Car occupant seat belt fit; the effect of belt pre-pretensioning Master’s thesis in Biomedical Engineering Louise Bohl and Klara Eliasson Department of Mechanics and Maritime Sciences CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2023 www.chalmers.se www.chalmers.se Master’s thesis 2023 Car occupant seat belt fit; the effect of belt pre-pretensioning Louise Bohl and Klara Eliasson Department of Mechanics and Maritime Sciences Division of Vehicle Safety Chalmers University of Technology Gothenburg, Sweden 2023 Car occupant seat belt fit; the effect of belt pre-pretensioning Louise Bohl and Klara Eliasson © Louise Bohl and Klara Eliasson, 2023. Supervisor: Amanda Hederskog, Autoliv Martin Östling, Autoliv Examiner: Johan Davidsson, Mechanics and Maritime Sciences Master’s Thesis 2023 Department of Mechanics and Maritime Sciences Division of Vehicle Safety Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: The test rig used to investigate shoulder belt repositioning Typeset in LATEX, template by Kyriaki Antoniadou-Plytaria Printed by Chalmers Reproservice Gothenburg, Sweden 2023 iv Car occupant seat belt fit; the effect of belt pre-pretensioning Louise Bohl and Klara Eliasson Department of Mechanics and Maritime Sciences Chalmers University of Technology Abstract In the event of a crash, the seat belt should load the occupant’s pelvis, thorax, and clavicle. A shoulder belt segment routed distal of the shoulder, i.e. positioned on the arm, may cause chest and abdominal injuries during a crash. The overriding aim of this study was to investigate if an improperly positioned shoulder belt can be repositioned to a proper position on the clavicle, with the help of a pre-pretensioner for front seat occupants. More specifically, the aims were to investigate if the location of the belt attachment points, occupant body characteristics, belt geometry, belt fit, and friction of clothing affected the ability of the pre-pretensioner to reposition the shoulder belt and from which distances down the arm it was possible. A volunteer study was conducted to investigate if the shoulder belt could be repo- sitioned for a nominal belt geometry similar to a Volvo S60 and for a belt-in-seat geometry in an adopted test rig. Several anthropometric, belt geometry, and belt fit measurements were collected and analyzed to identify why the belt did not repo- sition for some individuals compared to others. In addition, the ability of the pre- pretensioner to reposition the shoulder belt for different fore-aft seat positions, D- ring heights, and a belt-in-seat installation were investigated. 17 male and 18 female volunteers were tested. The study found that the location of the belt attachment points affected belt repositioning, since the shoulder belt was not repositioned for the majority of the volunteers in the belt-in-seat installation. The belt repositioned for all volunteers in the most common seat positions while the rate of unsuccessful repositionings increased for more forward seat positions. A high D-ring made belt repositioning possible for all volunteers. Measurements iden- tified as influencing belt repositioning were a taller shoulder height (measured while seated) and a smaller abdominal depth in seat positions forward of the mid position and for the belt-in-seat installation. The belt did not reposition with the lower fric- tion clothing material in the belt-in-seat installation but repositioned for some in a forward fore-aft position. The repositioning commonly failed from positions close to the acromion on the arm. The results indicate that the upper body shape influences belt repositioning. It could be linked to combinations of upper body measurements, shoulder belt routing, and different belt geometries. Based on the results, future studies should investigate shoulder belt repositioning for additional belt geometries and dynamic scenarios as well as the possibility to implement belt fit warning interventions. Keywords: belt repositioning, B-pillar installation, belt-in-seat installation, pre- pretensioner, D-ring attachment, and fore-aft position. v Acknowledgments We are immensely grateful for all the guidance and help we have received from our supervisor Amanda Hederskog at Autoliv. Thank you for your unwavering support, the hours you have spent discussing all aspects of this project with us, and giving constructive feedback on this report. We would also like to thank Martin Östling from Autoliv for the support and setting up the meetings with the project group "Passenger safety to the next level" who set the foundation for the thesis. A special thanks to our examiner Johan Davidsson for all his time, input, and advice. This study would not have been possible without all the practical help from the staff at Autoliv, and we especially want to express our gratitude to Mikael Enänger and Henrik Hermansson. We are extremely thankful for your practical wisdom, the help you have given us, and the way you always brightened up our days. Louise Bohl and Klara Eliasson, Gothenburg, June 2023 vii List of Acronyms Below is the list of acronyms that have been used throughout this thesis listed in alphabetical order: ASIS Anterior Superior Iliac Spine BMI Body Mass Index df degrees of freedom SD Standard Deviation ix Contents List of Acronyms ix List of Figures xv List of Tables xix 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.4 Ethical aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Developing test procedure 5 2.1 Overview of method . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 The test rig . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.3 Parameters influencing shoulder belt repositioning . . . . . . . . . . . 8 2.4 Parameter observations . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.4.1 Belt installation . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.4.2 D-ring height . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.4.3 Fore-aft position . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.4.4 Seat back angle . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.4.5 Seat height . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.4.6 Occupant clothing . . . . . . . . . . . . . . . . . . . . . . . . 12 2.4.7 Occupant posture . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.4.8 Force level of pre-pretensioner . . . . . . . . . . . . . . . . . . 13 2.4.9 Duration time of pre-pretensioner . . . . . . . . . . . . . . . . 14 2.5 Summary of the observations on parameters . . . . . . . . . . . . . . 14 2.6 Preliminary test conditions . . . . . . . . . . . . . . . . . . . . . . . . 15 2.7 Incremental process to find a potential point-of-no-return . . . . . . . 16 2.8 Definition of a repositioned shoulder belt . . . . . . . . . . . . . . . . 19 2.9 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.9.1 Anthropometric measurements . . . . . . . . . . . . . . . . . . 20 2.9.2 Belt geometry and belt fit measurements . . . . . . . . . . . . 22 2.9.2.1 Belt geometry angles . . . . . . . . . . . . . . . . . . 22 2.9.2.2 Belt fit distances . . . . . . . . . . . . . . . . . . . . 24 xi Contents 2.9.2.3 Belt wrapping distances . . . . . . . . . . . . . . . . 24 2.9.3 Retracted belt webbing and retraction time . . . . . . . . . . 25 2.10 Pilot study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.10.1 Evaluation of the pilot study . . . . . . . . . . . . . . . . . . . 26 3 Methods for volunteer study 27 3.1 The volunteers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.1.1 Volunteer distribution . . . . . . . . . . . . . . . . . . . . . . 28 3.2 Final test conditions for volunteer study . . . . . . . . . . . . . . . . 28 3.3 Test rig instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.4 The test procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.4.1 Volunteer preparation, instrumentation, and data collection . 30 3.4.2 Conducting test procedure . . . . . . . . . . . . . . . . . . . . 32 3.5 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.5.1 Statistical significance by two sample t-test with Bonferroni correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.5.2 Binary logistic regression . . . . . . . . . . . . . . . . . . . . . 35 4 Results 37 4.1 Volunteers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.2 Shoulder belt repositioning in the B-pillar installation . . . . . . . . . 38 4.3 Shoulder belt repositioning in the belt-in-seat installation . . . . . . . 41 4.4 Two sample t-test to test statistical significance . . . . . . . . . . . . 43 4.4.1 Two sample t-test and analysis of collected measurements at the second most forward test condition in the test rig . . . . . 43 4.4.2 Two sample t-test and analysis of collected measurements at the most forward test condition in the test rig . . . . . . . . . 47 4.4.3 Two sample t-test for belt-in-seat installation . . . . . . . . . 49 4.5 Logistic regression curves for belt-in-seat installation result . . . . . . 51 5 Discussion 55 5.1 Shoulder belt repositioning in the B-pillar and belt-in-seat installation 55 5.2 Effect of upper body characteristics, belt geometry, and belt fit on belt repositioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 5.3 Effect of friction on belt repositioning . . . . . . . . . . . . . . . . . . 60 5.4 Outliers at the test conditions where belt repositioning failed . . . . . 60 5.5 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 5.6 Evaluation of test procedure . . . . . . . . . . . . . . . . . . . . . . . 65 5.7 Excluded volunteers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 5.8 The data analysis methods . . . . . . . . . . . . . . . . . . . . . . . . 66 5.9 Suggestions for future research on shoulder belt repositioning . . . . . 67 5.10 Observations from performing the volunteer study . . . . . . . . . . . 68 6 Conclusion 71 Bibliography 73 xii Contents A Volunteer measurements I B Summary of volunteer measurements V C Failed belt repositioning images of volunteers at -346 mm VII D Failed belt repositioning images of volunteers at -387 mm IX E Failed belt repositioning images of volunteers at the belt-in-seat installation XI xiii Contents xiv List of Figures 2.1 The test rig and the coordinate system, with the origin in one H-point. 6 2.2 The D-ring range for the B-pillar installation and the fore-aft range in the test rig. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.3 The belt-in-seat installation . . . . . . . . . . . . . . . . . . . . . . . 8 2.4 A visualization of the car seat at the different fore-aft positions for the preliminary test conditions. The percentages are forward of the rearmost fore-aft position in the test rig, calculated relative the track length of a Volvo S60. The distances in millimeter are given rela- tive the D-ring attachment and the H-point. The final name is a description of its position in the fore-aft range. . . . . . . . . . . . . . 11 2.5 A photo of how the shoulder was covered with the lower friction ma- terial used in the study. . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.6 Belt fit in four postures with B-pillar installation. . . . . . . . . . . . 13 2.7 The position of the acromion, marked in red, from a frontal and side view of the shoulder. . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.8 The incremental steps down the arm, measured from the acromion process, with the B-pillar installation. . . . . . . . . . . . . . . . . . . 17 2.9 A flowchart of the incremental process used to analyze belt reposition- ing and identify the point-of-no-return if the belt fails to reposition. The blue/green boxes show the distance where the pre-pretensioner was activated. The orange rhombuses show the next step based on the outcome from the green boxes. The dark blue boxes were the last test at the identified point-of-no-return with a material with lower friction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.10 An illustration of how the distances down the arm were measured and fastened using masking tape in the B-pillar installation. The distance was measured from the acromion process to the upper edge of the belt. 19 2.11 An illustration of how the distances down the arm were measured and fastened using masking tape in the belt-in-seat installation. The distance was measured from the acromion process to the upper edge of the belt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.12 The caliper and the wooden caliper with spikes, both used to measure anthropometric measurements. . . . . . . . . . . . . . . . . . . . . . . 21 xv List of Figures 2.13 Illustration of the belt geometry angles and how they were measured using the digital protractor. The angle is presented in the first three figures and the corresponding measurement method in the three below. 23 2.14 An illustration of the measured belt angle in the XY-plane and how it was measured in both belt belt installations. . . . . . . . . . . . . . 23 2.15 Illustration of the two belt fit measurements, vertical belt distance and horizontal belt distance, and how they were measured between the suprasternal notch to the edge of the belt. . . . . . . . . . . . . . 24 2.16 The belt wrapping distance in the x-axis, from the acromion process to the outboard edge of the belt and how it was measured for the B-pillar installation. The wrapping distance is not measured if the belt covers the acromion process for the belt-in-seat installation. . . . 25 3.1 A visualization of the car seat at the different fore-aft positions for the test conditions. The percentages are forward of the rearmost fore- aft position in the test rig, calculated relative the track length of a Volvo S60. The distances in millimeter are given relative the D-ring attachment and the H-point. The final name is a description of its position in the fore-aft range. . . . . . . . . . . . . . . . . . . . . . . 29 3.2 Placement of the two Gopro cameras. . . . . . . . . . . . . . . . . . . 30 3.3 The photos taken of each volunteer. . . . . . . . . . . . . . . . . . . . 31 3.4 Testing area setup to take anthropometric measurements and photos. 32 4.1 Illustration showing to what extent the belt repositioned for the vol- unteers in the different fore-aft positions. All volunteers are depicted in gray whereas the males and females are depicted in blue and orange respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.2 The mean wrapping distance for all volunteers and the males and females separately in the different fore-aft positions. . . . . . . . . . . 39 4.3 The mean D-ring angle for males and females in the different fore-aft positions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 4.4 An overview of the belt sitting by itself without tape at the different positions in the fore-aft range. . . . . . . . . . . . . . . . . . . . . . 41 4.5 Illustration showing the relation between repositioned and not repo- sitioned shoulder belts in the belt-in-seat installation, and the corre- sponding points-of-no-return at 5 cm or 10 cm that belt repositioning failed for. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.6 The percentage of belt repositioning between the males and females at the identified point-of-no-return for the males and females. . . . . 42 4.7 An overview of the collected measurements that were statistically significant after Bonferroni correction in a two sample t-test for belt repositioning from 10 cm in the B-pillar installation at the second most forward test condition in the test rig, at -346 mm. . . . . . . . 45 4.8 The extra anthropometric measurements where the two volunteers that the belt did not reposition for were distinguishable compared to the other volunteers in the second most forward test condition in the test rig. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 xvi List of Figures 4.9 An overview of the collected measurements which were not statisti- cally significant after Bonferroni correction according to the two sam- ple t-test for belt repositioning from 10 cm in the B-pillar installation at the most forward position in the test rig, at -387 mm. . . . . . . . 48 4.10 An overview of the collected measurements which were statistically significant after Bonferroni correction according to the two sample t-test for belt repositioning from 10 cm in the belt-in-seat installation. 50 4.11 The probability curve of the belt not repositioning depending on shoulder height (sitting), abdominal depth, and D-ring angle for the belt-in-seat installation. . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.12 The probability curves of the belt not repositioning depending on three of the collected measurements: shoulder height (sitting), ab- dominal depth, and D-ring angle, in regard to three percentile values of the volunteer data for the belt-in-seat installation. . . . . . . . . . 53 5.1 Belt routing across the abdomen for two volunteers with different abdominal depths in the belt-in-seat installation, positioned 10 cm down the arm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 5.2 The outliers identified at the second most forward test condition in the test rig, at -346 mm. . . . . . . . . . . . . . . . . . . . . . . . . . 61 5.3 The two outliers identified at the most forward test condition in the test rig, at -387 mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 5.4 The three outliers identified at the belt-in-seat installation. . . . . . . 64 C.1 The shoulder belt position when positioned 10 cm down the arm, be- fore, and the position after the pre-pretensioner was activated (after). The after image is taken when the belt pre-pretensioner held the belt with a constant force in an improper position for the volunteers. . . . VII D.1 Images displaying the before and after for 4 out of the 7 volunteers the belt did not reposition for. The shoulder belt position when positioned 10 cm down the arm, before, and the position after the pre-pretensioner was activated for the most forward position in the test rig. The before images are depicted in the top row while the corresponding after images are depicted in the bottom row. The after image is taken when the belt pre-pretensioner held the belt with a constant force in an improper position for the volunteers. . . . . . . . IX D.2 Images displaying the before and after for the remaining 3 out of the 7 volunteers the belt did not reposition for. The shoulder belt position when positioned 10 cm down the arm, before, and the position after the pre-pretensioner was activated for the most forward position in the test rig. The before images are depicted in the top row while the corresponding after images are depicted in the bottom row. The after image depicts when the belt pre-pretensioner held the belt with a constant force in an improper position for the volunteers. . . . . . . X xvii List of Figures E.1 Images displaying the before and after for 4 out of the 25 volunteers the belt did not reposition for. The shoulder belt position when positioned 10 cm down the arm, before, and the position after the pre-pretensioner was activated in the belt-in-seat installation. The before images are depicted in the top row while the corresponding after images are depicted in the bottom row. The after image depicts when the belt pre-pretensioner held the belt with a constant force in an improper position for the volunteers. . . . . . . . . . . . . . . . . . XI E.2 Images displaying the before and after for 4 out of the 25 volunteers the belt did not reposition for. The shoulder belt position when positioned 10 cm down the arm, before, and the position after the pre-pretensioner was activated in the belt-in-seat installation. The before images are depicted in the top row while the corresponding after images are depicted in the bottom row. The after image depicts when the belt pre-pretensioner held the belt with a constant force in an improper position for the volunteers. . . . . . . . . . . . . . . . . . XII xviii List of Tables 2.1 The parameters influencing shoulder belt repositioning that were iden- tified along with the available options and settings for each parameter. 9 2.2 A summary of the parameters that were prioritized and fixed along with the final option and the total number of parameters (#). Cloth- ing was included for the tests when the belt did not reposition, thereby the (+1) sign. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3 The preliminary test conditions for the volunteer study. The fore-aft position is presented in distance of the D-ring attachment relative the H-point in the x-axis. The test conditions in parenthesis were only conducted if belt repositioning failed in the test with the low D-ring. 15 2.4 The anthropometric measurements measured of all volunteers that could be connected with belt repositioning . . . . . . . . . . . . . . . 21 3.1 Three percentile ranges with the corresponding measurement for three anthropometric measurements defining the Swedish population. All measurements are given in millimeter. . . . . . . . . . . . . . . . . . . 28 3.2 The final tests for the volunteer study. The fore-aft position is pre- sented in the x-axis distance of the D-ring attachment relative the H-point. The test conditions in parenthesis were only conducted if belt repositioning failed in the test with the low D-ring. . . . . . . . . 29 3.3 The denotations for sample mean, variance, standard deviation, and the sample size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.1 The number of volunteers that fit into the three percentile ranges for three anthropometric measurements presented in Section 3.1.1 of the Swedish population. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.2 A summary of the collected measurements at the second most forward position in the test rig, at -346 mm, that were statistically significant after Bonferroni correction between the two groups the volunteers that the belt repositioning repositioned for and not. All measure- ments are given in millimeter. . . . . . . . . . . . . . . . . . . . . . . 44 4.3 A summary of the data for the smallest p-values found after the two sample t-test analysis of the two groups: repositioned and not repo- sitioned, of the result obtained at -387 mm. All measurements are given in millimeter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 xix List of Tables 4.4 A summary of the data for the smallest p-values found after the two sample t-test analysis of the two groups: repositioned and not repo- sitioned, of the result obtained in the belt-in-seat. All measurements are given in millimeter expect the D-ring angle that is given in degrees. 49 4.5 The variables used as input to create the binary logistic regression curves. The input for the one independent variable and two indepen- dent variables are separated by the mid line. . . . . . . . . . . . . . . 51 B.1 A summary of all anthropometric measurements that were taken pre- sented for all volunteers. All measurements above the midrule were taken while standing and the ones below while seated. All data is given in millimeter except age and weight that is given in years and kilograms respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . VI xx 1 Introduction The 3-point seat belt is a restraint system implemented to reduce the injury risk of occupants in a crash. The seat belt is the standard occupant protection system in vehicles and it was implemented to reduce the motion of the occupant in the occur- rence of a crash (Kahane, 2015, Schoeneburg et al., 2011). It has been estimated that seat belts saved more than 12,000 lives annually between 2013 and 2017 in the USA, according to NHTSA’s National Center for Statistics and Analysis (2019). The shoulder belt should be positioned in the center of the clavicle according to a group of experts that defined proper belt fit for a belt fit intervention study con- ducted by Buckley et al. (2018). The lap belt should be positioned on the hips where it is in contact with the thighs and it should also be kept as tightly fitted as possible (Buckley et al., 2018). Seat belt routing is essential to safety and to ensure correct loading of the body. A study by Bohman et al. (2019) indicated that improper seat belt fit is an ongoing issue. The study showed that improper seat belt fit occurs among both the young and elderly participants, although most common among the elderly. Recently, a naturalistic driving study discovered that front seat passengers wear the shoulder belt down on the arm 22% of the time (Reed et al., 2020). Since an improper shoulder belt fit can increase the risk of injury, it is essential to investigate if an improper shoulder belt fit can be corrected into a proper shoulder belt fit to increase occupant safety. 1.1 Background The electrical pre-pretensioner is a safety system activated in the pre-crash phase with the aim to reduce the injury risk by improving the occupant’s seat position (Tobata et al., 2003). The pre-pretensioner retracts the seat belt when an unavoid- able crash has been identified or when the driver executes an emergency braking (Fujita et al., 2003). Thereby, the pre-pretensioner can move the occupant to an upright position against the seat back and restrict the occupant’s motion (Fujita et al., 2003). Mages et al. (2011) explains that the pre-pretensioner can limit the displacement of the occupant since it activates prior to the accident and thereby in- creases the effectiveness of the seat belt. The electrical pre-pretensioner is reversible and the retraction mechanism can be used repeatedly (Fujita et al., 2003). The 1 1. Introduction electrical pre-pretensioner is also known as an electrical reversible pretensioner, an electric pretensioner, a motorized shoulder belt tensioner, and a reversible seat belt tensioner, and is henceforth referred to as the pre-pretensioner. The pre-pretensioner was first launched in 2002 by Mercedes-Benz in a safety system called PRE-SAFE® (Schoeneburg and Breitling, 2005). The aim of the PRE-SAFE® system is to reposition the occupant in the seat during the pre-crash phase and it is activated by the braking behavior of the driver (Schoeneburg and Breitling, 2005). The pre-pretensioner can also be activated by sensors that identify an unavoidable crash situation (Fujita et al., 2003). Schoeneburg et al. (2011) explain that the PRE-SAFE® is activated when the time-to-collision is 1.6 seconds. The pre-pretensioner has been the focal point of several studies regarding occu- pant displacement since it was introduced. Fujita et al. (2003) compared the chest displacement in a head-on collision with and without a pre-pretensioner, and the results showed that the displacement was reduced by around 40 mm. Another test with a pre-pretensioner was conducted in a braking scenario with test subjects (Schoeneburg and Breitling, 2005). The results indicated that a vehicle with a pre- pretensioner was able to reduce both head and chest displacement of the occupant when compared to a vehicle without a pre-pretensioner. Mages et al. (2011) also performed studies on occupant displacement, both during an automatic emergency brake as well as during a double lane change event. During the automatic emergency brake scenario, the head and chest displacement were both reduced by around 40% when the pre-pretensioner was activated simultaneously as the automatic emergency breaking compared to a standard belt system without a pre-pretensioner. The dou- ble lane change event also showed reduced displacement of the chest and head when the pre-pretensioner was activated. As described above, the pre-pretensioner has been proven useful in different braking scenarios since it has reduced occupant displacement. Since the study by Reed et al. (2020) indicated that front seat occupants wear the belt down the arm 22% of the time, it is of interest to investigate means to reposition the shoulder belt from an improper shoulder belt fit to a proper. To the best of the authors’ knowledge, no study has been conducted on the pre-pretensioner’s ability to turn an improper shoulder belt fit into a proper shoulder belt fit. Neither has a corresponding test pro- cedure analyzing belt repositioning been developed. What is yet to be investigated is to what extent the shoulder belt can be repositioned when initially positioned off-shoulder. 2 1. Introduction 1.2 Objectives The objective of this thesis was to investigate for which positions on the arm a seat belt system, equipped with a pre-pretensioner, could reposition the shoulder belt to an on-clavicle position in a stationary car seat. The correlating research questions were as follows: • Does the seat belt installation affect the possibility to reposition the shoulder belt? • From which levels on the arm can the shoulder belt be repositioned? • Does upper body characteristics, belt geometry, and belt fit affect the possi- bility to reposition a shoulder belt? • Does friction of clothing affect the possibility to reposition the belt? 1.3 Delimitations The focus of this study was shoulder belt repositioning since the naturalistic driving study by Reed et al. (2020) indicated that front seat occupants more commonly wore the shoulder belt incorrectly down the arm than the lap belt incorrectly on the abdomen. The shoulder belt fit was analyzed in an upright standard seating position with pre-determined seat configurations. The belt repositioning testing was carried out in a stationary test rig, as this was an initial study of shoulder belt repositioning. Sled testing and dynamic testing were therefore excluded since it was considered a possible continuation depending on the thesis’ results. To narrow down and prioritize the parameters, only variations in the fore-aft position and D-ring setting were included in this study. Due to the geometrical boundaries of the test rig several parameters that could affect belt repositioning were kept fixed. These parameters include lap belt attachments and an additional pre-pretensioner. The lap belt attachments consist of the buckle and anchor, which are attached to the seat and therefore kept fixed. A lap belt pre-pretensioner was a possible addition. However, it was expected to mainly affect the lap belt fit, which was outside of the scope of this study, and therefore a lap belt pre-pretensioner was excluded from the study. 3 1. Introduction 1.4 Ethical aspects The aim of this thesis was to improve vehicle safety, reduce the risk of injuries of occupants who are part of a car crash, and thereby save lives. Society can benefit from the result of this thesis since it is meant to improve the general knowledge of belt repositioning and can increase the understanding of belt repositioning in the research area of vehicle safety. Further research could be conducted on this topic if the thesis discovered that belt repositioning is problematic when the shoulder belt is positioned below the shoulder. The study protocol was reviewed and approved by the Swedish Ethical Review Authority Application 2023-01920-01. Conducting this study with volunteers was crucial to get authentic, reliable, and trustworthy results that cannot be achieved by performing the same tests on crash test dummies or through simulations. Crash test dummies have a limited variety of shoulder shapes and torso heights, and although simulations can include multiple body shapes the volunteers have more trustworthy responses than the simulations. Volunteers were therefore the most suitable choice for investigating belt reposition- ing. However, volunteer testing comes with ethical aspects regarding consent and testing. The main ethical aspect to consider was regarding the volunteers participating in the study since anthropometric measurements, photos, and videos of them were taken. It was important that the volunteers consent to their data being collected and kept for the purpose of this study, and possibly saved for future analysis by employees at Chalmers. Volunteer data was kept on a hard drive which was placed in a safe at Chalmers when the study was completed. Another important ethical aspect was that the volunteers participate by their own volition. All volunteers were informed of what the study would entail before they participated through a document explaining the test procedure and what data would be saved. They were given the option to decline the invitation if they decided they did not want to participate and withdraw at any time throughout the testing. The testing area where the volunteer study was conducted was made as secluded as possible. Having a secluded testing area was important to ensure the anonymity of the volunteer throughout the study. It also helped keep the volunteers comfortable and ensured that non-authorized personnel did not enter the testing area. 4 2 Developing test procedure A literature study was conducted on volunteer studies within the research field of belt repositioning and belt fit. A test method to investigate belt repositioning had to be developed due to the lack of previous studies in the subject. The aim of the literature study was to understand the general setup and the important aspects of a volunteer study. When an understanding of the test procedure and the different aspects had been identified, they were implemented to create the volunteer study developed in this thesis. Meetings with a project group within the vehicle safety field, part of "Passenger safety to the next level", were conducted with representatives from Autoliv, Volvo, and Chalmers. The input from these meetings provided support in design decisions and helped form the volunteer study. The discussions and input mainly focused on parameter prioritization, how to perform a volunteer study, and standardized anthropometric measurements. In addition to the literature study and the meetings with the project group, the authors of this thesis participated in a research study in the vehicle and traffic safety field. The research study was conducted by a researcher at Chalmers. The experience was insightful and used as a foundation when developing and carrying out the volunteer study of this thesis. 2.1 Overview of method In the upcoming sections, the developing process of the volunteer study is presented. During this initial phase the test rig was investigated and necessary changes were made. The parameters influencing belt repositioning were identified and narrowed down by carrying out several different tests in the rig. The parameters were narrowed down to create the test conditions for the volunteer study. The participants in these tests were mainly the authors of this thesis and a few employees at Autoliv. The observations, motivations, and decisions made during the development phase are presented as part of the method since it laid the foundation for the test cases needed for the volunteer study. Based on the findings, the test conditions were assembled and the method of investigating belt repositioning was created. The method consists of incremental steps of shoulder belt repositioning to determine a specific region, called point-of-no-return, from where the belt does not reposition. 5 2. Developing test procedure In addition to investigating if the belt repositioned or not, several sets of data were collected from both the volunteers and the test rig. The data sets from the volunteers were anthropometric, belt geometry, and belt fit measurements whereas the data from the test rig consisted of belt retraction speed and time. A pilot study was conducted once the parts of the test procedure had been finalized. A summary of the observations made in the pilot study is presented. The next chapter presents a summary of the method developed for the volunteer study, where the final test procedure and test rig instrumentation along with the methods used for the data analysis is presented. 2.2 The test rig The test rig used to analyze belt repositioning was provided by Autoliv and consisted of a leather car seat (Volvo V60 2012) mounted on a movable aluminum rig, see Figure 2.1. The seat belt seen to the left in the figure was used in this study and it corresponds to a passenger seat in a right-hand traffic vehicle. The figure also depicts the coordinate system used for vehicles, where one H-point was set as origin. In order to relate the fore-aft position between the D-ring and the seat, the H-point was measured out with a SAE H-point mannequin when the seat torso angle was set to 25◦. This H-point will be used as a reference throughout the thesis to describe the D-ring attachment relative the seat position. The x-axis is the direction that the car travels, the y-axis is the lateral direction, and the z-axis is the vertical direction. The D-ring attachment is the part which the belt extends from, and it was mounted in the pillar, see Figure 2.2b. The D-ring was attached with a cap screw which it could rotate around. The fore-aft range is the range which the car seat can be moved in the x-direction, see Figure 2.2a. Figure 2.1: The test rig and the coordinate system, with the origin in one H-point. 6 2. Developing test procedure Adjustments to the test rig were done for the belt geometry to resemble that of a Volvo S60. The sedan vehicle type represented 41% of the total number vehicles that were included in the naturalistic driving study by Reed et al. (2020). Measurements of the D-ring’s position in relation to the tracks of the car seat were taken by manual means from a Volvo S60. The fore-aft position in the test rig was limited by the track length of 205 mm, compared to the 220 mm in the Volvo S60. The same z-axis distance to the D-ring attachment was measured in both the Volvo S60 and the test rig. The distance between the seat track and the D-ring, the y-axis, was similar to the D-ring’s position in the Volvo S60. The coordinates relative the H-point at the rearmost position for the B-pillar installation were (-174, -280, 632) mm for the lower end of the D-ring and (-174, -280, 702) mm for the upper end of the D-ring. The most forward position of the seat in the test rig was at (-287, -280, 632) mm for the lower installation and (-287, -280, 702) mm for the upper installation. (a) B-pillar installation: D-ring range. (b) Fore-aft range in the test rig. Figure 2.2: The D-ring range for the B-pillar installation and the fore-aft range in the test rig. In addition to the B-pillar installation, as seen in Figure 2.1, a mock-up of a belt-in- seat installation could be installed in the test rig. The seat belt installation resem- bling a belt-in-seat installation was achieved by mounting the D-ring attachment in the beam behind the head rest as shown in Figure 2.3. The coordinates relative to the H-point for the belt-in-seat installations were: (-369, -175, 514) mm and (- 345, -175, 514) mm for either end of the webbing, based on an existing belt-in-seat installation. Thereby, the D-ring and fore-aft position is fixed for the belt-in-seat installation. The buckle and anchor of the lap belt were the same for both the B-pillar and belt-in-seat installation. 7 2. Developing test procedure Figure 2.3: The belt-in-seat installation 2.3 Parameters influencing shoulder belt reposi- tioning To limit the amount tests and pre-pretensioner activations performed on each vol- unteer in the volunteer study, the parameters influencing belt repositioning were prioritized. These parameters created the basis for the test conditions used in the volunteer study. The seat belt system and the belt retraction components could add up to more degrees of freedom (test setups) than could be tested in a volunteer study. Additional parameters were occupant posture and clothing. A summary of all identified parameters and their respective options and settings that could affect shoulder belt repositioning are found in Table 2.1. The possibility to determine if shoulder belt repositioning was repeatable was in- vestigated. However, it would mean that the shoulder belt would be retracted in positions where it does not reposition. A shoulder belt that does not reposition causes more discomfort for the volunteer than when it does reposition. Therefore, repeatability testing for volunteers where the belt does not reposition would cause unnecessary strain. Also, to maintain a low amount of belt retractions with the pre-pretensioner the total number of belt retractions was kept as small as possible. To minimize the discomfort and strain for the volunteers and to maintain a low amount belt retractions, repeatability was excluded from the volunteer study. 8 2. Developing test procedure Table 2.1: The parameters influencing shoulder belt repositioning that were identified along with the available options and settings for each parameter. Parameter Options and settings Belt installation Two D-ring installations were available, one B-pillar and one belt-in-seat installation. D-ring height For the B-pillar installation: There were four available height options of the D-ring. In addition to the height options, the D-ring attachment on the pillar was mov- able in the y-axis which created more possible D-ring attachment options, see Figure 2.2a. For the belt-in-seat installation: One fixed D-ring height, see Figure 2.3. Fore-aft position For the B-pillar installation: The seat was mounted on tracks and could be moved forward and backwards in the test rig, meaning several fore-aft positions were avail- able, see Figure 2.2b. For the belt-in-seat installation: One fore-aft position is set depending on the belt-in-seat configuration. Seat back angle The seat back angle could be manually adjusted, creat- ing many options. Seat height The height of the seat could be manually adjusted. It simultaneously changes the seat back angle when ad- justed. Occupant clothing Many different clothing materials could be used: higher friction, lower friction, and bulkier shirts to name a few. Occupant posture In addition to a nominal posture, several other volunteer postures could be analyzed. The main identified out-of- position postures in this study were: leaning inboards (towards the center console), leaning outboards (towards the door), and slouching. Force level of pre- pretensioner The test rig was equipped with a pre-pretensioner that had the possibility to retract the belt with a force of 250 N and 450 N. Duration time of pre-pretensioner The duration time of the pre-pretensioner could be set to any arbitrary time period larger than 600 ms. 9 2. Developing test procedure 2.4 Parameter observations The parameters in Table 2.1 were investigated and analyzed to determine which parameters to include in the study. The parameters were tested on the authors of this thesis and on a few employees at Autoliv. The observations from this determined what test conditions to include in the volunteer study. A summary of the prioritized and fixed parameters can be found in Section 2.5. 2.4.1 Belt installation The belt installation was found to be an influential parameter when analyzing belt repositioning. Different B-pillar configurations and belt-in-seat installation were tested. The results showed that the belt installation in combination with the test persons’ body characteristics affected belt repositioning. Therefore, both a belt-in- seat and B-pillar installations were included in the volunteer study. 2.4.2 D-ring height The D-ring position is closely related to the belt installation and is subsequently equally important as a parameter. The D-ring height was tested with all other parameters fixed and the result indicated that the height affects belt repositioning. It was observed that shoulder belt repositioning at the low D-ring height was more difficult compared to the high. Based on these observations, the volunteer study includes belt repositioning for both the low and high D-ring heights. The two extremes were chosen to be analyzed since a larger contrast between the settings would yield a more distinctive difference in belt routing. For the test procedure, the low D-ring height was decided to be used as a nominal setting since it was identified as most difficult for belt repositioning. One test was used as a reference where both the high and low D-ring position were tested, to validate that the belt repositioned in the high D-ring position as well. The change in D-ring height is only applicable for the B-pillar installation since a singular belt geometry was tested for the belt-in-seat installation. 2.4.3 Fore-aft position Several fore-aft positions were tested and the findings showed that the further for- ward from the D-ring attachment the seat was positioned, the more difficult the belt repositioning became. Given these observations, the fore-aft position was found in- fluential to shoulder belt repositioning and was included in the test procedure. For the B-pillar installation, the y-axis distance from the H-point to the D-ring attach- ment was -280 mm and z-axis distance was 632 mm to the low D-ring and 702 mm to the high D-ring, see Figure 2.1 for coordinate system. 10 2. Developing test procedure The fore-aft span in the x-axis in the test rig between the D-ring attachment and the H-point range from -174 mm to -387 mm. These values translate to the rearmost position in a Volvo S60 to 93% forward, the shorter track length of the rig limiting the range. The fore-aft positions included in the test were based on the positions that passengers regularly sit. The study by Reed et al. (2020) found that the most common fore-aft position passengers used was between 0% to 50% forward from the aft position. The median seat position in their study was approximately 53 mm forward of the rearmost position, equivalent to 22% of the track length. The corresponding fore-aft position in a Volvo S60 is 48 mm forward of the aft position, adjusted to 49 mm due to the test rig tracks’ design. This gives the x-axis coordinate for the 22% forward position of -219 mm. Determining belt repositioning in the commonly used fore-aft positions is recognized by the authors as more important than less common positions (50% and further ahead). However, less common positions were included to investigate how the fore- aft position affects belt repositioning. Therefore, two additional fore-aft positions were included in the study: 50% and 75%. 50% forward corresponds to -284 mm and 75% to -346 mm in the x-axis distance between the D-ring attachment and the H-point. A visualization of the fore-aft positions is depicted in Figure 2.4, where the fore-aft position for the belt-in-seat installation at 0% forward of the rearmost position is also depicted. Figure 2.4: A visualization of the car seat at the different fore-aft positions for the preliminary test conditions. The percentages are forward of the rearmost fore-aft position in the test rig, calculated relative the track length of a Volvo S60. The distances in millimeter are given relative the D-ring attachment and the H-point. The final name is a description of its position in the fore-aft range. 2.4.4 Seat back angle The seat back angle mainly affects the contact point between the passenger and the belt. Increasing the seat back angle increases the distance between the occupant and the belt, thereby making belt repositioning easier. The test rig’s seat back angle was therefore kept fixed at 25◦, according to the seat torso angle in the Euro European New Car Assessment Programme (NCAP) protocol on mobile progressive deformable barrier collisions from 2023 (2023). The seat back angle was measured together with the H-point with a SAE mannequin. 11 2. Developing test procedure 2.4.5 Seat height Adjustments of the seat height affects the seat back angle, which as previously mentioned is kept fixed in this study. Changing the seat height created unwanted changes in seat back angle. The seat height was therefore kept fixed at the lowest seat height setting, according to the Euro NCAP protocol from 2023 on mobile progressive deformable barrier collisions (2023). 2.4.6 Occupant clothing Friction and bulkiness of occupant clothing was identified to potentially influence belt repositioning. Therefore, several clothing options hypothesized to make belt repositioning more difficult were tested: regular long sleeve shirts, a winter coat, knitted sweaters, and a jeans jacket. The findings indicated that clothing affects belt repositioning. The bulkiness of clothing hindered the belt from properly repo- sitioning at times. Reproducing such a hinder and positioning the belt similarly on the arm was problematic. In terms of reproducibility, the difference in cloth sizing and bulkiness was an issue since it is difficult to recreate it similarly for all vol- unteers. Making belt repositioning more difficult was therefore excluded from this study. All testing in the volunteer study was going to be conducted with a plain long- sleeved cotton T-shirt. A plain shirt made it easier to position the belt in a similar matter on the arm. In addition to the plain shirt, belt repositioning could be made easier by using a material with a lower friction since making repositioning more difficult was excluded. Using a material with a lower friction was more practical since a thin layer of fabric could easily be placed on the volunteer, see Figure 2.5. It did not interfere with belt positioning on the arm. The material with lower friction was included in the test conditions where belt repositioning failed. The influence that friction has on belt repositioning could therefore be investigated. Figure 2.5: A photo of how the shoulder was covered with the lower friction material used in the study. 12 2. Developing test procedure 2.4.7 Occupant posture The postures identified for this study were the nominal posture and out-of-position postures: leaning outboards (towards the door), leaning inboards (towards the center console), and slouching, see Figure 2.6. The nominal posture was identified to have a proper initial belt fit since it allows the belt to reposition to a position across the clavicle, thereby making the posture ideal for this study. The out-of-position pos- tures were not ideal. Leaning inboards did not allow for the belt to be repositioned to an on-clavicle position. Leaning outboards could cause injuries to the volunteers neck when the shoulder belt is retracted and reposition the occupant instead of the belt. Slouching was excluded since it was more relevant when analyzing lap belt fit and since it made the belt repositioning easier. Based on the findings, the nominal posture was the only posture suited for this study. Since the nominal posture was deemed fit for this study the lumbar support was kept in its most retracted position throughout the study. (a) Nominal posture (b) Leaning inboards (c) Leaning outboards (d) Slouching Figure 2.6: Belt fit in four postures with B-pillar installation. 2.4.8 Force level of pre-pretensioner The pre-pretensioner had two activation profiles which retract the belt with 250 N and 450 N respectively. The authors of this thesis decided that a force level of 450 N was too uncomfortable when retracted multiple times during a short period of time whereas the force level of 250 N was more manageable and acceptable to retract several times. In addition to the preferences of the test persons, the functionality of the higher force level was compared to the lower force level. It was investigated if the force level at 450 N could reposition the shoulder belt when the 250 N force level failed. The result showed that this was not the case, meaning the higher force did not reposition the belt better than the lower force. Based on these findings, the 250 N force level was therefore used in the volunteer testing. 13 2. Developing test procedure 2.4.9 Duration time of pre-pretensioner The shortest possible duration time of the retraction phase of the pre-pretensioner is 600 ms, due to the limits of the pre-pretensioner. Therefore, time periods larger than 600 ms were investigated. The time to reposition the belt was analyzed to determine for how long the pre-pretensioner had to be activated. The duration time had to be long enough to ensure that the belt is able to reposition to an on-clavicle position for all volunteers. In other words, the belt should not stop retracting before the pre-pretensioner has had the chance to reposition the shoulder belt. The duration time was tested, with the force profile of 250 N, in different seat configurations when the belt was placed in different positions on the arm. The findings showed that the belt repositioning time was less than 1 s. Based on these findings, the belt retraction time was set to 2 s to ensure that the possibility that belt retraction time does not affect the belt repositioning outcome. 2.5 Summary of the observations on parameters A summary of the final option(s) for the parameters that were identified to affect belt repositioning are found in Table 2.2 along with the total number of parameters represented by the octothorp (#) sign. The parameters in the table created the basis for the tests conducted in the volunteer study. As previously mentioned, the D-ring height and fore-aft positions were only adjusted for the B-pillar installation and not for the belt-in-seat installation. Table 2.2: A summary of the parameters that were prioritized and fixed along with the final option and the total number of parameters (#). Clothing was included for the tests when the belt did not reposition, thereby the (+1) sign. Parameter Final option(s) # Belt installation B-pillar and belt-in-seat installation. 2 D-ring height The high and low D-ring heights according to a Volvo S60 seat configuration for the B-pillar in- stallation whereas the D-ring position was fixed for the belt-in-seat installation. 2 Fore-aft position The car seat was set at three positions: 22%, 50%, and 75% forward of the rearmost position repre- sentative of a Volvo S60 seat configuration for the B-pillar installation. Corresponding to the x-axis position of the D-ring attachment at -219 mm, - 284 mm, and -346 mm relative the H-point respec- tively. Whereas the belt-in-seat installation had a fixed fore-aft position. 2 Seat back angle Fixed at 25◦. 1 Seat height Fixed at lowest setting. 1 14 2. Developing test procedure Occupant clothing Each participant was provided a long-sleeved shirt and additional tests with lower friction was carried out when the belt was not repositioned. 1 (+1) Occupant posture The volunteers were asked to sit in a nominal pos- ture. 1 Force level of pre- pretensioner A force profile of 250 N for the pre-pretensioner. 1 Duration time of pre-pretensioner A duration time of 2 s for the pre-pretensioner. 1 2.6 Preliminary test conditions From the summary above, five preliminary test conditions used to investigate the pre-pretensioners ability to reposition the shoulder belt were identified for the vol- unteer study: a fore-aft position where the D-ring attachment is at -219 mm, -284 mm, and -346 mm relative the H-point in the x-axis respectively (22%, 50%, and 75% forward of the rearmost position), a fourth test with a high D-ring at -284 mm (50% forward of the rearmost position), and a fifth test with a belt-in-seat instal- lation. The test conditions are preliminary since they are to be tested in a pilot study. A summary of the preliminary test conditions is found in Table 2.3, where the belt installation, D-ring height, fore-aft position, and the clothing are specified. For the B-pillar installations the test is complete if the belt is repositioned in the test conditions with the low D-ring. If the low D-ring instead fails to reposition two additional tests are carried out; the first one with lower friction and the second one with a high D-ring. Whereas if the belt repositioning failed for the belt-in-seat installation, the test was repeated with a lower friction material. Table 2.3: The preliminary test conditions for the volunteer study. The fore-aft position is presented in distance of the D-ring attachment relative the H-point in the x-axis. The test conditions in parenthesis were only conducted if belt repositioning failed in the test with the low D-ring. Belt installation D-ring height Fore-aft position [mm] Clothing B-pillar Low -219 Cotton shirt B-pillar Low -284 Cotton shirt B-pillar High -284 Cotton shirt B-pillar Low -346 Cotton shirt (B-pillar) (Low) (-346) (Lower friction) (B-pillar) (High) (-346) (Cotton shirt) Belt-in-seat Fixed Fixed Cotton shirt (Belt-in-seat) (Fixed) (Fixed) (Lower friction) 15 2. Developing test procedure 2.7 Incremental process to find a potential point- of-no-return To investigate to what extent the pre-pretensioner is able to reposition the belt, the shoulder belt had to be taped with masking tape down on the arm of the volunteer. The aim was to determine if the belt was repositioned or not. If the belt was not repositioned, it was of interest to determine if it got more difficult the further down the arm the belt was positioned and, in that case, determine where the point-of-no- return was. The point-of-no-return being the position furthest down the arm where the belt fails to reposition. The range down the arm was chosen to be from 5 cm down to 20 cm, in steps of 5 cm as is depicted in Figure 2.8. The distances were measured from the acromion since it could be identified on all volunteers. The acromion was palpated according to the description of its location in (Swedish Standards Institute, 2017), and the acromion’s position is depicted in red in Figure 2.7. The maximum distance down the arm was set to 20 cm to ensure that all volunteers could participate in the volunteer study. At around 30 cm down the arm it was also hypothesized that occupants are more likely to place the belt below the elbow or underneath the arm instead, which are cases where the pre-pretensioner cannot improve belt fit. The belt position at 0 cm is henceforth referred to as the reference position. To investigate belt repositioning, an incremental process was created, see Figure 2.9. The incremental procedure was carried out for each test condition in the volunteer study. A reference belt retraction is performed when the belt is positioned on the shoulder, referred to as 0 cm. The first belt retraction below the shoulder is at 10 cm and if the belt repositioned successfully the next position was 20 cm down the arm. If the belt failed to reposition at 10 or 20 cm, the belt is moved up the arm to 5 cm or 15 cm respectively to identify a potential point-of-no-return. The shoulder belt was taped with masking tape to the seat when positioned on the arm at all distances from the shoulder. How the belt was fastened using masking tape and how the distance was measured, from the acromion process to the upper edge of the belt, in the B-pillar installation and the belt-in-seat installation is depicted in Figure 2.10 and Figure 2.11, respectively. 16 2. Developing test procedure Figure 2.7: The position of the acromion, marked in red, from a frontal and side view of the shoulder. (a) 0 cm (b) 5 cm (c) 10 cm (d) 15 cm (e) 20 cm Figure 2.8: The incremental steps down the arm, measured from the acromion process, with the B-pillar installation. 17 2. Developing test procedure Figure 2.9: A flowchart of the incremental process used to analyze belt repositioning and identify the point-of-no-return if the belt fails to reposition. The blue/green boxes show the distance where the pre-pretensioner was activated. The orange rhombuses show the next step based on the outcome from the green boxes. The dark blue boxes were the last test at the identified point-of-no-return with a material with lower friction. 18 2. Developing test procedure (a) 5 cm (b) 10 cm (c) 15 cm (d) 20 cm Figure 2.10: An illustration of how the distances down the arm were measured and fastened using masking tape in the B-pillar installation. The distance was measured from the acromion process to the upper edge of the belt. (a) 5 cm (b) 10 cm (c) 15 cm (d) 20 cm Figure 2.11: An illustration of how the distances down the arm were measured and fastened using masking tape in the belt-in-seat installation. The distance was measured from the acromion process to the upper edge of the belt. 2.8 Definition of a repositioned shoulder belt To determine if the belt was repositioned to an on-clavicle position, a proper and improper shoulder belt fit had to be defined. A proper and improper belt fit has been defined in two studies on belt fit, by Fong et al. (2016) and Bohman et al. (2019). Both studies had to identify and classify belt fit, the former for elderly volunteers and the latter for younger as well as elderly volunteers. Both studies had a similar approach when defining proper and improper belt fit. Fong et al. classified a proper and improper belt fit in regard to the belts’ intended loading regions. Both studies classified a proper belt fit as having the shoulder belt in the center area of the shoulder and the lap belt having contact with the thigh. An improper belt fit was in both cases classified as when the shoulder belt was touching the neck, placed on the outer edge of the shoulder, or off the shoulder. In addition to the belt’s position on the shoulder, the shoulder belt’s position on the abdomen is also mentioned by Bohman et al. (2019). 19 2. Developing test procedure Based on the two studies on belt fit, the belt was defined as repositioned if the shoulder belt was medial of the acromion process and improper if the shoulder belt was lateral of the acromion process. The shoulder belt position across the torso, especially the belt routing on the abdomen, was considered repositioned if it crossed the sternum similarly as in the reference position. If the shoulder belt crossed the acromion process in the reference position, the seat belt was considered repositioned if it reached the reference position. 2.9 Data collection Measurements that were of interest for this study were anthropometric, belt geom- etry, and belt fit measurements. Additional measurements: retracted belt webbing, belt retraction time, and belt repositioning time, were recorded for future research purposes. The criteria to record and collect certain sets of data and the record- ing method for this study depended on the efficiency of respective data collecting method and to what extent that the belt interfered with the retraction results. The criteria were not met when shoulder belt force measurements were explored. Shoul- der belt force measurements were excluded due to the weight of the sensor since the weight could affect the belt angle from the D-ring to the shoulder and thereby influence belt retraction. More detailed information about the included data sets is presented below. 2.9.1 Anthropometric measurements Anthropometric measurements, in addition to age and sex, of the volunteer were taken to analyze if specific upper body characteristics were related to shoulder belt repositioning. The majority of the measurements and the corresponding procedure to take the measurements were according to SS-EN ISO 7250-1:2017, Swedish stan- dard institute (2017). Shoulder circumference and shoulder length were not standard measurements, they were defined and taken according to the procedure presented by Hotzman el al. (2011). Elbow-to-elbow breadth is the distance between the elbows when the arms are held out in 90◦. A summary of the anthropometric measurements and the equipment to take the measurement is found in Table 2.4, where the two calipers used are depicted in Figure 2.12. 20 2. Developing test procedure Table 2.4: The anthropometric measurements measured of all volunteers that could be connected with belt repositioning Posture Measurements Equipment Standing up Stature Measuring tape on wall Weight Scale Wall-acromion distance Caliper Chest depth Wooden caliper Chest breadth Caliper Chest circumference Measuring tape Waist circumference Measuring tape Shoulder circumference Measuring tape Shoulder length Measuring tape Sitting down Sitting height Measuring tape on wall Shoulder height (sitting) Wooden caliper Shoulder-elbow length Caliper Shoulder (biacromial) breadth Caliper Shoulder (bideltoid) breadth Caliper Elbow-to-elbow breadth Caliper Hip breadth Caliper Abdominal depth Caliper Thorax depth Caliper Figure 2.12: The caliper and the wooden caliper with spikes, both used to measure anthropometric measurements. 21 2. Developing test procedure 2.9.2 Belt geometry and belt fit measurements Belt geometry and belt fit data was hypothesized to influence shoulder belt reposi- tioning, especially since belt repositioning was observed to be more difficult with a belt-in-seat installation compared to a B-pillar installation. Based on this finding, more data regarding belt wrapping was determined to be included in the study. Wrapping is defined as how the belt encloses the shoulder. In total, six measure- ments were taken at the reference position for each of the test conditions in Table 2.3. Three measurements are related to belt geometry and three related to belt fit, one of which is wrapping. In total, six measurements were taken in the reference position for the B-pillar installation and five in the belt-in-seat installation. 2.9.2.1 Belt geometry angles The first belt geometry angle was taken where the belt extends from the D-ring. The second angle was taken where the belt first was in contact with the volunteer. Both these angles were measured from the XY-plane with a digital protractor. The angles are henceforth referred to as the D-ring angle and the contact angle, respectively. The D-ring and contact angle were measured for the B-pillar installation. For the belt-in-seat installation the D-ring and contact angle are a combined angle since the D-ring and contact point are at the same place. The D-ring and contact angles for respective belt installation are depicted in Figure 2.13. Figure 2.14 depicts the third belt geometry angle, in the XY-plane, and how it was measured in the two belt installations. It was measured with a L-square ruler and protractor. The L-square ruler was held horizontally against a flat surface on the test rig and the ruler at the end of the protractor was held against the outer edge of the seat belt. The angle was measured on the left side on the protractor. 22 2. Developing test procedure (a) B-pillar installation: D-ring angle (b) B-pillar installation: contact angle (c) Belt-in-seat installation: D-ring angle (d) How the angle in Figure 2.13a was measured (e) How the angle in Figure 2.13b was measured (f) How the angle in Figure 2.13c was measured Figure 2.13: Illustration of the belt geometry angles and how they were measured using the digital protractor. The angle is presented in the first three figures and the corresponding measurement method in the three below. (a) B-pillar installation: belt angle in XY-plane. (b) How the angle in Figure 2.14a was measured. (c) Belt-in-seat installation: belt angle in XY-plane. (d) How the angle in Figure 2.14c was measured Figure 2.14: An illustration of the measured belt angle in the XY-plane and how it was measured in both belt belt installations. 23 2. Developing test procedure 2.9.2.2 Belt fit distances Reed et al. (2013) and Bohman et al. (2019) both conducted studies using volunteers where they took measurements to investigate belt fit. The belt fit measurements were taken between the suprasternal notch and the shoulder belt, where Reed et al. (2013) measured it horizontally (y-axis) and Bohman et al. (2019) vertically (z-axis). The vertical belt and horizontal belt fit measurements and how they were taken with a measuring taped to the edge of the belt are depicted in Figure 2.15. (a) The vertical and horizontal belt distance. (b) How the vertical belt distance was taken. (c) How the horizontal belt distance was taken. Figure 2.15: Illustration of the two belt fit measurements, vertical belt distance and horizontal belt distance, and how they were measured between the suprasternal notch to the edge of the belt. 2.9.2.3 Belt wrapping distances The belt wrapping distance in the x-axis between the acromion and the outer edge of the belt was measured for the B-pillar and belt-in-seat installation. The belt wrapping distance is depicted in Figure 2.16 for the two installations. The wrapping was 0 cm in Figure 2.16c since there is no distance between the acromion process and the belt, which indicates a high degree of wrapping. 24 2. Developing test procedure (a) The belt wrapping distance with a B-pillar installation. (b) How the belt wrapping distance was measured. (c) The belt wrapping distance with a belt-in-seat installation. Figure 2.16: The belt wrapping distance in the x-axis, from the acromion process to the outboard edge of the belt and how it was measured for the B-pillar installation. The wrapping distance is not measured if the belt covers the acromion process for the belt-in-seat installation. 2.9.3 Retracted belt webbing and retraction time The retracted webbing and the belt retraction time provides an indication of the amount of belt slack and how much belt has to be retracted to reach a proper belt fit for occupants with different body characteristics. Several attempts at measuring the retracted webbing were tested, by optic devices, video, and software. The optical approach required a tape to be placed on the belt and this tape would interfere with the D-ring when the belt is retracted. The video approach required photo markers to be positioned on the belt which would have to be done for each volunteer, and thereafter use a software program to analyze the data. This option was excluded due to the large number of volunteers participating which led to large quantities of data. The most efficient method was to measure the retracted webbing using the software CANalyzer, which is used to activate the pre-pretensioner. The belt retraction velocity is recorded, logged, and integrated to obtain the retracted webbing length. The software CANalyzer was found to be the most efficient solution since retracted webbing could easily be recorded during the volunteer study. 25 2. Developing test procedure 2.10 Pilot study When the different parts of the test procedure described above in this chapter had been decided upon, they were all tested in a pilot study where four individuals participated. Conducting pilot studies is vital since it can help identify and pinpoint the most suitable test procedure, according to Teijlingen and Hundley (2002). They mention that adjustments can be made to the equipment needed for the study and that a research protocol can be established which in turn reveals potential issues with the test procedure. A pilot study was therefore conducted to identify and solve potential problems with the test procedure and protocol at an early stage. The pilot was not only necessary to implement changes to the test procedure, but also to learn and get a routine as to how to conduct the entire test procedure similarly for all volunteers. 2.10.1 Evaluation of the pilot study The major observation in the pilot study was regarding the belt installations. The result from the pilot study indicated that a belt-in-seat installation could be difficult to reposition since the belt did not reposition for two of the four participants. The findings regarding the B-pillar installation showed that the belt repositioned for all participants at the four test conditions when the D-ring attachment was at -219 mm, -284 mm, and -346 mm relative the H-point. Where -284 mm also was conducted with a high D-ring. Although the result for the B-pillar installation was promising in terms of vehicle safety, a possible connection between anthropometric measurements, belt geometry angles, and belt fit measurements could not be identified if the belt always repositioned. Therefore, an additional test condition was created with the expectation that the belt would not reposition for some participants. The added test case was the most forward position available in the test rig, with a fore-aft position 93% forward of the rearmost position in a Volvo S60. The fore-aft position can also be described as when the D-ring attachment was at -387 mm relative the H-point in the x-axis. It was anticipated that more data regarding failed belt repositioning could be gathered by adding this test case, which was hypothesized to bring light to the relevant data sets affecting belt repositioning. Since one test condition was added, it would affect the total amount of belt retractions. However, the findings regarding the test procedure indicated that the number of extra tests are relatively low and the full test procedure was judged to fit within a reasonable time frame. 26 3 Methods for volunteer study The overall structure and implementation of the developed test procedure for the volunteer study for this thesis is presented below. First the volunteer sample is presented and then the final test conditions that were adjusted after the pilot study. Thereafter the test rig instrumentation and the test procedure used during the volunteer study is described. The last section presents the methods used for the data analysis. 3.1 The volunteers The aim was to recruit around 40 volunteers. The volunteers were recruited through advertisement at Autoliv and several were also approached and asked if they would like to participate. Most volunteers therefore had a background within vehicle safety. In addition to the volunteers from Autoliv, some female students were recruited which was needed to have equally many males and females participants in the study. The students were provided with a gift card of 200 SEK each for their contribution. All volunteers were informed that the study was voluntary and that they did not have to participate. They were informed that they could decline the meeting in- vitation if they did not want to participate after reading the document explaining the test procedure. Ensuring that the approach and recruited individuals wanted to participate and did not feel pressured was essential. All volunteers needed to fulfill certain medical requirements, where individuals with previous or current pain in upper body regions could not participate. Recruitment was mainly performed at Autoliv, which resulted in a convenience sample since the volunteers are chosen since they are easily accessible according to MacFarlane et al. (2014). Few volunteers were approached based on body charac- teristics, known as purposeful samples (MacFarlane et al., 2014). Specific volunteers are then chosen based on wanting to obtain a certain set of data. Using these sam- ples was necessary to save time and to investigate if failed belt repositioning is more common among certain body characteristics. 27 3. Methods for volunteer study 3.1.1 Volunteer distribution If the anthropometric measurements of the sample of volunteers is normally dis- tributed according to the Swedish population, the result can be generalized for the Swedish population. A normal distribution is likely to be achieved if more than 30 volunteers participate, where fifteen or more are males and fifteen or more are females. Within each sex, the volunteer measurements should be equally divided in the three percentile groups: 0th - 33th, 33th - 66th, and 66th - 100th. One an- thropometric measurement from each dimension: height, depth, and breadth, were used to determine if the two groups were normally distributed. Sitting height, chest depth, and shoulder bideltoid breadth, were hypothesized to affect belt repositioning the most. Table 3.1 shows the threshold measurements for each percentile and the corresponding value from the Swedish population for both males and females. The measurement from the percentiles from the Swedish population were collected from Antropometri.se, which uses data collected from the Swedish population by Hanson et al. (2009). Table 3.1: Three percentile ranges with the corresponding measurement for three anthropometric measurements defining the Swedish population. All measurements are given in millimeter. Measurement Percentile groups <33th 33th - 66th >66th Males Sitting height 928.15 928.15 - 959.17 959.17 Chest depth 234.95 234.95 - 260.99 260.99 Shoulder bideltoid breadth 464.63 464.63 - 487.50 487.50 Females Sitting height 876.76 876.76 - 906.46 906.46 Chest depth 222.08 222.08 - 253.23 253.23 Shoulder bideltoid breadth 414.55 414.55 - 434.14 434.14 3.2 Final test conditions for volunteer study The final test conditions that were implemented in the volunteer study are presented in Table 3.2. The difference compared to the preliminary test conditions presented in Section 2.6 in Table 2.3 is the additional test at the most forward position in the test rig, at -387 mm in the x-axis, which was added after the pilot study. The distances of the D-ring attachment at -219 mm, -284 mm, -346 mm, and -387 mm relative the D-ring is also referred to as 22%, 50%, 75%, and 93% respectively. Also, as the most common seating position, the mid position, the second most forward position, and the most forward position in the test rig. The belt-in-seat installation is in the most rearward position, at 0%. An overview of the fore-aft positions and the corresponding denotations are found in Figure 3.1. 28 3. Methods for volunteer study Table 3.2: The final tests for the volunteer study. The fore-aft position is presented in the x-axis distance of the D-ring attachment relative the H-point. The test conditions in parenthesis were only conducted if belt repositioning failed in the test with the low D-ring. Belt installation D-ring height Fore-aft position [mm] Clothing B-pillar Low -219 Cotton shirt B-pillar Low -284 Cotton shirt B-pillar High -284 Cotton shirt B-pillar Low -346 Cotton shirt (B-pillar) (Low) (-346) (Lower friction) (B-pillar) (High) (-346) (Cotton shirt) B-pillar Low -387 Cotton shirt (B-pillar) (Low) (-387) (Lower friction) (B-pillar) (High) (-387) Cotton shirt Belt-in-seat Fixed Fixed Cotton shirt (Belt-in-seat) (Fixed) (Fixed) (Lower friction) Figure 3.1: A visualization of the car seat at the different fore-aft positions for the test conditions. The percentages are forward of the rearmost fore-aft position in the test rig, calculated relative the track length of a Volvo S60. The distances in millimeter are given relative the D-ring attachment and the H-point. The final name is a description of its position in the fore-aft range. 3.3 Test rig instrumentation Two GoPro Hero 11 cameras were used to record the belt retraction for all volunteers. One GoPro was positioned on a tripod and recorded the volunteer and the belt retraction from the front, see Figure 3.2a. The frame rate was set to 240 and standard linear field of view setting with no additional zoom was used. The video recorded by the front camera was used to observe the belt fit of the volunteer after the testing had been conducted. The second GoPro was mounted on the test rig and recorded the belt retraction at the spool, where the photo markers were positioned to record the motion of the belt, see Figure 3.2b. The video recorded by the second camera recorded the belt from behind the seat and could be analyzed in the future 29 3. Methods for volunteer study to investigate the time for the belt to move from an on-arm position to an on-clavicle position. In addition to the GoPro instrumentation, a box was placed in front of the test rig to raise the feet to resemble the foot position in a car. (a) Front view camera (b) Back view camera Figure 3.2: Placement of the two Gopro cameras. 3.4 The test procedure The test procedure includes all steps that were conducted when the volunteer was present at the testing area. The test procedure was carried out by two test leaders. One test leader instructed, took the anthropometric measurements, and performed the belt repositioning test conditions. The second test leader documented the an- thropometric data, filled in the test protocol, and was main responsible for activat- ing the pre-pretensioner. The entire test procedure lasted around one hour, starting from the moment the volunteer arrived until they left. Preparing, instrumentation, and taking anthropometric measurements of the volunteer took around 25 minutes. The volunteer was seated in the test rig for around 25 to 30 minutes. 3.4.1 Volunteer preparation, instrumentation, and data col- lection First, the volunteer was asked to sign a form giving consent to participation, in- cluding data collection, video recording, and storing. All information was provided before the testing and was kept available at the testing area. More in-depth informa- tion regarding the pre-pretensioner and the functionality was given to the volunteer if desired. 30 3. Methods for volunteer study The volunteer was prepared for the study by being asked to change into the assigned shirt. The volunteer was asked to remove their shoes in preparation for the anthro- pometric measurements. An adjustable necklace was placed on the volunteer which marked the neck region. The acromion processes and the suprasternal notch were palpated and marked with photo markers on the shirt respectively the skin. After the volunteer had been prepared, photos from the front and one from the side were taken while the volunteer was seated and standing, see Figure 3.3, with a Canon PowerShot G7 X Mark III camera. The area and setup where the photos were taken while seated and standing are depicted in Figure 3.3b and Figure 3.3c. (a) Seated front (b) Standing front (c) Seated side (d) Standing side Figure 3.3: The photos taken of each volunteer. The anthropometric measurements were taken and noted in Excel. The area where anthropometric measurements were taken is seen in Figure 3.4. A measuring tape was fixed along the edge of the room divider, see Figure 3.4a, which the volunteer had their back against to measure stature. The white foam could be moved along the edge to be positioned at the top of the head. On top of the white foam was a small spirit level to ensure a horizontal angle. To measure sitting height, the stool with a height of 470 mm and foot support (piece of foam) seen in Figure 3.4c, were positioned in front of the fixed measuring tape. When all measurements had been taken the volunteer was seated in the test rig. The seat belt was buckled by the volunteer and adjusted if needed, for example if the lap belt was not placed over the pelvic bones. The volunteer was given several instructions on how to sit. They were asked to sit as far back as possible with their back close to the backrest. They were told to keep their hands on their lap. They were informed to sit as similarly as possible throughout the tests to remain consistent throughout the different tests. The volunteer was instructed to fully extend the seat belt after each time the pre-pretensioner was activated. If the belt was not extended during the volunteer study it could also affect the recorded data by the software CANanalyzer, hence the belt speed and retracted webbing. The test leaders thereafter explained the test procedure and the belt repositioning test conditions, see Table 2.3, that were to be performed. 31 3. Methods for volunteer study (a) Setup to measure stature and sitting height (b) Setup to take photos and measurements while standing (c) Setup to take photos measurements while seated Figure 3.4: Testing area setup to take anthropometric measurements and photos. 3.4.2 Conducting test procedure The tests were performed in a randomized order. If the tests began with a B-pillar installation, all tests with that installation was performed before switching to the belt-in-seat installation and vice versa. This was done to remain efficient while conducting the volunteer study. The tests were performed according to the test procedure presented in the flowchart in Figure 2.9. The belt installation, seat posi- tion, D-ring height, clothing, and a comment section along with the six additional measurements: three belt geometry and three belt fit, were also noted for each test condition. Before the belt was retracted in the reference belt position, the three belt geometry and three belt routing measurements were taken, in all test conditions. When the belt was positioned down the arm, at either 5 cm, 10 cm, 15 cm, or 20 cm, it was observed if the belt would sit in the position without being taped or if the belt was retracted to the reference position. It was observed if it would sit by itself since it was hypothesized that a belt that stay by itself is less likely to reposition. The belt was taped for all volunteers regardless of its ability to sit by itself. The tape was torn before each test so that it would break when the belt was retracted. A visual observation was made by the test leaders to determine if the belt was repositioned or not, according to the set definition of repositioned shoulder belt. Belt repositioning was determined with a binary response, repositioned or not repostioned. If there were any uncertainties regarding belt repositioning, a note was made in the test protocol to check belt repositioning in the video recording afterwards. However, a decision on belt repositioning was made on-scene to continue testing according to the increment procedure, see Figure 2.9. The test could also be repeated if it was difficult to determine if the belt repositioned or not. 32 3. Methods for volunteer study 3.5 Data analysis To investigate what caused the belt to reposition for some volunteers but not for others, the collected data was analyzed using statistical methods. The collected data include the anthropometric, belt geometry, and belt fit measurements. First, a two sample t-test was conducted to compare the anthropometric, belt geometry, and belt fit measurements in the group of volunteers where belt repositioning was possible to the group where repositioning was not possible. All collected measurements were tested at each seat condition where belt repositioning failed. The aim of the t- test was to determine if there was a statistical difference in mean values between the two groups, repositioned or not repositioned. If a statistical difference between the groups would be identified, then there would be an indication that a certain measurement is linked with the belt repositioning outcome. 3.5.1 Statistical significance by two sample t-test with Bon- ferroni correction The two sample t-test was conducted using MATLAB (Inc., 2019a) with the function ttest2 from the Statistics and Machine Learning Toolbox (2019b). The method to conduct a two sample t-test described is described below and it was derived from Peck et al. (2008) unless otherwise specified. The implemented method assumes unequal variance since if a two sample t-test is conducted when the variance is not equal it could affect the p-values (Peck et al., 2008). Before a two sample t-test could be conducted the data within each group, repo- sitioned and not repositioned, need to fulfill certain requirements. Within each of the two groups the data should be independent, randomly sampled, and normally distributed. The collected measurements are independently collected but the ran- domness is limited since the volunteers were from a convenience sample. Normal dis- tribution is usually achieved by having more than 30 samples and the total amount of volunteers in the study was aimed to be around 40. A two sample t-test was deemed fit for this analysis since it would provide indications and show tendencies if the collected measurements affect the pre-pretensioners ability to reposition the belt. It was assumed that the data was normally distributed and quantile-quantile plots were used to check for normality. Table 3.3 shows the denotations that are used when conducting a two sample t-test. It presents the population mean, sample mean, variance, standard deviation (SD), and sample size. The table refers to the denotations for each group, repositioned (yes) and not repositioned (no). 33 3. Methods for volunteer study Table 3.3: The denotations for sample mean, variance, standard deviation, and the sample size. Group Population Sample Mean Mean Variance SD Sample size Repositioned µyes xyes s2 yes syes nyes Not repositioned µno xno s2 no sno nno The population means, µyes and µno, is the average value of each population and it is used to set the null hypothesis. The sample mean is the one calculated from the data collected from the volunteers and it was given by x = Σn i=1x n (3.1) where the collected measurement x is xyes or xno, and the corresponding sample size was nyes or nno. The sample variance, s2, for each group was found by s2 = Σ(x − x)2 n − 1 (3.2) where the collected measurement x is xyes or xno, and the sample mean is x is xyes or xno. To investigate statistical significance, the null hypothesis, H0, was set to H0 : µyes = µno (3.3) and alternative hypothesis, H1, was set to H1 : µyes ̸= µno (3.4) for all collected measurements. The null hypothesis states there is no difference be- tween the means of the two groups. If the null hypothesis is rejected, the alternative hypothesis is accepted, meaning there is not a statistical difference between the two mean values. If rejected, the result would therefore indicate a possible connection between the specific collected measurement(s) and belt repositioning. A significance level, α, was used to determine if the null hypothesis should be rejected or not. The significance level was set to 0.05. This significance level mean there is a 5% chance that a correct null hypothesis will be rejected. Collecting and performing 34 3. Methods for volunteer study two sample t tests on large quantities of data sets is called ”fishing”, which according to Rice (2007) means that it increases the chance that you incorrectly reject a null hypothesis. Rice mentions that a fishing study can be viewed as a step that is performed to provide recommendations for future research. A Bonferroni correction was also implemented to the significance level to compensate for the large amount of collected measurements that were analyzed. If not implemented it is more likely that a null hypothesis is incorrectly rejected when more measurements are analyzed (Rice, 2007). The correction was implemented by adjusting the size of α in regard to the total amount of collected measurements, denoted by k. The Bonferroni corrected significance value, αB, is therefore given by αB = α k (3.5) where k is the amount of collected measurements (Rice, 2007). The corresponding amount of collected measurements, k, in the study include the anthropometric, belt geometry, and belt fit measurements. Once the sample mean, sample variance, and sample, the null hypothesis, and sig- nificance level are known, the test statistic t and the degrees of freedom (df) could be calculated. The t-value and df are needed to find the corresponding p-value. The test statistic t was calculated by t = x1 − x2√ s2 1 n1 + s2 2 n2 (3.6) with the corresponding df given by df = (Vyes + Vno)2 V 2 yes nyes−1 + V 2 no nno−1 (3.7) where Vyes was replaced by s2 yes nyes and Vno by s2 no nno . The p-value is obtained through the software program, but it can also be obtained through a table by using the calculated t-value and df. If the p-value from the software program/table was less than the significance level, p-value < H0, the null hypothesis was rejected and the alternative hypothesis was accepted. 3.5.2 Binary logistic regression The software program IBM SPSS Statistics (IBM Corp., 2021) was used to perform binary logistic regression using the Binary Logistic Regression function. The fol- lowing description was implemented from Fritz and Berger (2015). Binary logistic regression is a method that creates and fit a model according to a collected mea- surement. Binary logistic regression was a suitable method for the data collected 35 3. Methods for volunteer study in this study since the belt repositioning outcome was binary, the belt had either repositioned or it had not repositioned, and since anthropometric, belt geometry or belt fit data had been collected from all volunteers. Thereby, it could be used to pre- dict the probability if the belt will reposition for an individual if their measurement is known. Binary logistic regression was implemented for the measurements that showed potential from the t-test results in predicting if the belt would reposition or not reposition. The logistic regression for one independent variable (one of the collected measure- ments) was given by Y = β0 + β1 · X1 (3.8) where Y is given by either 1 or 0, which represents the binary repositioning outcome, repositioned or not repositioned. The β0 is the intercept with the y-axis and β1 is the slope. The collected measurement is given by X1. If the number of independent variables is two (two of the collected measurements) the logistic regression was given by Y = β0 + β1 · X1 + β2 · X2 (3.9) where X2 is the data from the second measurement and β1 is the corresponding slope. The β0, β1, and β2 in the expressions (3.8) and (3.9) are found by using the software program IBM SPSS Statistics. The binary repositioning outcome, repositioned or not repositioned, and the collected measurement(s) was used as input. Once the variables β0, β1, and β2 (if two independent variables were used) were set, the final probability curve could be set up. The variables were used as input in the expression given by Y = e(β0+β1·X1) 1 + e(β0+β1·X1) (3.10) and by Y = e(β0+β1·X1+β2·X2) 1 + e(β0+β1·X1+β2·X2) (3.11) if two measurements were implemented. These were the final probability curves used to investigate how well the model fits to the data. How well the model fit the data was tested in SPSS by using the Hosmer and Lemeshow goodness-of-fit test. 36 4 Results The result is presented in five sections. First the volunteers that participated in the study and their anthropometric measurements are presented. Then overviews of the results from the B-pillar installation and the belt-in-seat installation are given. A more in-depth analysis follows, focusing on what causes the belt to reposition or not reposition in the different seat configurations. The analysis investigates if anthropometric measurements, belt geometry, and belt fit can be linked to belt repositioning. In the result section, not all data that were recorded during the volunteer study is presented. Additional data, such as the retracted belt webbing and belt retraction time, were recorded and could be used for future research purposes in the subject of belt repositioning. All test conditions, see Table 3.2, mentioned below are referred to by their fore-aft position where they are given as the distance between the D-ring attachment and H-point in x-direction, the H-point being the one presented in Section 2.2. The fore- aft positions for the B-pillar installation are therefore at -219 mm, -284 mm, -346 mm, and -387 mm, and the belt-in-seat installation at -369 mm. The test conditions are also referred to by their fore-aft position in the test rig as the most common seat position, the middle, the second foremost, and the foremost position respectively. Whereas the belt-in-seat installation is referred to by its name. Throughout the result section, males and females are represented in blue and orange respectively. A more opaque nuance of blue and orange in the figures indicate two or more data points overlapping from several volunteers having similar data values. 4.1 Volunteers In total, 37 volunteers participated in the volunteer study, 19 males and 18 females. Two males were excluded from the study. Their data is not presented as part of the result. In the end, the data collected from 35 volunteers, 17 males and 18 females, were included in the data analysis. 37 4. Results Table 4.1 shows the distribution of volunteers, males and females separately, di- vided into three percentile groups based on the anthropometric measurements of the Swedish population. Three chosen measurements are compared, corresponding to height, width, and depth of the upper body: sitting height, shoulder (bideltoid) breadth, and chest depth. The volunteers are divided into the percentile groups based on their individual measurements. An overview of the volunteer group’s mea- surements is presented in Table 3.1. Since the distribution of the volunteers is not equal between the groups, the volunteers are not normally distributed for any of the three dimensions. Therefore, the result from this study cannot be generalized to the Swedish population. The result is only representative of the sample collected in this volunteer study. The anthropometric measurements of the volunteers is found in Appendix A. A summary of the mean, standard deviation, minimum, and max- imum value of the volunteers’ anthropometric measurements is found in Appendix B. Table 4.1: The number of volunteers that fit into the three percentile ranges for three anthropometric measurements presented in Section 3.1.1 of the Swedish population. Measurement Percentile groups <33th 33th - 66th >66th Males Sitting height 7 3 7 Chest depth 5 8 4 Shoulder bideltoid breadth 1 7 9 Females Sitting height 5 9 4 Chest depth 17 1 0 Shoulder bideltoid breadth 2 5 11 4.2 Shoulder belt repositioning in the B-pillar in- stallation Shoulder belt repositioning at the different fore-aft positions is depicted in Figure 4.1. The shoulder belt repositioned for all volunteers in the most common seating position and in the mid position, when the x-axis position of the D-ring attachment was -219 mm and -284 mm relative the H-point. At the seat position -346 mm and forward, belt repositioning started failing. The belt did not reposition for 5.7% of the volunteers, corresponding to two volunteers. That corresponds to 5.9% of the males and 5.6% of the females. Whereas at -387 mm, the belt failed to reposition for 20% of the volunteers, a total of 7 volunteers. It failed to reposition for 23.5% of the males and 16.7% of females, corresponding to four males and three females. Images of the volunteers that the belt did not reposition for at -346 mm is found in Appendix C and the corresponding images at -387 mm is found in Appendix D. 38 4. Results -219 -284 -346 -387 D-ring relative H-point in x-axis [mm] 0 20 40 60 80 100 P e rc e n ta g e o f b e lt r e p o s it io n s [ % ] All volunteers Males Females Figure 4.1: Illustration showing to what ext