pushing the mask from the wearer's left side toward their right side (twisting from their left to right)
3. Pushing the mask from the wearer's right side toward their left side (twisting from their right to left)

Now my next task is to see if I can actually get some measurements on this for a few different mask types, to see where things line up. It'll be interesting to test this one. I suspect that twisting will actually be the biggest issue in most modern fencing masks, and masks that have a suspension system (like a construction helmet's, possibly with some additional strapping) will fair better in that use-case. But I also suspect that the frontal loading will also become an issue for worn and less well-fitting masks.

 

In the next post, we'll talk penetration... Garment and mask penetration. Get yer minds out of the gutter, people.

References

[1] D.C. Viano. Head Impact Biomechanics in Sport, IUTAM Symposium on Impact Biomechanics: From Fundamental Insights to Applications. Springer-Netherlands, 2005.
[2] R. Eppinger et al. Development of Improved Injury Criteria for the Assessment of Advanced Automotive Restraint Systems-II. 1999. Available from the NHTSA website, www.nhtsa.gov, accessed Nov 1, 2013.
[3] R. Eppinger et al. Supplement: Development of Improved Injury Criteria for the Assessment of Advanced Automotive Restraint Systems-II. 1999. Available from the NHTSA website, www.nhtsa.gov, accessed Nov 1, 2013.
[4] J.C. Parr et al. Neck Injury Criteria Formulation and Injury Risk Curves for the Ejection Environment: A Pilot Study. Aviation, Space, and Environmental Medicine. 84(12), Dec 2013.