Effects of restraint parameters using PIPER 6y in reclined seating during frontal impact

Abstract Objective This study explores possible challenges for child occupants in reclined seating positions, applying current protection systems. Using PIPER 6 y in frontal impacts, the aim was to investigate the effect of restraint parameters in reclined seating positions, in addition to an upright position, varying booster design, shoulder belt outlet, and pretensioner activation. Method Eighteen full frontal impacts were simulated using the PIPER 6 y human body model restrained on a booster in a front passenger seat. The type of booster, pretensioner activation and shoulder belt outlet were varied with the vehicle seat in ‘upright position’ (25°) and ‘reclined position’ (40°). Three booster principles were used: booster seat (BoosterA), booster cushion (BoosterB) and representing properties of a vehicle built-in booster cushion (BoosterC). The two shoulder belt outlets include ‘nominal D-ring’ and ‘rearward D-ring´. Results Overall, activation of the pretensioner reduced the overall body displacement as well as the head and neck response in both seating positions. Submarining occurred only in the case of BoosterB in ‘reclined position’ without pretensioner. Some differences were observed for the lap belt interaction with pelvis in the non-submarining simulations. Greater pelvis displacement was observed in ‘reclined position’ as compared to ‘upright position’. In both seating positions, greatest pelvis displacement was observed for BoosterB, due to relatively more forward initial lap belt position. While both provided favorable initial lap belt to pelvis contact, BoosterC offered more efficient lap belt restraint than BoosterA, since the lap belt remained lower on the pelvis and the vertical movement of the pelvis was more limited compared to BoosterA. When in ‘reclined position’, the ‘rearward D-ring’ position enabled earlier coupling of the torso due to initial shoulder belt to shoulder contact, resulting in lower head and neck responses as well as shorter head displacement compared to ‘nominal D-ring’ Conclusions Submarining can be addressed in reclined seating positions using current booster design in combination with a seatbelt pretensioner. Lap belt routing was influenced by booster design and reclined seating, affecting the overall kinematics and responses of the PIPER 6 y. This study highlights the importance of including the whole context of child occupant protection when investigating reclined seating, such as the interaction and compatibility of booster, vehicle seat and seatbelt.


Introduction
With increased vehicle automation, consumers may be desiring new seating configurations, such as reclined seating position as the occupants want to relax and sleep in vehicles (Jorl€ ov et al. 2017, € Ostling and Larsson 2019, Koppel et al. 2019. For adults in reclined seating positions, submarining has been identified as one challenge if exposed to frontal impacts. Several studies have addressed countermeasure to reduce the risk of submarining for adult occupants including pretensioners, active submarining restraints and belt in seat ( € Ostling et al. 2017, € Ostling et al. 2021, Richardson et al. 2020a, Rawska et al. 2019, while there are few studies addressing challenges with children in new seating configurations. Maheshwari and Belwadi (2020) conducted a CAE simulation study of upright and reclined seating positions in a rearward facing vehicle seat, using the PIPER 6 y human body model (HBM) restrained on a booster cushion and exposed to frontal impacts. They observed increased ramping up the seat back in a reclined seating position. Levallois et al. (2019) performed a static study highlighting incompatibility when fitting a booster seat into a reclined vehicle seat, as the booster's backrest could not align with the vehicle seat back, resulting in a wide space between the booster and the seat. Hauschild et al. (2021) conducted a sled test series with the large omnidirectional child crash test dummy in upright and reclined seating positions in frontal impacts. They concluded that a booster reduced the risk of submarining in a reclined seating position. In a frontal impact simulation study, Bohman et al. (2021) explored reclined seating positions, in addition to upright position, using the PIPER 6 y restrained in two different boosters, varying use of ISOFIX anchorages and pretensioner activation. Submarining occurred in reclined seating position with PIPER 6 y restrained on a booster cushion without pretensioner activation, whether ISOFIX anchorages were used or not.
Protection of booster-seated children is a combination of the vehicle restraints and the booster. Therefore, with the overall purpose to explore challenges for child occupants in reclined seating positions, this study aimed at investigating some potential influencing restraint parameters in the context of current protection systems. Specifically, the aim was to investigate the effect of restraint parameters in reclined seating, in addition to upright position, varying booster design, shoulder belt outlet, and pretensioner activation by using the PIPER 6 y exposed to a frontal impact.

Method
Eighteen simulations were conducted at 56 km/h, using a generic full frontal rigid barrier pulse ( Figure A1). Simulations were made using LS-DYNA MPP version R9.3.1 (ANSYS/LST, Livermore, CA) with 240 CPUs for a simulation time of 200 ms. The PIPER 6 y human body model (Giordano et al. 2017) was restrained on a booster cushion in a front passenger seat environment of a passenger car. The following four parameters were varied: booster design, vehicle seat back angle, shoulder belt outlet and with or without activation of the vehicle seatbelt pretensioner. See test matrix in Appendix, Table A1.
Three diverse booster designs were used, one with a backrest (BoosterA) and two without (BoosterB and BoosterC), see Figure A2 in Appendix. The three boosters vary in stiffness, belt guide design and with/without backrest, to provide a variation in belt routing. The models of BoosterA and BoosterB represent examples of two types of accessory boosters, while the BoosterC model was developed to represent the belt routing and built-in principles of a vehicle built-in booster. BoosterA was modeled with a deformable structure of a polypropene plastic, deforming 15 mm in crash. BoosterB and BoosterC were modeled with non-deformable shells, allowing only small deformations during crash. The BoosterA model represents an existing booster seat, although not validated in detail. The BoosterB model has been used and validated in a previous study (Bohman et al. 2020). All boosters were covered by a foam surface modeled with solid elements. BoosterC was based on BoosterB, but with removed belt guides, a reduced width by 54 mm and attached to the ISOFIX anchorages. The seatbelt was routed according to the manufacturer's guidelines ( Figure 1) and was done using the preprocessor PRIMER v18.0 (Oasys Ltd, Solihull, UK). The seatbelt system included a load limiter of 4 kN and a pretensioner with time to fire of 10 ms, when activated. An upright seating position corresponding to an H-point manikin torso angle of 25 , and a reclined seating position of 40 torso angle, were simulated, Figure 1. The positions are referred to as 'upright position' and 'reclined position' respectively. The initial position of the seatbelt buckle in BoosterC was rotated 14 and 18 for the upright and reclined seating positions respectively, to obtain proper lap belt position. The PIPER 6 y was positioned in LS-DYNA by using positioning cable nodes and was pulled toward the corresponding booster. The pelvis angle and arm angles were unchanged from the initial position of PIPER 6 y. The lower extremities were positioned when contact with the booster was reached during the presimulation. In the cases with reclined position, PIPER 6 y was rotated as a whole and thus the pelvis angle remained unchanged. The PIPER 6 y and booster were positioned on the seat until equilibrium was reached.
Two shoulder belt outlet positions were simulated to include a variation in shoulder belt geometry. In the 'nominal D-ring' configuration, the D-ring outlet was aligned with vehicle seat head restraint in upright position, as shown in Figure 1, while for the 'rearward D-ring', the outlet was translated 250 mm rearwards and 120 mm downwards, to obtain contact between shoulder and shoulder belt even in 'reclined position'.
The distance between the shoulder and the shoulder belt, defined as a shoulder belt gap, was measured by defining a longitudinal distance from the body at the mid clavicle to the shoulder belt. Jugular notch to shoulder belt distance was used to describe the lateral shoulder belt position and was measured as the lateral distance between the midline of the shoulder belt to the jugular notch. The pelvis angle was defined by measuring the angle as an average of left and right anterior superior iliac spine (ASIS) to the pubic symphysis in the sagittal plane (see Figure A3 in Appendix). The torso angle was defined by the angle between vertical plane and the sternum, measured as the line between lower and upper sternum. Submarining was defined as the lap belt completely moving above the ASIS, and each side was analyzed separately. If partial-only submarining occurred, the side of submarining was specified.
The maximum accelerations and trajectories from head, chest (T1 and T12) and pelvis were extracted, according to PIPER local accelerometers. The neck tension was measured by a cross-section at C2 C3, included in the PIPER 6 y. The chest deflection was measured at the upper, mid and lower sternum.
The kinematics of head, chest (T1 and T11 levels of the thoracic spine) and pelvis were analyzed with respect to displacement and excursion. Displacement was defined as the absolute values in x-and z-direction, while x-and z-excursions were defined relative the vehicle, with the SRP used as the origin of the coordinate system.

Results
Initial posture and beltfit PIPER 6 y had an initial torso angle of 40˚for all three boosters in the 'upright position'. In the 'reclined position', 55˚initial torso angle was obtained when seated in BoosterB and BoosterC, while a more upright torso angle of 47˚was obtained when using BoosterA, as a result of the booster backrest's interaction with the vehicle seat back (Figure 1). The initial pelvis angle for all three boosters was 39 for 'upright position' and a more posterior angle of 48 for 'reclined position'.
Due to the backrest, BoosterA positioned the pelvis more forward on the vehicle seat than BoosterB and BoosterC; 144 mm and 113 mm for 'upright position' and 'reclined position', respectively. Irrespective of booster, the 'reclined position' positioned the pelvis more rearward as compared to the 'upright position' (31 mm for BoosterA, 63 mm for BoosterB, and 22 mm for BoosterC).
The lap belt was positioned below the ASIS for all configurations. BoosterB positioned the lap belt more forward on the thighs, while for BoosterA and BoosterC the initial lap belt position was in initial contact with the pelvis, for both upright and reclined positions. For BoosterA and BoosterC similar initial lap belt positions were observed when in 'reclined position' as compared to 'upright position'. When using BoosterB, the horizontal distance from the lap belt to left ASIS was 35 mm longer than when in 'upright position' (65 mm and 30 mm, respectively), due to the design of the belt guides.
With 'nominal D-ring', the shoulder belt attained an initial mid-shoulder position in the 'upright position' for all three boosters, see Table A1 for detailed distances between jugular notch to shoulder belt. In the 'reclined position', the shoulder belt was slightly closer to the neck for BoosterC. There was initial contact with the shoulder belt (no gap) for all three boosters in the 'upright position'. While in the 'reclined position', no initial shoulder contact was obtained when in BoosterB (gap of 69 mm) and BoosterC (gap of 56 mm).
With 'rearward D-ring', the shoulder belt kept a similar mid-shoulder position for all boosters in 'upright position', while BoosterC had a position closer to the neck in 'reclined position'. With 'rearward D-ring', the shoulder belt had contact with the shoulder for all boosters in both 'upright position' and 'reclined position' (Figures A6-A8).

Shoulder and lap belt interaction
Submarining occurred in the configuration with BoosterB in 'reclined position' and 'nominal D-ring' without pretensioner activation. In this case, the lap belt slipped off both ASIS, and the abdominal pressure resulted in a pressure of 127.6 kPa, (Table A3). In all the other configurations, the lap belt stayed on the pelvis, and thus a lower abdominal pressure was measured, but variation in degree of interaction was observed. Even though no submarining occurred for BoosterA and BoosterC without pretensioner, the lap belt moved up along the pelvis and was partly above both ASIS (see Figure A4) due to delayed coupling of the pelvis by the lap belt. In simulations with pretensioners, the lap belt also had a more upward movement relative the pelvis in 'reclined position' compared to 'upright position', but this lap belt movement was less pronounced with activated pretensioner than for simulations without pretensioner ( Figure A4).
The differences in initial lap belt position for the boosters influenced the coupling of the pelvis. BoosterB had a more forward initial lap belt position, which resulted in a delayed contact between the lap belt and the pelvis during the crash. For the other two boosters, the initial lap belt to pelvis contact was kept during the crash.
For all three boosters, when evaluating the effect of shoulder belt outlet position (pretensioner activation only), the 'rearward D-ring' resulted in the lap belt moving more upwards on the pelvis compared to 'nominal D-ring'. This interaction was most pronounced for BoosterA in 'reclined position' (Figure A4 in Appendix).
The shoulder belt stayed on the shoulder during the whole forward movement phase, in all simulations. The shoulder belt moved more inboard in the 'reclined position' during the crash as compared to when in 'upright position', see Figures A6-A8 in Appendix. This applied for all three boosters.

Kinematics
The head, chest (T1 and T12) and pelvis x-displacements of the PIPER 6 y were longer when the pretensioner was not activated compared to when pretensioner was activated, for BoosterA and BoosterB in both 'upright position' and 'reclined position' (Table A2). The pelvis angle was not influenced by the pretensioner activation when in BoosterA, irrespective of seating position. The pretensioner helped to lower the pelvis angle by 8 from initial pelvis orientation when in BoosterB in both seating positions, and by 10 and 4 for BoosterC, in 'upright position' and 'reclined position', respectively ( Figure A5 in Appendix). All three boosters had greater head x-displacement in 'reclined position' compared to 'upright position'. The head x-displacement was lower in BoosterA in both 'upright position' and 'reclined position' irrespective of D-ring position, compared to the two other boosters (see Table A2). Still, the maximum head excursion was longer in BoosterA compared to BoosterB and BoosterC (see Figure 2). PIPER 6 y had a more forward initial position in BoosterA, due to the thickness of the booster backrest and due to the booster backrest interaction with the vehicle seat back, contributing to longest head excursion relative the vehicle during the crash.
The pelvis x-displacement increased in 'reclined position' compared to 'upright position', for all three boosters and BoosterB had the greatest x-displacement (Table A2). For both seating positions when on BoosterB, the pelvis had an upward excursion compared to the other boosters, while BoosterA and BoosterC had a downward pelvis excursion, being most pronounced for BoosterA (see Figure 1). During the crash event, the pelvis rotated posterior due to the lap belt interaction in BoosterA, see Figure A5. The posterior rotation was similar for BoosterA for both 'upright position' and 'reclined position', but with a smaller posterior rotation for BoosterB and BoosterC in 'upright position' compared to BoosterA. The 'rearward D-ring' resulted in shorter overall x and z displacements of the head in both seating positions, as compared to 'nominal D-ring' position (Figures 3 and 4). The difference in head x-displacement between the two D-ring positions when using BoosterA was 59 mm shorter and the corresponding delta displacements for BoosterB and BoosterC were 108 mm and 99 mm, respectively (Table A2). All three booster had a greater posterior rotation of the pelvis in 'rearward D-ring' compared to 'nominal D-ring'. Of all three boosters, BoosterA had the greatest increase in posterior rotation of 27 for 'upright position' and 30 for 'reclined position', compared to the initial pelvis angle. For BoosterB, the greatest posterior pelvis rotation was observed in the configuration with submarining, Figure A5.

Responses
Activation of the pretensioner generally resulted in lower head, chest and pelvis accelerations as compared to without pretensioner, Table A3. In BoosterA, the head acceleration observed without pretensioner was influenced by the head impacting the knee in one simulation and the booster guide in another simulation. The pelvis acceleration was low with the exemption for the acceleration of 110 g when in BoosterC without pretensioner. This was due to PIPER 6 y moving forward relative the booster, loading the front part of the booster downwards causing a booster to seat structure contact. For all three boosters, the head, chest, and pelvis accelerations were higher in the 'reclined position' compared to the 'upright position', while the opposite was observed for the chest deflections in 'reclined position' compared with 'upright position'. The 'rearward D-ring' generally resulted in lower head, chest, and pelvis acceleration for all boosters, in both seating positions. BoosterA had small differences in head acceleration, and BoosterC had minor differences in the chest acceleration in 'rearward D-ring' compared with 'nominal D-ring'. The chest deflections (upper, mid, lower) for BoosterA was similar in 'nominal D-ring' as in 'rearward D-ring'. BoosterB and BoosterC resulted in lower chest deflections (upper, mid and lower) with 'rearward D-ring position' in both seating positions. In general, the upper and mid chest deflections were higher than the lower and this trend was more pronounced for BoosterB and BoosterC compared to the BoosterA.

Discussions
This exploratory study demonstrated that the PIPER 6 y human body model was well restrained by the seatbelt in the reclined as well as the upright seating position, using a conventional type of booster, given an activated pretensioner. For all three boosters, the lap belt stayed on the pelvis and the shoulder belt stayed on the shoulder with activated pretensioner. Submarining occurred only in one case in the study, BoosterB in 'reclined position' without pretensioner. Irrespective of booster, a more robust lap belt to pelvis interaction was observed when using pretensioner, achieved by tightening the belt and thereby also providing a lower lap belt position on the pelvis resulting in an early engagement of the pelvis.
The lap belt interaction was also influenced by the booster's design. In 'reclined position' the initial pelvis position was more rearward compared to 'upright position'. BoosterA and BoosterC offered an initially more rearward position of the lap belt, as compared to BoosterB, resulting in an earlier coupling of the pelvis and thereby less pelvis forward displacement during the crash. This early coupling influenced both kinematics and responses favorably for both seating positions when using BoosterA and BoosterC.
Comparing BoosterA and BoosterC, despite similar pelvis displacement, BoosterC had a lap belt engagement lower on the pelvis, which resulted in less pronounced pelvis posterior rotation and thereby a more efficient lap belt interaction. Limiting the pelvis posterior rotation may reduce the risk of submarining (Richardson et al. 2020b). In addition, BoosterA moved more downwards during the crash which contributed to a lap belt interaction higher up on the pelvis during the crash. This was partly due to its initially more forward position on the vehicle seat, due to the thickness of the backrest and the backrest's incompatibility with the vehicle seat's head restraint, and that it thereby interacted with other parts of the vehicle seat cushion. In a user study by Baker at al. (2021) investigating a wide range of boosters, similar trend was observed with a more forward position with children seated on a booster seat compared to a booster cushion, with a difference up to 154 mm in supra sternal x-position. Downwards movement during impact may also be influenced by the booster stiffness, as shown by Forman et al. (2021 andBohman et al. (2020).
The different booster designs controlled the lap belt interaction by either the booster's belt guides or by the design of the booster height and width together with the vehicle belt geometry. Durbin et al. (2003) demonstrated that boosters nearly eliminate the risk of abdominal injuries in nominal seating positions, in the age group 4-7 years. Efficient boosters are designed to help provide children a proper lap and shoulder beltfit, with the purpose to reduce risk of submarining. The current study shows that boosters can be efficient even in reclined seating, especially when used together with a pretensioner, although the boosters were not designed for reclined seating position. This puts child occupants in a more favorable situation when moving toward a possible increased exposure for reclined seating, as compared to adult occupants likely requiring additional restraints beyond what is commonly used in cars today (Gepner et al. 2019a, Rawska et al. 2019. One of the key parameters found in the current study was to control the lap belt to pelvis interaction, achieved by a low initial lap belt geometry, together with seatbelt pretensioner and limited booster displacements during crash.  The head x-displacement was smallest with activated pretensioner, as compared to without, for all the three boosters. The greatest head x-displacement was observed when using BoosterB in the 'reclined position' and with 'nominal Dring'. However, the longest head x-excursion relative the car was observed when in 'upright position' using BoosterA. This can be explained by the more forward initial position, which was due to the static incompatibility of the booster's backrest and the vehicle seat back, which limited the booster seat's possibility to recline to the same extent as the vehicle seat back. This incompatibility together with the thickness of the booster's backrest positioned the head more forward than when in BoosterB or BoosterC, in 'upright position'. The influence of the two shoulder belt outlet positions had greatest effect for BoosterB and BoosterC. An initial shoulder belt to shoulder contact was seen in 'rearward D-ring', while a gap was seen in 'nominal D-ring'. The initial contact enabled an earlier coupling to the torso, resulting in lower responses to the head and neck, as well as shorter head displacement. When using the BoosterA, these differences were less, due to initial shoulder belt contact for both shoulder belt outlet positions. Furthermore, for all three boosters, the 'rearward Dring' resulted in the lap belt moving slightly more upward the pelvis, leading to a larger posterior pelvis angle as compared to 'nominal D-ring'. Hence, although an initial shoulder belt contact is desired, more parameters need to be addressed to achieve a desired balance of pelvis and torso interaction, including the controlled upper body forward movement (torso pitch), even in reclined seating positions.
This study included three different designs of boosters installed in one front passenger seat, and it resulted in different initial beltfit and different kinematics. Baker et al. (2021) studied static beltfit of ten different boosters, both with and without backrest, showing a range of different lap and shoulder beltfit. Bohman et al. (2020) exposed three types of booster cushions to frontal impacts, showing poorer shoulder belt interaction with boosters with deformable characteristics. Forman et al. (2021) studied 24 parameters potentially influencing booster performance in nominal seating and they identified soft, deformable booster as one major contributor for increased risk of submarining. The boosters in the current study differed both with respect to having a backrest or not, and by different designs of the lap belt guides. These two parameters could not be separated clearly in this study, due to the 8 degrees more upright posture of the PIPER 6 y due to the booster seat backrest's interaction with the vehicle seat back, when in the 'reclined position'. Hence, there is a need to study a greater variation of booster designs and vehicle seat positions, and to further identify boosters design features which are relevant for occupant protection in reclined seating positions, while monitoring the potential effect when used in upright seating positions.
Pelvis orientation and pelvis position in relation to the lap belt are essential when evaluating risk of submarining. In the current study, the PIPER 6 y was positioned based on engineering judgment, lacking representative data of 6-yearolds. There is a need to quantify the initial posture in reclined seating positions for real children, especially gathering data on the ranges of pelvis positions and orientations. Furthermore, there were technical challenges when positioning the PIPER 6 y model. Due to lack of an LS-Dyna positioning tree in PIPER, difficulties in orienting the pelvis angle were experienced. Additional work is also encouraged in this area to enable positioning of the PIPER 6 y in other software environments.
The PIPER 6 y was not developed for reclined seating position. There are ongoing efforts worldwide to develop and validate tools for adult occupants in reclined seating positions (Richardson et al. 2020a, € Ostling et al. 2021. Similar activities are also needed for the pediatric tools, to address the upcoming challenges with reclined seating for child occupants as well.
As child occupant protection is being developed to meet future challenges, this likely needs to address reclined as well as upright seating positions. This study highlights the importance of including the whole context of child occupant protection when addressing reclined seating. The design of booster and vehicle seat both influences their capability to fit together. This is exemplified in this study by BoosterA, which would not allow for the same sitting posture as the other boosters, due to the interaction between backrest and vehicle seat back. As an exploratory study, the three selected boosters offering some diverse belt geometries and sitting postures served as a first step. Additional parameters should be investigated in further studies.
The benefits of the seatbelt pretensioner was evident. The influence of shoulder belt outlet position was shown, both with respect to the shoulder interaction by the two shoulder belt geometries, but also the lap belt routing by the different booster designs. Although reclined seating is challenging, the same principles of protection apply, as for upright seating. It includes an early and tight contact between the lap belt and the pelvis, and to keep this contact during the whole crash (Adomeit and Haeger 1975;Adomeit 1977). Furthermore, it includes a robust coupling of the torso by the shoulder belt, also allowing a balanced torso retention enabling a torso pitch, to help limit the posterior rotation of the pelvis.
In conclusion, within the limitations of this study it could be seen that submarining can be addressed in reclined seating using current booster design in combination with a vehicle seatbelt pretensioner. Overall, the pretensioner contributed to a more robust lap belt to pelvis interaction. Lap belt interaction with pelvis was influenced by booster design features and seating position, affecting the overall kinematics and responses of the PIPER 6 y human body model. Although child occupant protection in reclined seating is more challenging than upright seating, the same principles of protection apply, and the optimal protection is achieved by allowing the vehicle and the child restraint system to work together.

Disclosure statement
Katarina Bohman, Sarah El-Mobader and Lotta Jakobsson are employees of Volvo Car Corporation.