Interactions between rearward-facing child restraint systems and the front row seatback in frontal impact sled tests

Abstract Objective Some child restraint system (CRS) manufacturers specify a minimum distance between the CRS and the seatback, whereas others require that the CRS may contact the seatback but cannot be “braced”; however, few studies have investigated these interactions. Therefore, the aim was to investigate the interactions between the front row seat and rearward-facing CRS models with and without a support leg during frontal crashes. Methods Sled tests using the FMVSS 213 frontal crash pulse were performed with the Q1.5 and Q3 anthropomorphic test devices (ATDs) seated in rearward-facing infant and convertible CRS models, respectively. A front row vehicle seat was in front of the test bench in three track positions: brace, touch and gap. For the touch condition, the front row seat was translated aftward until the seatback contacted the CRS. For the brace condition, the front row seat was translated 20 mm aftward. For the gap condition, the front row seat was translated 50 mm forward. Each condition was tested with and without the support leg of the CRS. Results The tests with a support leg were associated with significantly (p = 0.007) lower resultant linear head acceleration 3 ms clip compared to the tests without a support leg, but the reduction of head injury criterion 15 ms (HIC15) was not significant (p = 0.057). The Q1.5 ATD in the rearward-facing infant CRS with a support leg had the lowest injury metrics for the touch and gap conditions, whereas the Q3 in the rearward-facing convertible CRS had the lowest head injury metrics for the brace condition. Conclusions The use of a support leg provided a clear benefit in terms of reducing head injury metrics for the Q1.5 in the rearward-facing infant CRS, especially for the touch and gap conditions. The rearward-facing convertible CRS in the current study appears to benefit from being braced against the front row seat. However, further tests are required to allow further statistical comparisons and determine if these preliminary findings extend to other rearward-facing CRS models.


Introduction
Use of a correctly used child restraint system (CRS) is associated with substantial reductions of injury and mortality risks in motor vehicle crashes (Arbogast et al. 2004, Elliott et al. 2006, Durbin and Hoffman 2018. In the late 1980s, some CRS models in Sweden were designed to be installed in the rear seat with a support leg, which extended to the floor of the vehicle. The purpose of a support leg (also referred to as a load leg) is to reduce the forward rotation of the CRS during a frontal impact. Although CRS models with a support leg have been common in Europe for several decades, it was only introduced to the United States market in 2004 and remains a relatively rare feature of current CRS models.
Few previous studies have investigated the interactions between a rearward-facing CRS and a front row seat in frontal impacts. Sherwood et al. (2004) tested the Child Restraint/Air Bag Interaction 12-month-old (CRABI-12) anthropomorphic test device (ATD) in different rearward-facing CRS models using the Federal Motor Vehicle Safety Standards (FMVSS) 213 frontal crash pulse. One CRS model had a support leg and was tested with no front row structure, whereas another CRS model did not have a support leg and was tested in three different conditions: no front structure, semi-rigid front structure and empty front row vehicle seat. For the latter two conditions, the CRS was initially contacting the front row structure. The test of the rearward-facing CRS model with a support leg resulted in the lowest ATD head injury metrics. For the rearward-facing CRS model without a support leg, the tests with no front row structure resulted in lower ATD head injury metrics compared to the tests with a front row structure (i.e., semi-rigid or empty vehicle seat).
In a subsequent study, Sherwood et al. (2005) tested the CRABI-12 ATD in three different rearward-facing infant CRS models in frontal impacts using the FMVSS 213 crash pulse. Each CRS model was tested in four restraint conditions: no front row structure, empty front row vehicle seat, rigid structure with no gap and rigid structure with a 150 mm gap. It was observed that the empty vehicle seat flexed forward during the impact and there was no contact between the CRS and the seatback; therefore, head injury metrics were similar to the condition with no front row structure. The rigid structure with no gap condition typically had lower head injury metrics compared to the no front row structure condition. In contrast, the rigid structure with a 150 mm gap condition had substantially greater head injury metrics compared to the other conditions. Tylko (2011) reported a large series of 131 rearward-facing CRS models (117 infant and 14 convertible) evaluated in 85 vehicle crash tests (offset deformable barrier tests at 40 km/h and rigid barrier tests at 40, 48 and 56 km/h) across a range of vehicles. The CRS were installed in the second or third vehicle rows. Adult ATDs were positioned in the front row driver and passenger seats, which were in the foremost track position for most tests, but were in the mid-or rearmost track positions for some tests. All but one test used the CRABI-12 ATD with the remaining test using a 3-year-old Q-series (Q3) ATD. Of the 107 tests with CRS models installed behind front row seats, the CRABI-12 ATD head contacted the seatback in 35 instances, 5 of which were associated with peak resultant linear head acceleration values above 80 g. In a further 36 instances the CRS contacted the seatback, 7 of which were associated with peak resultant linear ATD head acceleration values above 80 g. In the single test with the Q3, the CRS contacted the seatback and a peak resultant linear head acceleration above 80 g was recorded.
For the 15 tests in which the CRS models were initially contacting the front row seat back, none were associated with peak resultant linear ATD head acceleration values above 80 g.
More recently, Patton et al. (2020) performed sled tests to evaluate the effects of using a support leg on the head injury metrics of ATDs (CRABI-12, Q1.5, Q3 and Q6) in rearwardfacing CRS models (infant and convertible) during frontal impacts. The test buck comprised a blocker plate installed in front of the test bench seat to represent a front row passenger seatback. The frontal impact crash pulse was based on full-scale frontal crashes into a rigid-barrier , which had a delta-v of 56 km/h and a peak acceleration of 35 g with an associated rise time of 35 ms. It was found that the presence of a support leg in rearward-facing CRS models was associated with reductions in head injury metrics, which was attributed to the reduction in forward rotation that prevented contact with the blocker plate and/or reduced the contact force. However, the fidelity of the interaction between the CRS and blocker plate as an adequate representation of the interaction that would occur in a real vehicle is not well understood.
An additional gap in knowledge is the lack of understanding regarding the optimal positioning of a rearward-facing CRS relative to the front seat back. Some CRS manufacturers do not specify whether or not the rearward-facing CRS can make contact with the front row seatback during installation, whereas others specify that the CRS requires a minimum distance between the CRS and the seatback or that the CRS may make light contact with the seatback. For example, Diono state that "a rear-facing Diono car seat can touch the vehicle seat in front of it as long as it does not lift the bottom of the car seat off of the vehicle seat" (Diono 2021). In contrast, Evenflo states "unlike its convertible seats, however, Evenflo does require 1.5 inches of clearance between a front vehicle seat and any of its rear-facing only infant car seats." (Evenflo Company 2022). Evidence is lacking as to which approach represents best practice.
The current study addresses the aforementioned gaps in the research by quantifying the head injury metrics associated with CRS installation conditions to further the understanding of interactions between the front row seat, in various track positions, and rearward-facing CRS models, with and without a support leg, during frontal crashes.

Methods
The sled was driven by a 12-inch hydraulic-controlled gas energized (HYGE) Crash Simulation System: stroke length, 1.54 m; maximum payload, 2268 kg; maximum thrust; 1000 kN; maximum acceleration, 24-50 g; maximum velocity, 20.56-29.50 m/s. The design of the HYGE provides extremely repeatable and reproducible acceleration pulses, enabling accurate simulation and modeling of crash conditions (HYGE Inc 2020). The FMVSS 213 frontal impact crash pulse is used to certify the safety performance of CRS in the United States and was used for all tests in the current study ( Figure 1). The National Highway Traffic Safety Administration (NHTSA) recently explored revising the FMVSS 213 frontal impact crash pulse in a Notice of Proposed Rulemaking, but ultimately decided this to be unwarranted.
The test buck was attached to the sled deck and comprised a test bench seat, front row vehicle seat and sled-mounted highspeed camera system ( Figure 2). The test bench seat was based on the second-row seat of a 2010-2011 Ford Flex sport utility vehicle, which was rigidized to increase durability for repeated testing. The test bench seat springs and cushions were replaced every six tests. The front row vehicle seat was from a contemporary model General Motors pickup truck and had manual track and recline adjustment. The tracks of the front row vehicle seat were fitted to a track mount that was fixed to the sled deck. A conductive foil panel was attached to the rear surface of the front row seatback and head restraint. A new front row vehicle seat was used for each test. Between the test bench and the front row seat, a force plate was installed on the floor of the test buck to measure the ground reaction force during tests involving a support leg. The force plate was covered with non-slip grit tape. Four high-speed video cameras were attached to an on-board camera frame, which provided top, rear and side views that moved with the test buck during each test. An additional high-speed video camera was mounted to the ceiling and provided a stationary top-down view of each test. The high-speed video cameras recorded frames with a resolution of 1600 Â 1228 at a rate of 2000s À1 for a duration of approximately 450 ms.
Two calibrated Q-series ATDs were used to represent pediatric occupants aged 18 months and 3 years: Q1.5 and Q3, respectively. The heads of the ATDs were instrumented with 3axis accelerometers located at the center of gravity of the head. A strip of conductive foil was attached to the top of the head of each ATD so that a voltage signal identified contact with the conductive foil on the front row seatback, herein referred to as head contact. The Q1.5 (mass, 11.1 kg; stature, 0.80 m) and Q3 (mass, 14.6 kg; stature, 0.99 m) ATDs were seated in the rearward-facing infant or convertible CRS models, respectively.
The rearward-facing infant CRS could accommodate occupants with a body mass from 2.0 to 14.5 kg and a stature of less than 0.81 m. The base of the rearward-facing infant CRS had rigid lower anchor connectors. The rearward-facing infant CRS was attached to the base and the handle was positioned vertically, as specified by the manufacturer. When the support leg was not used, it remained stowed in the available space underneath the rearward-facing infant CRS. The rearward-facing infant CRS had a mass of 9.6 kg. The rearward-facing convertible CRS could accommodate occupants with a body mass less than 18.5 kg and a stature from 0.61 to 1.05 m. The rearwardfacing convertible CRS had rigid lower anchor connectors. The support leg of the rearward-facing convertible CRS could not be stowed; therefore, it was removed for tests in which the support leg was not used. The rearward-facing convertible CRS had a mass of 13.1 and 13.9 kg with and without the support leg, respectively.
The ATDs were secured in the CRS models as specified in FMVSS 213 (National Highway Traffic Safety Administration 2014). The CRS models were installed on the test bench by locking the rigid lower anchor connectors onto the anchor points in the seat bight. Initially, the front row vehicle seat was reclined to 26 from the vertical and was translated aftward (i.e., toward the CRS) until the seatback contacted the CRS. This condition is hereinafter referred to as the touch condition. For the brace condition, the front row seat was translated 20 mm aftward from the touch condition. For the gap condition, the front row seat was translated 50 mm forward from the touch condition. The two rearward-facing CRS models were tested in each condition with and without a support leg for a total of 12 tests (Table 1, Figure 1A).
Data from ATD instrumentation were filtered, as specified in SAE International J211 (SAE International 2014), and full time-histories of head accelerations were recorded from which the following injury metrics were calculated: resultant linear head acceleration 3 ms clip and head injury criterion 15 ms (HIC15). Injury metrics were compared to previously published injury tolerance values (Table 2). Separate multiple linear regressions for resultant linear head acceleration 3 ms clip and HIC15 were used to identify significant (p < 0.05) associations between the test conditions (independent variables) and injury metrics (dependent variables). CRS model, support leg use and head contact were coded as categorical variables and track position was coded as a continuous variable. Linearity for the categorical independent variables was assumed and standard multiple linear regression diagnostics were performed for the normality (i.e., normal probability plot), homoscedasticity (i.e., scatterplots of standardized residuals) and multicollinearity (i.e., variance inflation factors) assumptions.

Results
The tests with a support leg were associated with significantly (p ¼ 0.007) lower resultant linear head acceleration 3 ms clip compared to the tests without a support leg (Tables 3, 4 and A2). Although HIC15 values were lower for most tests with a support leg compared to the respective tests without a support  leg, no significant association was found (p ¼ 0.057). All head injury metrics were below head injury tolerance values, except for the resultant linear head acceleration 3 ms clip of the Q1.5 ATD in the rearward-facing infant CRS without a support leg for the brace and touch conditions, which exceeded the head injury tolerance value scaled from UN ECE R94 ( Figures A2  and A3). The time-histories of linear ATD head acceleration for all 12 sled tests are depicted in Figure A4. Minimal forward rotation of the front row seatback was qualitatively observed in all tests. Greater forward rotation of the CRS without a support leg compared to the respective tests with a support leg was qualitatively observed for both the Q1.5 ATD in the rearward-facing infant CRS and the Q3 in the rearward-facing convertible CRS ( Figure A5). For the tests of the Q3 ATD in the rearward-facing convertible CRS in the touch and gap conditions, the CRS models were observed to slide forward along the adjustment track of the CRS base, which decreased the recline angle of the CRS ( Figure A6). This phenomenon did not occur for the brace condition.
The Q1.5 ATD in the rearward-facing infant CRS with a support leg in the gap condition was the only test in which the CRS did not contact the front row seatback. For the Q1.5 ATD in the rearward-facing infant CRS, head contact with the front row seatback was identified in all tests without a support leg, but only for the brace condition with a support leg. For the Q3 ATD in the rearward-facing convertible CRS, head contact was observed in the touch and gap conditions with a support leg, but only for the gap condition without a support leg.
Peak resultant support leg reaction force values ranged from 4950 to 5124 N for the tests of the Q1.5 ATD in the rearwardfacing infant CRS. For the tests of the Q3 in the rear-facing convertible CRS, peak resultant support leg reaction force increased across the brace, touch and gap conditions (3987, 4773 and 5837 N, respectively).

Discussion
A correctly used CRS is associated with substantial reductions of injury and mortality risks in motor vehicle crashes (Arbogast et al. 2004, Elliott et al. 2006, Durbin and Hoffman 2018. In addition, previous studies of rearward-facing CRS models in frontal impacts have found that the presence of a support leg is associated with a reduction in ATD head injury metrics (Sherwood et al. 2004, Sherwood and Crandall 2007, Patton et al. 2020, Patton et al. 2021. Some CRS manufacturers specify a minimum distance between the CRS and the seatback, whereas others require that the CRS may contact the seatback but cannot be "braced"; however, few studies have investigated these interactions. Therefore, the current study used sled tests to investigate the interactions between the front row seat, in various track positions, and rearward-facing CRS models, with and without a support leg, during frontal crashes. The tests with a support leg were associated with significantly (p ¼ 0.007) lower resultant linear head acceleration 3 ms clip compared to the tests without a support leg. Such a finding adds to the growing body of biomechanical evidence that   support legs are an effective injury prevention design feature of rearward-facing CRS models in frontal impacts (Sherwood et al. 2004, Sherwood and Crandall 2007, Patton et al. 2020, Patton et al. 2021. Although not significant (p ¼ 0.057), HIC15 values were lower for all tests with a support leg, compared to the respective test without a support leg, except for the Q3 ATD in the rearward-facing convertible CRS with a gap. For the Q3 ATD in the rearward-facing convertible CRS installed in the touch and gap conditions, the CRS models were observed to slide forward along the adjustment track of the CRS base. The cause of this movement is unknown, but may be a result of using the rearward-facing convertible CRS in a misuse condition (i.e., removing the support leg). The HIC15 values from these tests were approximately 10% lower compared to values from the equivalent test that used a support leg. Similar to the damping systems in some bicycle helmet designs (Halldin et al. 2001), this mechanism of relative motion between CRS and base may be interesting to explore for CRS design. For tests with the Q1.5 ATD in the rearward-facing infant CRS with a support leg, the brace condition had the highest head injury metrics compared to the touch and gap conditions, which was likely due to the head contact observed in the brace condition. The head of the ATD was observed to contact the top of the front row seatback, which was attributed to the height of the rearward-facing infant CRS relative to the front row seatback. Arbogast et al. (2012) impacted front row vehicle seatbacks with a pediatric ATD headform to evaluate the energy management characteristics of relevant contact points for head injury. Impacts to the top of the seatbacks resulted in some of the highest peak resultant linear ATD head accelerations compared to the other seatback regions. For the tests with the Q1.5 ATD in the rearward-facing infant CRS without a support leg, the brace and touch conditions were associated with resultant linear head acceleration 3 ms clip values exceeding the injury tolerance value scaled from UN ECE R94 of 70 g. The full time-histories of resultant linear head acceleration for these two tests include high magnitude spikes at the timepoint of ATD head contact with the top of the front row seatback. It has previously been suggested that there may be potential to reduce head injury metrics for rearward-facing CRS models in frontal impacts by improving the energy absorption properties of the front row seatback (Patton et al. 2020).
For the Q3 ATD in the rearward-facing convertible CRS, head injury metrics increased as the front row seat moved further away from the CRS in the brace, touch and gap conditions. This finding supports previous studies of full vehicle and sled tests. Tylko (2011) conducted a large series of 131 rearward-facing CRS models evaluated in 85 full vehicle crash tests and reported that none of the 15 tests in which the CRS was initially touching the front row seatback were associated with peak linear ATD head accelerations above 80 g. Sherwood et al. (2005) reported that rearward-facing CRS models installed with a 150 mm gap from a rigid front row structure resulted in higher head injury metrics recorded by the CRABI-12 ATD compared to the condition with no gap during frontal impact sled tests. In the current study, a gap between the rearward-facing convertible CRS and a front row seatback provided the space to increase the closing velocity of an impact between the CRS and the seatback. The head restraint of the convertible rearwardfacing CRS was relatively higher than the top of the front row seatback. Therefore, the front row seat supported the rearwardfacing convertible CRS, similar to the role of a support leg, which is likely why peak resultant support leg reaction force increased as the front row seatback was positioned further from the CRS (i.e., from the brace condition to the touch condition to the gap condition). In contrast, the head restraint of the rearward-facing infant CRS was below the top of the front row seatback and peak resultant support leg reaction force values were similar across track positions. Therefore, there was likely minimal support from the front row seatback. Sherwood et al. (2005) performed a test with the rearward-facing CRS models installed against an empty front row vehicle seat, the seatback of which rotated forward approximately 16 during the impact and resulted in no contact between the CRS and the seatback. The front row seat was from a fourth generation Ford Mustang (1994)(1995)(1996)(1997), whereas the front row seat used in the current study was from a contemporary model General Motors pickup truck, which exhibited minimal forward rotation during tests. This finding was expected due to the stiffening of front seat structures in recent years (Parenteau and Viano 2020).
There are several limitations associated with the current study. Most importantly, sled tests involve an idealized set of crash conditions for a simplified vehicle surrogate. For example, the test bench used in the current study was developed to be representative of a vehicle rear seat; however, seat dimensions, materials and properties vary across vehicle types, makes and models. The location and angle of a front row seat during a crash depends on the vehicle type, initial track position and recline angle of the seat, which were standardized for the tests in the current study. The pediatric ATDs used in the current study are some of the most advanced available; however, the biofidelity of any ATD is an idealized simplification of a human occupant. Also, injury tolerance values for pediatric ATDs are mostly scaled from adult data due to the paucity of pediatric injury data. Another limitation is that the CRS models used in the current study were selected for their features (i.e., support leg) and do not represent the full range of CRS models available; other CRS model with different design features may show varying results. The rearward-facing infant CRS model in the current study was designed to be used with and without a support leg. In contrast, the rearward-facing convertible CRS model was designed to be used with a support leg. Removing the support leg for certain tests changed the mass of the rearward-facing convertible CRS by 0.8 kg and represented a misuse condition (i.e., not used in the manner required by the manufacturer). Similarly, the front vehicle seat selected for the current study does not represent the full range of front vehicle seats in the modern fleet. In addition, the front row seat was empty and, therefore, the influence of a front row seat occupant was not assessed, which would likely affect the dynamics (Kuppa et al. 2005). Lastly, no repeat tests were performed. Although the test buck used in the current study has exhibited good repeatability in previous studies, the repeatability of the test setup with the addition of the front row vehicle seat has yet to be established.
In summary, tests with a support leg typically resulted in lower head injury metrics compared to frontal impact sled tests without a support leg. The use of a support leg provided a clear benefit in terms of reducing head injury metrics for the Q1.5 in the rearward-facing infant CRS, especially for the touch and gap conditions. The use of a support leg reduced resultant linear head acceleration 3 ms clip for the Q3 ATD in the rearward-facing convertible CRS, but the effect on HIC15 was less clear. For the Q3 ATD in the rearward-facing convertible CRS, head injury metrics increased as the front row seat moved further away from the CRS in the brace, touch and gap conditions. Therefore, the rearward-facing convertible CRS in the current study appears to benefit from being braced against the front row seat. Further tests are required to determine if these findings extend to other rearward-facing CRS models.