Abstract
The performance of rear facing child restraints in frontal crashes can be determined by controlling a) the child’s kinematics and b) interactions with vehicle structures. Twelve sled tests were performed to analyze the effect of the location and structural properties of vehicle interior components. The role of restraint kinematics was studied by developing computational models which underwent idealized motions. Stiff structures originally offset from the restraint, but which contact the restraint late in the test, cause increased injury values. Attachment methods which reduce child restraint rotation and more rigidly couple the restraint to the vehicle result in the best safety performance.
Vehicle restraint systems are typically designed for adults and do not fit the smaller anatomy of children. Restraint designs for children less than approximately 4 years of age require two restraint systems; first the child restraint must be attached to the vehicle, and then the child must be attached to the child restraint. Each system is important and contributes to the overall performance.
The attachment of the child restraint to the vehicle, however, is especially crucial because this attachment determines the kinematics of the child restraint, and subsequently the kinematics of the child. The attachment methods for forward facing child restraint systems (FFCR) have evolved to improve safety and simplify installation. LATCH systems and upper tethers reduce incorrect installation of child restraints (Lowne et al., 1994) and result in more rigid attachment of the child restraint to the vehicle (Lowne et al., 1997; Turbell et al., 1993). This system reduces forward rotation and translation of the child restraint which lowers acceleration forces and excursion amounts in frontal crashes.
The use of LATCH anchors affect rear facing child restraints (RFCR) to a much lesser extent. Only the lower LATCH anchors are used, and although the use of the lower anchorages simplifies installation, it does not result in significant changes to RFCR kinematics. The restraints undergo translation and rotation in frontal crashes. The translation is largely controlled by the lower LATCH anchorages, but rotation of the restraint is the primary motion and is determined by the interaction with the vehicle seat cushion. Vehicle seats are designed primarily for adult passenger comfort and safety, however, and may not result in optimal RFCR performance (Le Claire et al., 2000).
The kinematics of RFCR are important for two reasons. First, significant movement of the restraint in frontal impacts may allow contact with other vehicle structures such as the front vehicle seat when restraints are placed in rear rows, or the dash in pickup trucks. In a limited test series conducted by the National Highway Traffic Safety Administration, contact of RFCR with front vehicle seats appeared to result in high injury values (NHTSA, 2005) in a small number of tests. Computational studies have also shown the possibility of increased injury values due to contact with the front seat (van Rooij et al., 2003). The properties of the structure and the initial gap between the structure and the child restraint will determine injury values. Contact with other structures may help explain why infants have higher rates of skull fracture and intracranial injury than children one to four years of age (Gotschall and Luchter, 1999). Second, the kinematics of the RFCR determine the kinematics of the child, which is the most important factor for determining injury measures. This is particularly true for RFCR because the structure of the restraint provides support for the entire body and plays a more significant role than the harness (Kamren et al., 1993; Weber, 2002). Only the internal harness restrains the child in FFCR, while both the harness and the child restraint structure support the child in RFCR. More simplified investigations of the effect of RFCR kinematics on injury values may assist in the design of the methods used to attach RFCR to vehicles.
The objective of this study was to investigate a) the effect of the location and structural properties of vehicle interior structures, and b) the effect of different RFCR kinematics on injury measures. Sled tests and computational models were used to accomplish these objectives.
METHODS
SLED TESTS
Sled tests were performed to measure how the interaction of a RFCR with different vehicle structures may affect the dummy injury measures. A total of 12 sled tests were performed using a 48 km/h impact speed, with a pulse comparable to the FMVSS 213 acceleration pulse (Figure 1). The tests were performed on a 3rd row Ford Windstar minivan bench seat. The seat pan and seat back were rigidly attached to the buck to create a durable, consistent seat system.
Figure 1.
Sled acceleration pulse
A CRABI 12 month old dummy was positioned in the restraints, and was equipped with head, chest, and pelvis accelerometers, and upper neck (UN) and lower neck (LN) load cells. The coordinate system and electronic data collection procedures followed Society of Automotive Engineers (SAE) Recommended Practice J211. Each test was recorded at 1000 frames/sec with side and overhead digital video cameras. Planar video analysis using the side camera views was used to calculate child restraint rotation angular data.
The dummy was tested in three different popular convertible child restraint models (Britax Roundabout (BTX), Evenflo Vanguard 5 Comfort Touch (EF), Safety 1st Comfort Ride (SF)). Each restraint was tested in four restraint conditions: No Structure, Vehicle Seat, Rigid 0, Rigid Gap (Figure 2). The rigid structure was a welded frame made of steel tubing, and had no measurable deformation. An additional condition with an initial gap between the child restraint and the vehicle seat was planned, but as described in more detail in the Results section, the lack of any contact between the child restraint and the vehicle seat even with no gap eliminated the need for testing this restraint condition. Each restraint was attached to the vehicle using the lower LATCH belt with an initial angle of 40 degrees (measured at the dummy’s back, with respect to vertical). This angle was chosen because children at 12 months of age can sit more upright than the 45 degree angle recommended for newborns. Foam spacers were placed under the base of the restraint in order to achieve this angle.
Figure 2.
Pre-crash photographs of Safety 1st restraint in each restraint condition (some photos reversed for easier comparison)
The lower LATCH belt was tensioned as tightly as possible. The internal harness, however, had no slack, but was not placed in tension to match what is commonly seen in real-world installations (Decina and Knoebel, 1997).
The No Structure condition represented the FMVSS 213 test, and a condition in which only the LATCH belt and vehicle seat cushion coupled the restraint to the vehicle. The Vehicle Seat condition represented the common real-world condition in which a RFCR may be in contact with the front vehicle seat after installation. The vehicle front seats were Ford Mustang (MY 1994–97) front driver or passenger bucket seats, and were positioned to provide initial contact with the child restraint.
The Rigid 0 condition represented a scenario in which a RFCR might be placed against a pickup truck dash (with airbag deactivated), although the dash would not be as stiff as the rigid structure. The Rigid 0 condition also represented possible RFCR designs with built-in components, such as anti-rotation legs. As in the Vehicle Seat test, there was initial contact between the structure and the child restraint. Finally, the Rigid Gap condition modeled a scenario in which there was an initial gap between the restraint and the truck dash. The structure was moved forward until there was an initial gap of 15 cm between the structure and the child restraint.
COMPUTATIONAL MODELS
The typical motion of a RFCR in a frontal crash is a combination of translation and rotation. To gain an understanding of how each of these components of the RFCR motion affect injury values, computation models were developed which allowed each motion to be controlled individually.
A MADYMO multi-body model of the No Structure condition was constructed and validated with the sled test. The model simulation was performed in the multibody simulation environment MADYMO 6.1 (TNO 2003). The RFCR was modeled as a single rigid body with a meshed surface. The child occupant was modeled using the CRABI 12 month old database included in the MADYMO 6.1 release. As in the sled tests, the combination of the LATCH belt and the vehicle seat cushion allowed the RFCR to both translate and rotate.
In order to study the effect of varying kinematics in a more idealized environment, the RFCR was isolated from all vehicle components and its motions defined mathematically relative to the vehicle reference frame. A kinematic joint was used to allow the following types of motion: 1) Rigid, 2) Translation only, 3) Rotation only, as well as 4) Combined translation/rotation. The joint was located on the rear edge of the child restraint base (near the child’s feet) which approximated the center of rotation of the child restraint in the sled test. Translational and rotational springs with linear stiffnesses were added to the joint to limit the motion of the RFCR to approximately the same distances measured in the sled test. The forward translation of a point on the restraint, located just behind the head of the occupant, was approximately 9 cm. Head and chest accelerations along with upper and lower neck loads were used to evaluate the different idealized motions.
RESULTS
SLED TESTS
The kinematics for the Safety 1st child restraint in each restraint condition are shown in Figures 3 and 4. The maximum net rotation angle for each test are located in Table 1, and the time based rotation data for the Safety 1st restraint are shown in Figure 5.
Figure 3.
Images of Safety 1st restraint in No Structure (left) and Vehicle Seat (right) conditions at 0, 25, 50, and 75 ms
Figure 4.
Images of Safety 1st restraint in Rigid 0 (left) and Rigid Gap (right) conditions at 0, 25, 50, and 75 ms
Table 1.
Child restraint maximum net rotation angle
| No Structure | Vehicle Seat | Rigid 0 | Rigid Gap | |
|---|---|---|---|---|
| Britax (deg) | 32 | 29 | −29 | 8 |
| Evenflo (deg) | 16 | 18 | −9 | 8 |
| Safety 1st (deg) | 24 | 24 | −4 | 9 |
| Mean (SD) | 24 (8) | 24 (6) | −14 (13) | 8 (1) |
Figure 5.
Rotation data for the Safety 1st child restraint
The No Structure and Vehicle Seat tests had the largest rotation values, each rotating an average of 24 degrees. The vehicle seats underwent an average of 16 degrees of forward rotation due to their own inertia. At the onset of the crash pulse, the front vehicle seats underwent enough forward rotation due to their own inertia that there was no substantial interaction with the child restraints.
The rigid structure prevented excursion and actually caused the child restraint to move vertically and rotate in the rearward direction. The variation of this rotation in each RFCR was due to variations in the design of each child restraint, including the geometry of the child restraint seat and its interface with the LATCH anchors. When the rigid structure was placed 15 cm forward of the child restraint, the kinematics followed the No Structure test until contact was made and the angle (8 deg) remained relatively constant.
Measured injury data for all restraints and conditions are shown in Tables 2–4. The Vehicle Seat test data were very similar to the No Structure tests, as suggested by the kinematic data. There were, however, substantial differences in the Rigid 0 and Rigid Gap tests. Figure 6 shows the effect of the addition of the rigid structure, measured with respect to the No Structure tests, using the data averaged from each child restraint.
Table 2.
Injury data for Britax restraint conditions
| BTX No Structure | BTX Veh Seat | BTX Rigid 0 | BTX Rigid Gap | |
|---|---|---|---|---|
| Head Res Acc 3 ms clip (g’s) | 52.4 | 51.7 | 55.2 | 68.6 |
| HIC 36 | 560 | 476 | 318 | 1677 |
| Upper Neck Fx (N) | −219.5 | −407.6 | −279.3 | −439.4 |
| Upper Neck Fz (N) | 1394.6 | 1149.9 | 441.7 | 1421.6 |
| Upper Neck My (Nm) | −13.1 | −13.9 | −11.4 | −15.1 |
| Lower Neck Fx (N) | 324.1 | 222.3 | 4.4 | 452.2 |
| Lower Neck Fz (N) | 1482.2 | 1294.8 | 499.7 | 1517.8 |
| Lower Neck Mz (Nm) | −10.5 | −8 | −2.3 | −12.2 |
| Chest Res Acc 3 ms clip (g’s) | 47.7 | 43.7 | 52.2 | 87.7 |
Table 4.
Injury data for Safety 1st restraint conditions
| SF No Structure | SF Veh Seat | SF Rigid 0 | SF Rigid Gap | |
|---|---|---|---|---|
| Head Res Acc 3 ms clip (g’s) | 55.9 | 56.9 | 50.5 | 150.3 |
| HIC 36 | 436 | 524 | 265 | 1633 |
| Upper Neck Fx (N) | −408.6 | −279.8 | −513.8 | −358.8 |
| Upper Neck Fz (N) | 1190.1 | 1319.1 | 1040.5 | 1385.9 |
| Upper Neck My (Nm) | −13.1 | −8.8 | −17.3 | −12.5 |
| Lower Neck Fx (N) | 437.4 | 495.2 | 77.8 | 435.6 |
| Lower Neck Fz (N) | 1298.7 | 1400 | 1214.2 | 1457.2 |
| Lower Neck Mz (Nm) | −14 | −15.1 | −4.6 | −12.4 |
| Chest Res Acc 3 ms clip (g’s) | 30.6 | 32.5 | 41.5 | 68.1 |
Figure 6.
Percentage change in injury values (with respect to No Structure test) due to addition of vehicle seat and rigid structure
The Rigid 0 test resulted in decreased values for almost all injury measures, with respect to the No Structure test. No measures were increased by more than 15%, while several measures had average decreases greater than 40% (HIC36, UN Fx, UN Fz, LN Fx, LN Fz, and LN My). Although there were variations in magnitude, the addition of the rigid structure caused the same trends in each child restraint.
The Rigid Gap tests, however, caused increases in injury values for almost all injury values, when compared to the No Structure test. When averaged for all restraints, the rigid structure caused all injury values to increase. In particular, the head and chest values increased by more than 70%.
COMPUTATIONAL MODELS
The following four conditions were simulated: Rigid, Translation, Rotation, and Combined (Translation/Rotation). Figures 7, 8, and 9 show the motion of the RFCR at a point near the head of the occupant (the rigid condition had no motion of the RFCR.)
Figure 7.
Translation System
Figure 8.
Rotation System
Figure 9.
Combined (Translation/Rotation) System
The Rigid and Translation RFCR configurations resulted in much lower neck injury measures when compared with configurations that had a rotational component. Figure 10 shows the upper and lower neck loads, while Figure 11 shows the upper and lower neck moments.
Figure 10.
Upper and lower neck load values for idealized motions
Figure 11.
Upper and lower neck moment values for idealized motions
The head injury levels were also lower for the Rigid and Translation systems when compared to configurations that had a rotational component, except for the HIC36 value in the translational system which was higher than any other condition (Figure 12). The resultant head acceleration and thorax acceleration time histories are show in Figures 13 and 14. The early onset of the acceleration values in the rigid configuration resulted in lower durations of head acceleration, and lower peak values in chest acceleration.
Figure 12.
HIC data for idealized motions
Figure 13.
Head resultant accelerations.
Figure 14.
Thorax resultant accelerations.
DISCUSSION
The front vehicle seats used in the sled tests underwent enough forward rotation due to their own inertia that there was no substantial interaction with the child restraint after the crash pulse began. Qualitative analysis of publicly available sled tests suggests that the rotational motion of these seats was similar to other seats, although it is very likely that many seats exist which undergo less rotation. Further research is required to quantify the range of stiffnesses in currently manufactured seats, so that the interaction between the front seat and RFCR can be analyzed.
While the Rigid Gap restraint condition does not accurately represent interactions with front vehicle seats due to the high stiffness of the structure, it provides a reasonable approximation of possible contact of RFCR with the vehicle dash. Although vehicle dashes are certainly more compliant, these tests do suggest that RFCR positioned in the front seat of vehicles with an initial gap result in higher injury values. Even contact with front vehicle seats has the potential to increase injury values if the properties of the front vehicle seat are such that a differential velocity develops between the two structures at the time of contact.
The Rigid 0 test condition had the best overall results and illustrates the benefits of early ride down and reduced amounts of rotation. The early onset of acceleration reduced the peak head acceleration values, while the reduction in child restraint rotation decreased loads on the neck. Attachment methods which can rigidly couple the RFCR to the vehicle while reducing restraint rotation may result in decreased injury measures.
The computational models resulted in similar findings. Idealized motions which allowed restraint rotation resulted in higher neck loads, while rigid fixation of the child restraint reduced acceleration levels. The Translation condition had higher injury values than the Rigid condition, although excursion should be able to reduce values by extending the timing of the deceleration. In these idealized cases, the translation spring did not provide any initial “installation force”. If the spring properties modeled those of force limiting belts, high initial stiffness followed by a constant force value, the Translation condition could be optimized to perform better than the Rigid condition. In this case, however, excursion distances should be kept low to prevent contact with other structures. The general goal should be to reduce translation and excursion by more rigidly attaching the restraint to the vehicle.
LIMITATIONS
Although the front vehicle seats used in these tests resulted in no interaction with the child restraints, further testing to characterize the behavior of a larger sample of seats is suggested, particularly seats with integrated belts which would likely have stiffer structural components. The computational models represented idealized child restraint motions and do not take specific CRS and vehicle variability into account. These particular motions are believed to be achievable with different coupling systems, but testing of real-world systems must be performed.
CONCLUSIONS
Attachment methods affect the safety performance of RFCR in frontal crashes by controlling the overall kinematics of the restraint and its interaction with other vehicle structures. The possible interaction with front vehicle seats requires further study, while restraints placed near stiff structures, such as pickup truck dashes, may result in higher injury measures. The ideal attachment method for RFCR, in frontal crashes, reduces rotation and provides high initial translational stiffness to provide early ride down of the crash.
Table 3.
Injury data for Evenflo restraint conditions
| EF No Structure | EF Veh Seat | EF Rigid 0 | EF Rigid Gap | |
|---|---|---|---|---|
| Head Res Acc 3 ms clip (g’s) | 76.7 | 71.9 | 68.6 | 176.1 |
| HIC 36 | 690 | 730 | 348 | 2353 |
| Upper Neck Fx (N) | −371.6 | −339.4 | −290.6 | −346.3 |
| Upper Neck Fz (N) | 1332.3 | 1387.6 | 325.2 | 1832.3 |
| Upper Neck My (Nm) | −12 | −9.9 | −13 | −12.3 |
| Lower Neck Fx (N) | 459.7 | 387.8 | 191.4 | 544.7 |
| Lower Neck Fz (N) | 1468.7 | 1570.6 | 481.5 | 1908.4 |
| Lower Neck Mz (Nm) | −14.3 | −11.7 | −7.2 | −15.2 |
| Chest Res Acc 3 ms clip (g’s) | 55.3 | 54.9 | 59.7 | 77.2 |
ACKNOWLEDGEMENTS
This work was funded by the National Highway Traffic Safety Administration Contract DTRS 57-03-D-30005. The views expressed are those of the authors and not necessarily those of NHTSA.
REFERENCES
- Decina LE, Knoebel KY. Child safety seat misuse patterns in four states. Accid Anal Prev. 1997;29(1):125–132. doi: 10.1016/s0001-4575(96)00051-6. [DOI] [PubMed] [Google Scholar]
- Gotschall CS, Luchter S. Head injuries to motor vehicle occupants aged 0–5 years. Child Occupant Protection in Motor Vehicle Crashes. 1999:17–27. [Google Scholar]
- Kamren B, Koch M, Kullgren A, Lie A, Tingvall C, Larsson S, Turbell T. Protective Effects of a Rearward Facing CRS: An Overview of Possibilities and Problems Associated with Child Restraints for Children 0–3 Years. Child Occupant Protection Symposium Proceedings; Warrendale, PA: SAE; 1993. pp. 113–119. SAE 933093. [Google Scholar]
- Le Claire M, Lowne R, Langwieder K, Hummel Th. Loadings to Children in Cars in Frontal Collisions – Experience from ECE-R44 Test Bench Results Compared with Car Results. Proc. IRCOBI. 2000:335–343. [Google Scholar]
- Lowne R, Roy P, Paton I. A Comparison of the Performance of Dedicated Child Restraint Attachment Systems (ISOFIX). Child Occupant Protection 2nd Symposium Proceedings; Warrendale, PA: SAE; 1997. pp. 71–86. SAE 973302. [Google Scholar]
- Lowne RW, Turbell T. The Development of a Unified Child Restraint-to-Car Attachment System: A Contribution to the ISOFix Discussions. The Fourteenth International Technical Conference on the Enhanced Safety of Vehicles. 1994:1598–1605. Paper No. 94 S10 O 02. [Google Scholar]
- National Highway Traffic Safety Administration. 2005. [Accessed May 27, 2005]. http://dms.dot.gov/search/document.cfm?documentid=243764&docketid=4962; http://dms.dot.gov/search/document.cfm?documentid=243766&docketid=4962.
- van Rooij L, Sherwood C, Crandall J. A Comparison of the Injury Risk for 12-Month-Old Children in Forward and Rearward Facing Seats and the Effect of a Front-row Seat. TNO MADYMO 5th Users’ Meeting of the Americas.2003. [Google Scholar]
- Weber K Child Passenger Protection. Accidental Injury: Biomechanics and Prevention. York, PA: 2002. pp. 529–549. [Google Scholar]