Quantifying the Relationship Between Vehicle Interior Geometry and Child Restraint Systems

Abstract

The prevention of interactions of children or child restraints with other vehicle structures is critical to child passenger safety. Fifteen current vehicles and seven rear and forward facing child restraint systems were measured in an attempt to quantify the available distance between child restraints and these vehicle structures. Rear facing child restraints exhibited such small amounts of clearance that contact would be expected in the majority of frontal crashes. Upper tethers are critical in the prevention of head contact, while head contact is likely when the upper tether is not used.

One of the most important aspects of child passenger safety is the interaction of the child and child restraint with other vehicle structures. In frontal crashes, rear facing child restraints (RFCR) rotate and translate forward and this movement may result in the restraint contacting the front vehicle seat or the truck dash. Previous studies have shown the potential danger of this interaction with the front dash (Sherwood et al., 2005), while NHTSA tests have shown high injury values as a result of RFCR restraints striking front vehicle seats (NHTSA, 2005). For forward facing child restraints (FFCR), dangerous interactions may occur if the restraint system is unable to limit the child’s forward excursion. In fact, studies suggest that head contact with other vehicle structures are the most frequent source of injury in restrained children (Arbogast et al., 2002).

In an attempt to prevent these interactions, the NHTSA included dynamic performance requirements as part of the Federal Motor Vehicle Safety Standard (FMVSS) 213 for child restraints. In addition to dummy injury criteria specified by the standard, RFCR cannot rotate past 70 degrees, while there are excursion limits for the child dummy’s head and knees in FFCR. These FMVSS tests are performed on a test buck intended to represent average vehicle rear seats, and at a speed and crash pulse that simulate the accelerations seen in a representative frontal crash. Several of these requirements were addressed in the update to FMVSS 213 in 2002. While NHTSA measured the vehicle interior geometry (Glass, 2005) in over 40 vehicles for this update, there are several reasons for a reevaluation of the vehicle interior measures for child safety. First, vehicle designs are constantly changing, and these changes must be quantified in order to understand their effect on child passenger safety. Second, the NHTSA study measured geometry based on the seat bight. Newer vehicles have LATCH bars which provide a more consistent and representative comparison to the FMVSS 213 test buck. Third, in the NHTSA study the front vehicle seat was positioned in a nominal upright position rather than the typical reclined driving position, which changes the available excursion distance. Finally, the report was not intended to measure the actual installation position of child restraints in the vehicles. Differences between the installed position of child restraints between the FMVSS 213 test buck and fleet vehicles must be measured since these differences may affect the results of these compliance tests, particularly excursion values.

Although sled tests can never completely reproduce the many variations experienced in actual crashes, there are significant differences between the FMVSS 213 test procedure and real-world crashes. To fully understand how the test conditions and results relate to the real world child restraint environment, these differences need to be analyzed and quantified. The primary factors which are different in the FMVSS 213 dynamic performance test are:

• ▪ Crash pulse

• ▪ Dynamic vehicle seat properties

• ▪ Vehicle seat/dash geometry vs. FMVSS 213 excursion limits

• ▪ Initial position of the child restraint

• ▪ Human vs. dummy response

• ▪ Real world child restraint misuse

While the first two factors are of critical importance, they would require a large number of full scale tests to fully address their importance, due to the variation of these factors in fleet vehicles. The remaining factors, however, can be analyzed and estimated in order to better predict the real world behavior of child restraint systems.

The goals of this study are to better understand the likelihood of the interaction of children and child restraint systems with other vehicle structures in frontal crashes. The first objective is to quantify the distance between RFCR and the front vehicle seat or truck dash in the current real world environment. The second objective is to estimate the likelihood of children in FFCR to contact these structures due to their forward excursion.

METHODS

Fifteen 2005 model year vehicles were chosen using data from the Partners for Child Passenger Safety Study (Arbogast, 2005). The PCPS data allowed the selection of vehicles which most commonly had crashes with child occupants. When a popular model was no longer manufactured in 2005, the replacement model was chosen (e.g., Pontiac G6 replaced Pontiac Grand Am). Four vehicle types were chosen (passenger cars, minivans, SUVs, and trucks), and the four most popular models were chosen within each category, except trucks in which three models were chosen (Table 1).

Table 1.

Vehicles measured in study (all 2005 model year)

Passenger Cars	Ford Taurus	SUVs	Chevrolet Suburban
	Honda Civic		Chevrolet Trailblazer
	Pontiac G6		Ford Explorer
	Toyota Corolla		Jeep Grand Cherokee
Minivans	Chevrolet Uplander	Trucks	Chevrolet Silverado 1500
	Ford Freestar		Dodge Ram 1500
	Dodge Gr Caravan		Ford F150
	Toyota Sienna

The back of the front passenger seat was positioned as specified by the vehicle manufacturer for the National Highway Traffic Safety Administration (NHTSA) New Car Assessment Program frontal crash testing. For vehicles with adjustable second row seat backs, the seat back was adjusted to the same angle as the front passenger seat. The geometry of the front vehicle seat was measured in the forward (FWD) and rear (REAR) most adjustment positions, which also allowed all intermediate positions to be calculated. In this study, four fore-aft adjustment positions were quantified: FWD, MID, UMTRI, and REAR. The MID position was the mid-track position, while the UMTRI position was calculated in an effort to more accurately describe the position used by a 50^th percentile male. Reed et al. (2001) developed the UMTRI model based on vehicle geometry and occupant stature to predict the seat track position. They found that the mid-track position specified in federal frontal crash standards is not an accurate reflection of where mid-size males position their front seat. Using this procedure for updated measurements on fifty five new model vehicles, the Insurance Institute for Highway Safety determined that the average predicted seat track position is 48 mm rear of the mid-track position (Arbelaez, 2006).

The second row seat geometry was also measured. In vehicles with second row fore-aft adjustment (Chevrolet Uplander, Toyota Sienna), both FWD and REAR positions were measured although only the REAR position was used for calculations in this study.

A three dimensional measuring device (FARO Technologies, Inc.) was used to measured the geometry of the vehicle interior. The origin of the reference frame for all measurements was the outboard LATCH bar (Figure 1). The FMVSS 213 standard uses the Z-Point as the reference origin, and excursion limits are based on this origin. This Z-Point is 102 mm rearward of the LATCH bars on the test buck. In order to make the values equivalent, the FMVSS 213 excursion reference values must be reduced by 102 mm as shown in Table 2. The LATCH reference frame will be used throughout the paper.

Figure 1.

Diagram of vehicle and child restraint measurements

Table 2.

Z-point and LATCH reference frame excursion limits

	Z-point reference frame		LATCH reference frame
Excursion limits	Head	Knee	Head	Knee
Tether	720	915	618	813
No Tether	813	915	711	813

Seven child restraints were chosen to represent popular convertible and forward facing only child restraints (Table 3). This analysis was part of a larger study investigating the benefit of RFCR for older children, thus the focus was on children one to three years old. Infant seats and booster seats were not studied, but the measured excursion values are applicable to all FFCR. Four of the restraints were convertible (rear or forward facing) and three were forward facing only restraints.

Table 3.

Child restraints measured in study

Child Restraint	Model	Rear Facing 12 month	Forward Facing 12 month	Forward Facing 3 year
Britax Marathon	E9L0688	X	X	X
Evenflo Triumph 5	3591208P1	X	X	X
Graco Comfort Sport	8630MTR	X	X	X
Cosco Alpha Omega	22-155TRP	X	X	X
Britax Husky	E9L3004		X	X
Graco Platinum Cargo	8489PEN		X	X
Cosco Summit	22-260HOU		X	X

Each child restraint was measured in all appropriate combinations of orientation and child size using the CRABI 12 month and Hybrid III 3 year dummies (Table 3), and was positioned on the second row passenger seat (first row in trucks). Restraints were installed appropriately using both lower LATCH and upper tether (FFCR) by a certified Child Passenger Safety technician, but without the internal harnesses being attached. Rear facing restraints were positioned such that the angle at the child’s torso was 40 degrees (with respect to vertical). This angle was chosen because while 45 degrees is the specified angle for infants, older children may sit more upright once the neck has strengthened sufficiently to support the head. In order to achieve this angle, spacers (placed under the base of the restraint near the seat bight) were frequently required, as specified by the Child Passenger Safety training manual. Once the dummies were positioned, coordinates of the child restraint, dummy head, and dummy knees were recorded.

The same measurement procedure was repeated on the FMVSS 213 buck for each of the restraints in the forward facing orientation. The rear facing child restraints were not measured because the pre-crash installation position for these restraints was determined from their actual FMVSS 213 compliance tests.

The measurements made in the vehicles and on the test buck provide data that can be used to better estimate the likelihood of contacting vehicle structures in real world crashes. For FFCR, these measurements determined the available excursion distance to the front vehicle seats and truck dash, and the different initial positions between the FMVSS 213 buck and in real vehicles. Two other factors (human vs. dummy response, real world misuse) must also be estimated. Although there are still many unknowns about the biofidelity of child dummies, it is generally accepted that the dummies are stiffer than children and this results in the dummies underestimating forward excursion distances. There is some research which allows estimates to be made. Brun Cassan et al. (1993) compare a three year old cadaveric test to a three year old dummy and found that the dummy’s head underestimated total head excursion by 43 mm. This value can be used to predict the additional excursion amounts expected by real children when compared to FMVSS 213 estimates.

The second factor which can only be approximated is the effect of misuse of child restraints in the real world. While gross misuse cannot and should not be considered in compliance testing, there are other misuse conditions that will likely always exist in the real world child restraint environment (Arbogast et al., 2002). One of the these scenarios, a loose internal harness, is perhaps the most common type of misuse. While attempts to decrease all misuse continue through education and labeling, loose harnesses will likely always remain, especially when compared to the FMVSS 213 installation procedure (stationary dummy, tensioned harness, thin dummy clothing). Although the amount of slack observed in the field has not been quantified, this study used an estimate of 40 mm. This magnitude of slack was felt to be a conservative estimate, because other types of common misuse that would result in additional frontal excursion (loose LATCH belt, loose vehicle belt, loose upper tether) were not included. Since tests in our laboratory have shown that the amount of extra slack in an internal harness results in approximately the same amount of additional frontal excursion, we have estimated the additional forward excursion amount of 40 mm. Each of these factors was considered in an attempt to determine a realistic estimate of the risk of head contact in real world frontal crashes.

RESULTS

REAR FACING

Measurements of the RFCR and vehicle were used to determine the clearance between the forward-most position of the RFCR and the rear-most position of the front seat/dash (Figure 1). This data was averaged for each vehicle type (Figure 2). Trucks were not included in the average for all vehicles since the number of children transported in trucks (9%) was substantially smaller than each of the other vehicle types that accounted for at least 24% of the vehicles transporting children (PCPS 2005).

Figure 2.

The distances between RFCR and the front seat/dash

The data in Figure 2 show the range of clearances between the RFCR and the front vehicle seat or dash as a function of front seat position. When the front seat is positioned in the forward most position (REAR in trucks), the average clearance values range from 58 mm (trucks) to 114 mm (minivans). All other seat positions, however, resulted in no clearance between the RFCR and front seat (except MID position in minivans). It should be noted that because the rear most point on the front seat and the forward most point on the RFCR may not be at the same height (Figure 1), this method may underestimate the absolute distance between these two points. This underestimate is not considered to be extreme because the height of these points were generally in the same vicinity. True clearance could only be determined by testing due to the complex combined motions of translation and rotation.

An attempt was made to compare RFCR installation positions in FMVSS 213 relative to the installation in real vehicles, particularly because spacers are not used to adjust the rear facing child restraint installation angle in federal standards. The use of spacers often results in the child restraint being more upright than the position typically specified by the CPS training manual. The four RFCR tested in this study were also tested for compliance in FMVSS 213 standards. The different installation angles are shown in Table 4, with the restraints being an average of 6 degrees more upright in the compliance tests. Differences in the translational position of the restraints were not measured.

Table 4.

RFCR installation angles

RFCR	In vehicles	FMVSS 213
Britax Marathon	40 degrees	35 degrees
Evenflo Triumph 5	40 degrees	36 degrees
Graco Comfort Sport	40 degrees	32 degrees
Cosco Alpha Omega	40 degrees	32 degrees

FORWARD FACING

The available excursion distance in vehicles was measured with the front row seat in various fore-aft adjustment positions. This measure is a vehicle based measurement based solely on vehicle geometry, as measured from the LATCH origin. The values are presented in the LATCH reference frame with the FMVSS 213 standards included for comparison (Figure 3). The only vehicle without a LATCH bar was the Ford F-150 so it was not included in measurements for FFCR. The average values for each vehicle type are presented in Table 5.

Figure 3.

Available excursion distances in vehicles, measured from the LATCH bar to the front seat/dash. The FMVSS 213 head excursion limits (Tether 618 mm, No Tether 711 mm) are included for reference.

Table 5.

Average available excursion distances (mm, ± standard deviation) in vehicles

Vehicles	Front Row REAR	Front Row UMTRI	Front Row MID	Front Row FWD
Pass Cars Average	552 ± 42	622 ± 40	670 ± 40	788 ± 38
Minivans Average	596 ± 64	650 ± 71	698 ± 71	799 ± 80
SUVs Average	516 ± 62	586± 54	634 ± 54	753 ± 47
Average (no trucks)	555 ± 62	619 ± 58	667 ± 58	780 ± 56
	Front Row FWD	Front Row MID	Front row UMTRI	Front Row REAR
Trucks Average	514	625	673	736

Depending on the vehicle, available excursion distances were both greater than and less than the FMVSS 213 excursion limits. On average, the SUVs had the smallest available excursion distances, while the minivans generally had the largest available distances. In the UMTRI position, the SUV available excursion distance of 586 mm was less than the Tether limit of 618 mm, while the trucks had the largest UMTRI available distance of 673 mm. When all non-truck vehicles were averaged, the average UMTRI position (619 mm) was almost identical to the FMVSS 213 With Tether limit (618 mm).

The different installation positions of the FFCR between the test buck and vehicles were also measured. Again, the Ford F150 is not included in this set of data because of the lack of the LATCH bar. The head and knee positions of both the 12 month and 3 year old dummies were measured and averaged for each vehicle type, and then compared to the same measures on the FMVSS 213 buck. In every vehicle type, the dummy was positioned farther forward in the vehicles (Figure 4). The head of the CRABI 12mo dummy was positioned an average of 40 mm farther forward, while the knees were positioned an average of 21 mm farther forward (not including trucks). The three year old dummy had a similar value for the head position (43 mm) while the knees were positioned 34 mm farther forward.

Figure 4.

Graph of differences in dummy position between FMVSS 213 and vehicle measurements. All dummy positions were farther forward in the vehicles.

One of the reasons for the more forward position of the dummy in the vehicles was the presence of head restraints, as shown in Figure 5. Head restraints were removed if possible, or adjusted rearward to minimize interference with the child restraint. In the twelve non-truck vehicles, there were four vehicles in which the head restraint was not removable and resulted in some sort of interference with the child restraint. Even when the vehicles with head restraint interference were removed, however, the average values were only minimally reduced. The twelve month dummy head value changed from 40 mm to 34 mm, and the three year old dummy head value changed from 43 mm to 37 mm.

Figure 5.

An example of how a non-removable head restraint may result in interference with a FFCR.

In order to get a more accurate estimate of the likelihood of children making contact in actual vehicles in real world crashes, several factors were used to adjust the FMVSS 213 test results for the three year old dummy (Table 6). First, both Tether and No Tether child restraint scenarios were considered. Second, both the maximum allowable excursion limit as well as actual excursion values measured in compliance tests were analyzed. The average FMVSS 213 excursion value was found by averaging the values of all forward facing child restraints tested with the three year old dummy during fiscal year 2005 compliance testing (NHTSA 2005). Excursion values were averaged for both Tether and No Tether tests.

Table 6.

Predicted average head excursion distances in real world crashes

	FMVSS 213 Limit Tether	Actual Avg. Excursion Tether	FMVSS 213 Limit No Tether	Actual Avg Excursion No Tether.
Comparison value	618 mm *	405 mm	711 mm	618 mm *
Initial distance vs. FMVSS 213	43 mm	43 mm	43 mm	43 mm
Additional excursion in child(vs. dummy)	40 mm	40 mm	40 mm	40 mm
Loose internal harness	40 mm	40 mm	40 mm	40 mm
Predicted real world excursion distance	741 mm	528 mm	834 mm	741 mm
Average available space measured in vehicles (UMTRI)	619 mm	619 mm	619 mm	619 mm

Note: The value of 618 mm for the average excursion value in No Tether tests for all FMVSS 213 results is correct. The same value of 618 mm for the FMVSS 213 Tether limit is only coincidence.

The analyses showed that when the front seat is placed in the UMTRI position, current child restraint systems would be expected to prevent head contact with the front seat when the upper tether is used (528 mm vs. 619 mm). This is primarily due to the fact that on average, current FFCR surpass head excursion limits by 213 mm. Child restraints without Tethers which met the excursion requirement without additional space, however, would be expected to allow head contact (741 mm vs. 619 mm). When upper tethers are not used, the average child restraint would be expected to allow head contact by a significant margin, even accounting for the lower excursion values allowed in child restraint testing in FMVSS 213 (741 mm vs. 619 mm).

DISCUSSION

REAR FACING

The adjustment range for vehicle front seats (seat back angle and fore-aft adjustment) resulted in a large number of possible positions with respect to the rear seat. When this range of variation is combined with multiple vehicle models and multiple child restraints, the available excursion distance data must be condensed for a meaningful analysis. For this discussion, the available excursion distance with the front seat positioned for a mid-size male was used as the reference value for comparison with federal safety standards.

When vehicle front seats are placed at the mid-track position or farther rearward in non-truck vehicles, there is generally no clearance space for a RFCR. Since the majority of front seat adjustment positions would result in insufficient space to allow a RFCR, it is likely that the most common situation is for the child restraint to be in contact with, or to be very close to, the front seat back. While there is limited research on the effect of contact between the child restraint and the front vehicle seat during a frontal crash, a small number of NHTSA tests suggest the potential for high injury values. Further research is needed to quantify the possible effect of these interactions.

In the trucks measured in this study, there was sufficient space for installing RFCR only when the front seat was at or close to the REAR adjustment position. Given the amounts of forward translation and rotation experienced by the typical RFCR in frontal crashes, contact with the dash is very likely even when positioned in the REAR adjustment position. A simple analysis of this data, conservatively assuming that forward movement was only due to rotation of the child restraint (no translation), suggested that only one RFCR/truck combination would not result in contact with the dash. Thus in the large majority of trucks, a RFCR would contact the front dash even with the front seat positioned in its full rear position. Based on research findings that RFCR perform better when more rigidly attached to the vehicle (Sherwood et al., 2005), in combination with data suggesting the potential for high injury values when a RFCR is positioned with an initial gap between the child restraint and a rigid structure, we suggest that when placed in trucks, RFCR should be installed such that they are in contact with the truck dash.

Spacers are frequently used in real world vehicles to allow RFCR to be more reclined, but are never used in FMVSS 213 tests. The result is a more upright test position of the restraints in the compliance tests. It is unknown if a less reclined installation angle has a significant difference on injury values, but it is likely that the more upright angle makes it less probable that the restraints will surpass the 70 degree angle limit. In addition, it is not known if the addition of spacers in real world vehicles results in detrimental child restraint performance.

FORWARD FACING

The available excursion distances, based on vehicle geometry, were smaller than the results from the 2002 NHTSA study (Glass 2002) by amounts ranging from 4 to 128 mm (Pass cars – 38 mm, Minivans 4 mm, SUVs - 128 mm, Trucks – 84 mm). The small difference observed in minivans is likely due to the fact that the 2^nd row minivan seats were positioned fully rearward in this study. It is probable that the other differences in values were largely due to the fact that the front seats were not reclined in the NHTSA study. The available excursion distances when the front seat was placed in the UMTRI positions were obviously smaller than the NHTSA mid-track values.

The goals of this study were to predict the likelihood of head contact in the real world child restraint environment. This was done by quantifying the excursion distances allowed by the FMVSS 213 standard, actual excursion distances measured in FMVSS 213 test results, the different initial position of child restraints between the FMVSS 213 test buck and real vehicles, the additional excursion distance for children in comparison to dummies, and the effect of real world misuse increasing excursion distances. These factors were combined for a predicted average real world excursion distance. This predicted average value of excursion distance was then compared to the available excursion distance measured in current vehicles. The analysis was performed based on data for the three year old passenger.

The predicted likelihood of head contact in the current child restraint environment had mixed results. When an upper tether is used, the results suggest that head contact will be prevented by a relatively large margin (91 mm). This is primarily due to the fact that the average new restraint model allows much less excursion than allowed by FMVSS 213 (405 mm vs. 618 mm). However, restraints which pass FMVSS 213 excursion requirements by only small amounts would allow head contact (741 mm vs. 619 mm).

The results are less encouraging for restraints when the upper tether is not used. In this case, the margin of safety for new restraints is much smaller (93 mm). Head contact is predicted even when the actual FMVSS 213 test results were used in the analysis (741 mm vs. 619 mm). This is particulary bad news because the upper tether is used in less than 20% of all FFCR (PCPS 2005).

In the process of installing the child restraint models in each of the vehicle models, several practical considerations were noted. Several vehicles had non-removable head restraints that caused interference with the FFCR and resulted in the child restraint being farther forward and closer to the vehicle front seat, although the head restraints themselves were not the primary cause of the difference between FMVSS 213 and vehicle initial position. In addition, in several vehicles it was difficult to tightly attach many upper tethers. Either a solid head restraint did not allow a direct routing from the child restraint to the tether anchor, or the ergonomics of the combination of the tether adjustment hardware and the anchor position was poor. Typically, the tether could still be attached, but reasonable tightening was not possible. Loose upper tethers will increase head excursion distances, but should not result in higher injury values than the same child restraint installed without the tether (Legault et al., 1997).

The findings in this study are not meant to necessarily suggest a change in FMVSS 213 excursion limits. Both child restraints and vehicles play role in protecting children. Many other factors are also involved. The findings do, however, provide more insight into how compliance tests and their results relate to the real world crash environment.

LIMITATIONS

The primary anslysis performed in this study used a single front vehicle seat position. In addition, only one front seat back angle was chosen. Although these represent the seating position of only the mid-size male, this seating position seems to represent a reasonable reference value.

Forward movement of front seats during a crash may increase available excursion distance. However, there is no published data on the movements of these seats in frontal crashes available for inclusion in this study.

CONCLUSIONS

The results of this study suggest that the interaction of RFCR with front seats or truck dashes is a very likely occurrence due to limited clearance distances at the time of installation. In order to reduce the risk of high energy impacts with the truck dash, it is recommended that in trucks RFCR be installed so that the rear of the RFCR is in contact with the truck dash. For FFCR, the analysis suggests that upper tethers are a crucial component in the prevention of head contact with the vehicle front seat or dash. In fact, when the upper tether is not used, the results predict that head excursion values will be large enough to allow head contact with these structures, in crashes similar to the FMVSS 213 test.