First, the airbag provides primary restraint to the head and face so that contact with the steering wheel is mitigated. This benefit has been documented thoroughly in frontal impact tests. This is not to say that an airbag precludes any risk of head or face contract, especially in the absence of concomitant belt use. Crandall et al. [29] and others have observed that head and face injuries are over-represented in unbelted occupants with an airbag compared to belted occupants with no airbag. Berg et al. [30] presented a series of seven cadaver tests, where an airbag restraint with no belt was used. They noted head impacts into the upper region of the windscreen or the roof in all tests. Depending on the size of the occupant, the severity of the collision, the geometry of the steering column and the vehicle interior, and the characteristics of the airbag, it is possible for an occupant to translate vertically over the airbag and sustain head or face loading via the steering wheel or the windshield/windshield header (Fig. 5.6). This injury mechanism emphasizes, again, the importance of the belt as a primary restraint and the airbag as a supplemental restraint. With a properly positioned belt restraint, the whole-body motion shown in Fig. 5.6 does not occur.

Second, the airbag distributes forces on the chest and shares loading with the belt system. Numerous studies have shown that cadavers restrained by a belt and an airbag sustain lower levels of injury than cadavers restrained by a belt alone or by an airbag alone [32] [7, 21, 33]. Shoulder injuries related to belt loading also seem to be reduced with the addition of an airbag restraint [34] (Fig. 5.7). These injury reductions are primarily due to the load sharing and force distributing effects of an airbag. It is well established that a person’s tolerance to a force applied on the thorax is highly dependent upon the area over which that force is applied. For example, Patrick et al. [35] performed a series of sled tests with embalmed cadavers impacting padded load cells anteriorly. They found that a 3.3 kN hub load to the sternum resulted in minor trauma, while similar injuries required approximately 8.8 kN if the load was distributed over the shoulders and chest. Bierman et al. [36] tested volunteers with a drop device that loaded a four-point belt harness 490 cm² in area. Painful reactions and some minor injuries occurred when loads exceeded 8.9 kN. When the load area was increased to 1,006 cm², loads up to 13.3 kN were sustained without pain or injury. The potential exists, therefore, to increase substantially the work done on an occupant and to perform this work at a non-injurious force level if the force can be distributed over the entire anterior thorax.

This work can be increased even further, without increasing thoracic injury risk, if the occupant is tensed during the impact. In a series of human volunteer sled tests performed at 48 km/h, loading through the steering wheel via tensed arms and through the floorpan via tensed legs had a pronounced effect in terms of controlling occupant kinematics and minimizing restraining loads through the belt and airbag. The role of arm bracing has been studied by Horsch and Culver [37] and shows a considerable ability to restrain the upper torso away from the steering wheel when the driver is alert of the pending crash. Similar effects can be seen with bracing by volunteers in simulated crashes [38]. However, the proportion of supplemental restraint by bracing decreases with increasing crash severity, simply because the occupant’s strength becomes negligible compared to the inertial forces exerted on the upper body. In these cases, the airbag plays a greater role. Nonetheless, comparison of the tensed human volunteer in Fig. 5.8 with the atonic cadaver in Fig. 5.6 indicates the substantial effect of musculature in a crash as severe as 48 km/h. Armstrong et al. [40] estimated that as much as 55 % of an occupant’s kinetic energy in a tolerable frontal impact may be dissipated via the work done by “propriotonic” restraint.

The load-sharing characteristics of airbags have led to the observation that thoracic acceleration, a widely used injury indicator, may not reflect an airbag’s benefit to the occupant. By distributing forces on the chest, the total force on the chest can be higher than the force generated by belt loading, thereby resulting in higher chest acceleration. Furthermore, the fundamental of upper torso lean is an important attribute of belt restraint and loading through the bony structures of the body. Forces concentrated on the shoulder or upper thorax are more tolerable than forces concentrated lower on the thorax or abdomen. With this background, it is necessary to evaluate chest acceleration critically when comparing belt-dominated loading with airbag-dominated loading since a higher level of force, and therefore acceleration, can be tolerated if the force is distributed or if the torso is pitched forward. Grosch [41] found that the expected benefit due to a supplemental airbag was not reflected by the acceleration levels measured in laboratory tests and concluded that it is necessary to consider other indicators of injury risk. Others have observed that the chest acceleration peak due to bag slap during deployment can be the global maximum, though the magnitude of this peak (approximately 50 g) is insufficient to cause acceleration-induced injury [42]. An acceleration peak such as this, having a different cause than a comparable sled test with belt loading, can lead to misinterpretation of the relative benefits of different restraint conditions. In the Grosch test series, chest acceleration and chest deflection exhibited contradictory trends, with acceleration increasing when an airbag was added, while deflection decreased. Further complicating the assessment is the fact that this phasing is different for drivers and passengers. Driver-side airbags typically deploy rearward nearly far enough to load the occupant prior to any forward translation of the driver in the vehicle. The result of the driver’s proximity to the deploying bag is that the airbag loading begins sooner after the bag deploys (approximately 50 ms after impact in a 48-km/h full-frontal sled test) than it does for the passenger. The driver’s chest acceleration therefore typically exhibits a single peak, which occurs during combined belt and airbag loading. A passenger-side occupant, on the other hand, sustains primarily belt loading over the first approximately 75 ms after impact, which results in peak chest acceleration due primarily to belt loading (Fig. 5.9). The later airbag loading then typically generates a second chest acceleration peak. Depending on the characteristics of the belt, airbag, vehicle interior, and occupant, a passenger’s global acceleration maxima may occur either under belt loading or under combined loading, while a normally seated driver’s maximum acceleration occurs under combined loading. This phenomenon was observed by Vezin et al. [43] and its importance for thoracic injury prediction in the laboratory is discussed in more detail by Kent et al. [31]. The combinatorial nature of belt and airbag loading also necessitates appropriate phasing of belt and airbag loading so that the combined effect does not exceed the acceleration tolerance of the human body, though cases exceeding the acceleration tolerance of a properly restrained occupant in an otherwise survivable frontal crash are assumed to be extremely rare (by inference from Melvin et al., 1998) [44]. Biomechanics researchers are currently working on the development of injury criteria that apply to combined belt and airbag systems. These criteria incorporate load distribution and load phasing within the functional form of the criterion, through multipoint measurement methods, or through the integration of the criterion within an injury risk function [5, 45].

While the airbag does share loading with the seat belt and thereby reduces the level of concentrated belt force exerted on the chest, it is typically an insufficient restraint by itself. As shown in Fig. 5.6, injurious thoracic loading can occur from steering wheel contact, even with an airbag restraint. This behavior was observed by Smith et al. [39] in volunteer tests and produced injuries in cadaver tests conducted by Yoganandan et al. [33]. Because airbags neither remain inflated nor provide lateral restraint, seat belts are needed to adequately control occupant kinematics over the range of crash types, including rollovers and side impacts.

Third, an airbag facilitates the use of force-limiting belts, and allows for the use of lower force limits than would be possible without a supplemental restraint. While numerous experimental studies have documented the benefits in the laboratory, the value of lower belt force limits (e.g., 4 kN vs. 6 kN) has also been observed in the field [8]. By distributing loads, mitigating head and face contact, and sharing loading with the belt system, an airbag allows for lower levels of belt force to be applied to the occupant. This has two primary benefits. First, a force-limiting belt can be designed to yield before injurious belt forces are developed (Fig. 5.10). Second, a force-limiting belt promotes forward rotation of the torso, which generates a “higher quality” biomechanical restraint condition (Figs. 5.3 and 5.11) [23]. While a reduction of belt force is associated with a lower risk of belt-induced thoracic injury, there is a limit on how low the shoulder belt force should be reduced without compromising protection in crashes involving long-duration deceleration or multiple impacts. If the belt has yielded in a previous collision, there can be a loss of upper torso control and (with a sliding latch plate design) lap control in subsequent collisions. Since the airbag will deflate after the primary impact, it cannot be relied upon to provide restraint in secondary impacts. In this case, the belt restraint system must be available and functioning to provide retention and restraint of the occupant.

The discussion above describes the mechanisms by which airbags supplement seat belts in frontal collisions, and also illustrates how some benefit can be realized by unbelted occupants. The discussion also implies, however, the biomechanical tradeoffs associated with an inflatable restraint. First, occupant loading should occur as early as possible in a crash, so that the maximum ride-down benefit may be achieved. This emphasizes early deployment of a relatively large airbag and also the use of belt pretensioning with early activation and fast retraction to initiate early belt restraint. Second, the magnitude of restraint loading applied to the occupant should be the maximum tolerable so that injurious contact with interior components can be avoided. These requirements emphasize an airbag with relatively high internal pressure so that large forces can be developed. In contrast, the work performed on the occupant should occur over as much distance as possible prior to contact with the vehicle interior in order to minimize the magnitude of force that must be applied. Depending on the crash severity and occupant size, this emphasizes an airbag with lower internal pressures. Furthermore, occupants of different mass and geometry must be considered. It is a substantial design challenge to optimize the airbag system’s performance given the range of conditions in which it must perform, including a single crash that may have both belted and unbelted occupants.