Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Furthering the Pro-Neck-Tor helmet technology through multibody modeling and design Thomson, Vanessa 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_september_thomson_vanessa.pdf [ 3.28MB ]
Metadata
JSON: 24-1.0165740.json
JSON-LD: 24-1.0165740-ld.json
RDF/XML (Pretty): 24-1.0165740-rdf.xml
RDF/JSON: 24-1.0165740-rdf.json
Turtle: 24-1.0165740-turtle.txt
N-Triples: 24-1.0165740-rdf-ntriples.txt
Original Record: 24-1.0165740-source.json
Full Text
24-1.0165740-fulltext.txt
Citation
24-1.0165740.ris

Full Text

FURTHERING THE PRO-NECK-TOR HELMET TECHNOLOGY THROUGH MULTIBODY MODELING AND DESIGN by  Vanessa Thomson  BSc. (Eng.), Queen’s University, 2013  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Biomedical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2015  © Vanessa Thomson, 2015   ii Abstract The Pro-Neck-Tor (PNT) helmet was developed to reduce the incidence of cervical spine injuries in head-first impacts. By guiding the head out of the path of the following torso, the helmet reduces the compressive loads on the spine. It consists of an inner- and outer-shell that are connected by an internal guide mechanism. In an impact the mechanism deploys and guides the inner shell and head relative to the outer shell. The helmet is intended to induce extension in slightly anterior-to-vertex impacts, and flexion in slightly posterior-to-vertex and vertex impacts. A passive deployment mechanism was previously designed, but was not thoroughly tested. The objective of this work was to assess the functionality of the passive PNT deployment mechanism, and suggest design improvements accordingly.  The selector mechanism was prototyped and tested in a drop tower. The impact platform was positioned at three different angles (-15⁰, 0⁰ and 15⁰) to generate posterior, vertex, and anterior impacts. The mechanism deployed in the correct direction in the -15⁰ and 15⁰ impact conditions, but deployed incorrectly in the 0⁰ impact condition. Additionally, the mechanism did not induce sufficient head rotation prior to head rebound to reduce the loads on the cervical spine.   A multibody model of the experimental apparatus was constructed to determine how the impact forces were transmitted to the mechanism to initiate flexion or extension deployment. A dynamic analysis revealed that the net moment on the head determined the deployment mode. This moment was affected by loading from the torso. Ultimately, the deployment mode should depend only on the location of an impact on the head.    iii A new selector mechanism was designed, which is capable of sensing anterior impacts according to the direction of the loading vector on the head. By default, the mechanism deploys in flexion. In anterior impacts a switch is triggered, which changes the deployment mode to extension. A computational model of this mechanism showed the mechanism deployed appropriately in anterior, posterior and vertex impacts. The mechanism also produced greater head rotation prior to head rebound than the passive mechanism. Physical prototyping of this mechanism is recommended in the future.     iv Preface This thesis was written entirely by Vanessa Thomson. Dr. Peter Cripton guided the development of the test methodologies and revised this thesis. All drop tower testing, data analysis and computational modeling was performed entirely by Vanessa Thomson.   Chapter 4 is based on a theoretical design developed by Vanessa Thomson. The design was revised by Kurt McInnes to improve the manufacturability of the design.     v Table of Contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of Contents ...........................................................................................................................v List of Tables ................................................................................................................................ ix List of Figures .................................................................................................................................x Acknowledgements .................................................................................................................... xiii Dedication ................................................................................................................................... xiv Chapter 1: Introduction ................................................................................................................1 1.1 Motivation ....................................................................................................................... 1 1.2 Cervical Spine Injury ...................................................................................................... 2 1.2.1 Axial Loading and Cervical Spine Injury ................................................................... 2 1.2.1.1 Epidemiological Evidence: Cervical Spine Injury in Sports .............................. 4 1.2.1.2 Theory: Buckling and the ‘Most Vulnerable Posture’ ........................................ 6 1.2.1.3 Experimental Evidence: Axial Loading in the Laboratory ............................... 10 1.2.2 Factors Affecting Injury ............................................................................................ 11 1.2.2.1 Posture............................................................................................................... 11 1.2.2.2 Torso Interaction ............................................................................................... 12 1.2.2.3 Head Constraint ................................................................................................ 13 1.2.2.4 Neck Muscles .................................................................................................... 17 1.2.3 Injury Tolerance of the Cervical Spine ..................................................................... 18 1.3 The Pro-Neck-Tor: Concept and Development ............................................................ 19   vi 1.4 Existing Devices ........................................................................................................... 23 1.4.1 MIPS Helmet ............................................................................................................ 24 1.4.2 LEATT Brace............................................................................................................ 25 1.5 Objectives & Scope....................................................................................................... 26 Chapter 2: Evaluating the Existing PNT Helmet Mechanism .................................................28 2.1 Introduction ................................................................................................................... 28 2.2 Methods......................................................................................................................... 29 2.2.1 Selector Mechanism Design ..................................................................................... 29 2.2.2 Drop Tower Apparatus ............................................................................................. 31 2.2.3 Data Collection ......................................................................................................... 33 2.3 Results ........................................................................................................................... 34 2.4 Discussion ..................................................................................................................... 40 Chapter 3: Multi-body Modeling of the PNT Helmet ..............................................................45 3.1 Introduction ................................................................................................................... 45 3.1.1 Modeling Contact in MSC Adams ............................................................................ 46 3.2 Methods......................................................................................................................... 48 3.3 Results ........................................................................................................................... 52 3.3.1 Tuning ....................................................................................................................... 52 3.3.2 Validation .................................................................................................................. 56 3.3.2.1 Kinematics ........................................................................................................ 57 3.3.2.2 High-Speed Video ............................................................................................. 59 3.3.3 Dynamic Analysis of Deployment Mechanism ........................................................ 62 3.4 Discussion ..................................................................................................................... 65   vii Chapter 4: Conceptual Design of an Improved Selector Mechanism .....................................69 4.1 Introduction ................................................................................................................... 69 4.1.1 Design Problems ....................................................................................................... 69 4.1.2 Design Objective ....................................................................................................... 71 4.2 Design Methodology: Parameter Analysis ................................................................... 72 4.3 Design Process .............................................................................................................. 74 4.3.1 Need Identification.................................................................................................... 74 4.3.2 Design Requirements ................................................................................................ 75 4.3.3 Technology Identification ......................................................................................... 76 4.3.4 Parameter Identification I ......................................................................................... 77 4.3.5 Critical Synthesis I .................................................................................................... 77 4.3.6 Evaluation I ............................................................................................................... 79 4.3.7 Parameter Identification II, Creative Synthesis II and Evaluation II ........................ 80 4.4 Final Conceptual Design ............................................................................................... 83 4.5 Analysis......................................................................................................................... 84 4.6 Discussion ..................................................................................................................... 88 Chapter 5: Conclusion .................................................................................................................91 5.1 Summary ....................................................................................................................... 91 5.2 Strengths and Limitations ............................................................................................. 93 5.2.1 Strengths ................................................................................................................... 93 5.2.2 Limitations ................................................................................................................ 94 5.3 Recommendations and Future Work ............................................................................ 96 5.4 Conclusion .................................................................................................................... 97   viii Bibliography .................................................................................................................................99 Appendices ..................................................................................................................................104 Appendix A Machine Drawings for Passive Deployment Mechanism .................................. 104 Appendix B Machine Drawings for New Deployment Mechanism ....................................... 106    ix List of Tables  Table 2-1: Deployment direction observed in -15⁰, 0⁰, and 15⁰ impacts. .................................... 35 Table 2-2: Peak force measurements ............................................................................................ 37 Table 3-1: Model contact parameters ........................................................................................... 50 Table 3-2: Force magnitudes of the first and second impact peaks in the model and experiment........................................................................................................................................................ 55 Table 3-3: Comparison of deployment modes observed in model and experiment ..................... 62 Table 4-1: Design requirements for a new PNT selector mechanism .......................................... 75 Table 4-2: Possible design configurations for a selector mechanism with a default deployment mode .............................................................................................................................................. 78 Table 4-3: Potential sensors for use in an active guide mechanism ............................................. 82    x List of Figures  Figure 1-1: Resting lordosis and forward flexed postures of the cervical spine ............................. 5 Figure 1-2: Euler buckling  ............................................................................................................. 6 Figure 1-3: Fundamental buckling mode shape and higher order buckling modes ........................ 7 Figure 1-4: Orientations of the stiffest and second-stiffest axes of the straightened cervical spine ......................................................................................................................................................... 8 Figure 1-5: Image of the segmented column model proposed by Voo et al. .................................. 9 Figure 1-6: Relationship between end condition and failure mechanism ..................................... 15 Figure 1-7: The Pro-Neck-Tor helmet .......................................................................................... 20 Figure 1-8: Early cylindrical prototype of the PNT helmet .......................................................... 21 Figure 1-9: Early deployment guide prototypes ........................................................................... 22 Figure 1-10: Examples of previous PNT helmet prototypes ........................................................ 23 Figure 1-11: LEATT™ brace ....................................................................................................... 26 Figure 2-1: Aluminum PNT helmet prototype.............................................................................. 30 Figure 2-2: Passive deployment mechanism ................................................................................. 31 Figure 2-3: Schematic of the head and PNT helmet in drop tower .............................................. 33 Figure 2-4: Typical force response for impacts to a horizontal surface ....................................... 36 Figure 2-5: Repeatability of force response .................................................................................. 37 Figure 2-6: Average outer shell kinematic response in 0⁰, 15⁰ and -15⁰ impact conditions ........ 39 Figure 2-7: Time lapse of head and helmet throughout 0⁰, 15⁰ and -15⁰ impacts ........................ 40 Figure 2-8: The anterior position of the carriage attachment relative to the deployment pin generates a moment that biases the head to rotate in extension .................................................... 41   xi Figure 3-1: Multibody model of experimental drop tower apparatus ........................................... 49 Figure 3-2: When the contact stiffness is low, the deployment pin is able to penetrate the guide and move through an escape path without rotating ...................................................................... 52 Figure 3-3: Sensitivity of impact force to changes in contact parameters .................................... 53 Figure 3-4: Agreement between representative experimental response and model force response in 0⁰ impact after tuning ............................................................................................................... 54 Figure 3-5: Comparison of experimental and model force response in 15⁰ and -15⁰ impact conditions ...................................................................................................................................... 55 Figure 3-6: Relationship between angular position and changes in contact stiffness ................... 56 Figure 3-7: Validation of model kinematics according to the helmet angular position and angular velocity .......................................................................................................................................... 58 Figure 3-8: Forward flexion of the head by 4⁰ improves the model agreement with experimental results ............................................................................................................................................ 59 Figure 3-9: Frame-by-frame comparison of model and experiment in a 0⁰ impact ..................... 61 Figure 3-10: Frame-by-brame comparison of model and experiment in a 15⁰ impact ................. 61 Figure 3-11: Frame-by-frame comparison of model and experiment in a -15⁰ impact ................ 62 Figure 3-12: Forces are transmitted to the head through the occipital condyles and the deployment pins ............................................................................................................................ 63 Figure 3-13: The moment acting on the head is the resultant of the moments due to the forces acting at the deployment pins and the occipital condyles. ............................................................ 64 Figure 4-1: When the impact platform is inclined at -13⁰ the moments arising from the forces at the occipital condyles and deployment pins are balanced ............................................................ 71   xii Figure 4-2: Conceptual design process using parameter analysis ................................................ 73 Figure 4-3: Anterior impacts can be identified by posteriorly directed reaction force ................. 80 Figure 4-4: Possible configuration of spring sensor ..................................................................... 81 Figure 4-5: Final selector mechanism design ............................................................................... 83 Figure 4-6: Action of the new selector mechanism in anterior impact ......................................... 84 Figure 4-7: Deployment of the PNT helmet with an active selector mechanism ......................... 86    xiii Acknowledgements  To Dr. Peter Cripton: thank you for your positivity and support throughout this process. This project would not be what it is without your enthusiasm and dedication. Your cause is a good one.  To my lab mates: what would I do without you? Angela, you have been so patient with me, and have taught me so much. I hope you know how wholly your help is appreciated. Hannah, thank you for refueling with me every day at 2:30. The coffee was good, but the company was better. To everyone else, thanks for making the lab so sunny, even on the cloudy days.   Thanks to my wonderful family, for guiding me along my ever-changing path. Pops, the breadth of your knowledge will never cease to amaze me. I hope I can one day have a fraction of your wisdom. Mom, thank you for being both my friend and my role model; and for being so fast.  I will always be racing to keep up with you. Alec, you are the better sibling.   And last, but not least, Meghan: thank you for being my constant reminder that life is an adventure. The next step in my journey would be far more daunting were it not for your example.  P.S. Come visit me in Sweden.    xiv Dedication  Mountains, You keep me sane.  1 Chapter 1: Introduction  1.1 Motivation Spinal cord injuries (SCI) have devastating consequences, and typically result in partial or complete paralysis. Additional complications may include loss of bowel control or the ability to participate in sexual intercourse [1], [2]. These symptoms can substantially reduce quality of life, and there is no cure.  Sporting accidents are a leading cause of SCI [3]. These injuries have been reported in football [4], mountain biking [5], snow sports [6], motocross [7] and hockey [8]. Head-first impact resulting in cervical spine fracture-dislocation is the predominant mechanism of injury [4], [8], [9]. During head-first impacts, axial loading of the cervical spine occurs and can cause fracture or dislocation of the vertebrae. This can cause bony fragments or vertebral components to enter the spinal canal and impact the spinal cord, resulting in spinal cord injury [10].  The annual incidence of spinal cord injuries in the United States is 40 cases/million individuals. Of those, nine percent occur due to sporting accidents [3]. Injuries sustained by helmeted individuals, as a result of head-first impacts, may be preventable with the development of new helmet technologies. While the size of this target population is not known, the incidence of catastrophic neck injuries in football players, specifically, was estimated to be 0.52-14/100,000 players [4]. Despite the rarity of these accidents, the economic costs are great. An individual who sustains an upper cervical spine injury at 25 years old will incur an average of $4,724,181 in health care and living expenses over the course of his or her life [11]. Efforts to prevent these   2 injuries are warranted by the substantial economic and personal costs that they yield. By employing our understanding of injury mechanisms, we may be able to reduce the prevalence of these events in sports ranging from football to motocross.  1.2 Cervical Spine Injury The mechanics of SCI are complex. Irregular vertebral geometry and complex articulations result in coupled motions that make predicting cervical spine motion challenging, during both normal physiological motion and complex injurious loading scenarios [12]. Furthermore, without muscles the osteoligamentous cervical spine is unstable and buckles under the weight of the head. Therefore, the neck can only be stabilized through activation of the surrounding musculature [13]. However, this neuromechanical system is highly redundant and poorly understood [14], which limits our ability to accurately replicate the in vivo mechanics of the cervical spine experimentally or with computational models. As a result, it is challenging to accurately determine injury tolerances and mechanisms. The following section explains our present understanding of cervical spine injury, obtained through epidemiological, computational and experimental studies. The focus is on axial loading, given its relevance to cervical spine injuries in sports.  1.2.1 Axial Loading and Cervical Spine Injury Cervical spine injuries are classified according to the perceived loading conditions at injury (flexion, extension, tension, compression, etc). Originally, investigators hypothesized that these loading conditions were the result of excessive head motion. For example, forward dislocation – a flexion injury – was believed to result from hyperflexion of the head and neck [15]. However,   3 failure to produce pure moment injuries in the laboratory eventually led to the assertion that injury due to pure hyperflexion or hyperextension is an ‘anatomical impossibility’ [16].   The role of axial loading in cervical spine injury has since gained increasing attention. In 1978 Bauze and Ardran produced bilateral facet dislocations by loading isolated cervical spines in compression [17]. In these experiments, an unrealistic stiffening constraint was applied to the inferior end of the specimens, by inserting a steel spindle into the lower vertebral canal. However, the results suggested that axial loading can give rise to complex intervertebral loading conditions, including shear and rotational forces, which may lead to flexion or extension-type injury mechanisms. Bauze and Ardran postulated that forward dislocations result from axial loading, rather than hyperflexion, and theorized that the injury mechanism does not necessarily reflect the motion or loading of the head. This was later corroborated by Nightingale et al., who conducted a series of dynamic drop tests with head-neck specimens. The injuries produced were indicative of the local deformations of the cervical spine, rather than the motion of the head [18]. Despite loading the specimens axially, the entire spectrum of cervical spine injury mechanisms (distractive/compressive, and flexion/extension) were observed due to the combined loading conditions that developed at the vertebrae. Thus, axial loads applied externally are not necessarily indicative of the intervertebral loading conditions and resulting injury mechanisms.  Numerous studies have provided supporting evidence for axial loading of the head and neck as the primary cause of injury in head-first impacts. Video recordings of sporting accidents have provided evidence of axial loading in clinical cases, while cadaveric tests have been used to produce various injury mechanisms under axial loading conditions in the laboratory.   4 Furthermore, the observed behavior of the axially loaded cervical spine is consistent with mathematical models, which approximate the spine as an ideal, slender column. Epidemiological, theoretical, and experimental findings that have contributed to the current understanding of axial loading in head-first impacts will be presented in the following sections.  1.2.1.1 Epidemiological Evidence: Cervical Spine Injury in Sports Epidemiological reviews have revealed considerable evidence of axial loading in sport-related cervical spine injuries. In 1975, Torg et al. conducted a four-year retrospective review of data from the National Football Head and Neck Injury Registry and attributed the majority of football-related severe cervical spine injuries to head-first impacts. Thereafter, the National Collegiate Athletic Association (NCAA) prohibited intentional striking with the head, known as spearing. Over the following 12 years, the incidence of these injuries decreased by 65%-70% [9]. The authors used video recordings to identify the mechanism of injury in 51 cases, and in every case the injury was attributed to head-first impact with another player or object. Similarly, Broglio et al. used video footage to investigate one incidence in which a football player fractured his sixth cervical vertebra. The injury was attributed to axial loading of the cervical spine, arising from an impact to the top right side of the helmet [19]. An investigation into hockey-related cervical spine injuries was performed by Tator et al., and revealed impact to the top of the head as the cause of injury in every one of the 42 cases investigated [8].   Torg et al. proposed that the forward-flexed, spear-tackling posture in football removes the natural lordosis, or curvature, of the cervical spine (Figure 1-1). In an impact, the straightened spine is constrained by the head and the torso. The torso continues to move towards the head   5 after the head is stopped, which causes compressive loading of the cervical spine [9]. The cervical spine then behaves as an axially loaded column and becomes prone to buckling failure, which causes sudden angular deformation of the vertebral elements that can lead to injury [20]. In non-injurious impacts the head is able to flex or extend away from the path of the following torso and safely dissipate impact energy [9].                       Figure 1-1: Resting lordosis and forward flexed postures of the cervical spine. Forward flexion of the head removes the resting lordosis and increases the alignment of the cervical spine. The straightened spine is vulnerable to axial loading injury. Adapted from Torg et al. [9] and used with permission from SAGE Publications.   Epidemiological and cinematographic evidence is insufficient to draw conclusions regarding the failure mechanics of the spine, given that the vertebrae cannot be directly visualized. However, we can postulate that (1) the forward-flexed cervical spine may be more vulnerable to injury than other postures, and (2) buckling may be the predominant failure mode.    6  1.2.1.2 Theory: Buckling and the ‘Most Vulnerable Posture’ Euler buckling theory describes the fundamental failure mechanism proposed by Torg et al. when the cervical spine is approximated as a slender column. The theory states that a slender column can support loads up to a critical point, after which it becomes unstable and fails by sudden outward deflection, known as buckling (Figure 1-2). The critical buckling load is the maximum load a column can support before buckling failure. If the applied load exceeds the material strength of the column before the critical buckling load is reached, the column will fail due to compression, rather than buckling. Typically, a buckled column will exhibit the fundamental buckling mode shape. However, rapidly applied loads may generate higher-order buckling modes (Figure 1-3).   Figure 1-2: When the critical buckling load, Pcr, of a column is reached Euler buckling occurs, characterized by sudden outward deflection away from the midline of the column (left). If the column is sufficiently short, the material failure will occur before the critical buckling load is reached (right). © eFunda, Inc., adapted from http://www.efunda.com/formulae/solid_mechanics/columns/intro.cfm with permission.    7  Figure 1-3: Buckled columns most commonly exhibit the fundamental buckling mode shape (left). However, rapidly applied loads can give rise to higher order buckling modes. © Indian Institute of Technology, adapted from http://www.nptel.ac.in/courses/Webcourse-contents/IIT-ROORKEE/strength%20of%20materials/lects%20&%20picts/image/lect36/lecture36.htm CC BY-SA  Failure of the cervical column may be attributed to either buckling or compressive material failure. Compressive failure gives rise to crush injuries, such as burst fracture of the vertebral bodies. Buckling failure causes rotation of the vertebrae which can lead to injuries such as wedge fractures or dislocations. It should be noted that buckling deformation can, but does not necessarily, result in material failure (i.e. injury).  As the stiffness of a column increases, the critical buckling load increases as well. This increases the loads that can be transmitted to the column in an impact, making the column more prone to compressive material failure. In other words, a low-stiffness column is unable to sustain large loads without deforming, and will buckle before injurious loads can develop. Conversely, a high-stiffness column will sustain large loads, but is more prone to catastrophic failure if the critical buckling load or material failure load of the column is reached. The cervical column is believed   8 to be stiffest – and therefore, most vulnerable to injury – when the spine is straightened and the centroidal axis is aligned with the impact load [21], [22]. However, there exists only one configuration in which these are precisely aligned, making this loading scenario improbable.   Liu and Dai mathematically demonstrated the existence of a ‘second-stiffest axis’ [22]. If a load is applied along this axis, the stiffness of the column will be a portion of the stiffness when the column is loaded in pure compression. Assuming the spine is represented as a homogeneous column this axis can be oriented in an infinite number of directions forming the surface of a cone (Figure 1-4). Thus, the straightened spine is vulnerable to injury in an infinite number of load orientations. The model predicts that the alignment of the cervical spine and orientation of the applied load are both important factors in axial loading injuries.     Figure 1-4: Figure 1-4 has been removed due to copyright restrictions. The figure showed the orientations of the stiffest and second-stiffest axes of the straightened cervical spine. Original source: Y. K. Liu and Q. G. Dai, The second stiffest axis of a beam-column: implications for cervical spine trauma, J. Biomech. Eng., 111(2), pp. 122–127, 1989 [22].   The model proposed by Liu and Dai contributes to our understanding of axial loading, but is limited by the assumption of a homogeneous column and a lack of experimental validation. A better approximation is achieved by modeling the spine as a segmented column (Figure 1-5). Voo et al. demonstrated that the critical load of a segmented column, like that of a homogeneous column, is strongly influenced by end constraint [23]. Therefore, reducing the constraint on the   9 head should reduce the loads on the cervical spine, and the risk of injury. Furthermore, the segmented column model predicts dislocation as the primary buckling mode, which is representative of the most common injuries observed both clinically and experimentally [24]. Computational modeling has shown good agreement between the segmented column model and cadaveric buckling kinematics [25].   Figure 1-5: Image of the segmented column model proposed by Voo et al. (right), compared to the ideal, homogeneous column (left) assumed in Euler buckling theory. The spheres in the segmented column model are assigned stiffness parameters for the intersegmental joints.  Adapted from L. Voo and Y. K. Liu, “Segmented Column Model to Predict Human Cervical Spine Buckling,” in Frontiers in Head and Neck Trauma: Clinical and Biomechanical, vol. 21, IOS Press, 1998, p. 442., [23] with permission from IOS Press and L. Voo.  Theoretical models are inherently limited by their simplification of the real world. However, they have offered insight into a complex system by demonstrating the importance of factors such as end constraint and spinal alignment. These models provide a mathematical basis for   10 understanding axial loading injury mechanisms observed clinically and experimentally, and exhibit parallels with real world observations.  1.2.1.3 Experimental Evidence: Axial Loading in the Laboratory Cadaveric experiments have shown that increasing vertebral alignment increases the forces on the cervical spine at failure [26], [27]. Therefore, the aligned cervical spine is more likely to undergo catastrophic bony failure, causing injuries such as burst fractures. As a result, the aligned cervical spine has been generally accepted by researchers as the most vulnerable posture. However, Pintar et al. refer to a ‘window’ of cervical spine alignment, within which cervical spine injuries can be produced by externally applied axial loads [26]. Thus, ‘axial’ loading applies broadly to approximately superoinferior loading. In fact, injuries due to axial loading have also been produced in neutrally aligned (lordotic) specimens [20], [28], [29]. These specimens have exhibited first-order, and higher-order, buckling behavior [20].   Nightingale et al. used a drop track to generate head-first impacts in lordotic specimens and observed first-order dynamic buckling behavior. Some specimens also exhibited a transient, higher order buckling mode. The buckling deformation corresponded to the local injury mechanism in 14 out of 15 injuries, and all injuries were sustained prior to significant head motion. They demonstrated that the neutrally positioned cervical spine is sufficiently slender to buckle prior to compressive failure. Furthermore, they showed that buckling causes local deformations in the axially loaded cervical spine, which can give rise to simultaneous flexion and extension-type injuries. Numerous other researchers have produced axial loading cervical   11 spine injuries under a variety of loading conditions [26], [30]–[33]. As a result, axial loading is generally accepted as the primary cause of injury in head-first impacts.   1.2.2 Factors Affecting Injury Axial loading injuries are affected by a number of parameters, including: cervical spine posture, alignment of the cervical spine with the major loading vector, head constraint, head orientation and neck muscle activation. These factors affect injury tolerance, as well as the types of injuries produced in head-first impacts. The effects of these parameters are described in the following sections.  1.2.2.1 Posture Cervical spine posture is challenging to define. It is usually described by eccentricity (the anteroposterior distance between the occipital condyles and T1), or curvature (kyphosis, lordosis, or aligned). Some studies have also attempted to quantify the orientation of the cervical spine axis relative to a reference plane [30], [34]. However, none of these methods provide a complete description of posture, and each metric may affect injury independently.  Pintar et al. quantified the pre-impact posture of head-neck specimens using eccentricity [26]. Aligned specimens (0±0.5 cm eccentricity) failed primarily in compression. However, some specimens maintained a slight initial curvature that altered the mechanism of failure. Thus, eccentricity does not fully describe posture, or govern the failure mechanics of the cervical spine. Still, eccentricity has been shown to significantly affect both the mechanism and severity of injuries. As eccentricity increases, injury type changes from fracture to soft tissue injury.   12 Maiman et al. tested specimens with eccentricities ranging from -0.5 to 10.2 cm, and grouped them by injury type. On average, the eccentricities of the fractured and non-fractured specimens were 0.6 cm and 5.2 cm respectively [32].   Pre-impact posture can also alter the trajectory of the head after impact [30]. Nusholtz et al. showed that when the cervical spine is flexed beyond a critical angle, impact causes ‘parabolic’ forward flexion of the head; less flexion results in axial ‘serpentine’, or ‘S’ shape, motion. This behavior is consistent with the escape concept described by Torg et al., which says that pre-flexed or pre-extended spines will bend out of the path of the major loading vector; whereas axial loading will occur when the spine is aligned with the direction of impact [9], [20].    Measures of cervical spine posture are not well defined in the literature. However, eccentricity and curvature are two important factors that affect the mechanism of injury. The combined effect of a straightened spine and low eccentricity increases the overall alignment of the cervical spine, which alters the type of injury and increases injury severity [32].     1.2.2.2 Torso Interaction Cervical spine injuries in head-first impacts have been attributed to loading from the following torso, rather than the loads resulting from the primary head impact [35]. The head and torso induce inertial constraints on the superior and inferior ends of the cervical spine, respectively. When the head, neck, and torso are axially aligned during an impact, the cervical spine becomes trapped between the torso and the head. The spine is then prone to compressive injury as the head is stopped by the impact and the torso, which still has momentum, continues to move   13 towards the head. When the alignment of the torso with the head and neck is decreased, torso interaction is reduced. This decreases the inferior inertial constraint imposed on the spine by the torso, which decreases the stiffness of the cervical column. As a result, the cervical spine cannot sustain the loads from the incoming torso without deforming. Thus, the spine bends out of the path of the torso before injurious loads can develop.   Nusholtz et al. confirmed the involvement of the torso in head-first impacts using a series of pendulum tests. Cadaveric subjects were impacted to the crown of the head with various torso and cervical spine orientations. When head-neck-thorax alignment was greatest, thoracic spine injuries were produced [36]. The authors concluded that torso involvement depends on the orientation of the thorax relative to the impact axis. They further concluded that the injuries produced were affected by head-neck-thorax alignment [34], [36].  1.2.2.3 Head Constraint Torg et al. postulated that potentially injurious axial loads may be dissipated through motion of the cervical spine, and the spine is therefore more vulnerable to injury when the head is constrained from motion [9]. Theoretically, this can be described by buckling mechanics. When a column is constrained, the stiffness of that column increases [23]. This increases the critical buckling load, and the compressive loads that can be transferred to the column in an impact.   Numerous researchers have confirmed that head constraint can affect injury potential. These studies include computational models as well as cadaveric experiments. Nightingale et al. applied axial loads to isolated cadaveric cervical spines, with the inferior end fixed to the upper   14 surface of a materials testing machine. One of three constraints was applied to the superior end: unconstrained, rotationally constrained or fully constrained (Figure 1-6) [29]. Increasing constraint increased the axial stiffness and the axial load at failure. End constraint also affected motion of the spine during loading, and the subsequent failure mode. Fully constrained specimens sustained compression or wedge compression injuries, while rotationally constrained specimens sustained bilateral facet dislocations. In contrast, the unconstrained specimens flexed out of the load path and were uninjured. These results agree with the model developed by Voo et al., which predicted that the critical load of a segmented column was primarily affected by end constraint [23]. Constraining end conditions may contribute to catastrophic injuries in head-first impacts.    15  Figure 1-6: Relationship between end condition and failure mechanism. Increasing end constraint alters the failure mechanism of axially loaded specimens. Figure reproduced from Nightingale et al. [29].   16 Nightingale et al. performed a series of dynamic drop tests to further explore the effects of end condition on cervical spine injury. They demonstrated that the inertia of the head is sufficient to constrain the cervical spine in impacts to rigid surfaces, by preventing the cervical spine from moving out of the path of the following torso. These inertial effects could be overcome if the impact surface was angled away from the axis of the cervical spine, to direct the head away from the following torso. The cervical spine was at the greatest risk of injury during impacts to 15° inclined surfaces (causing anterior head impact), when the impact surface was nearly perpendicular to the axis of the lordotic cervical spine [20], [35] Compliant padding further constrained the head due to ‘pocketing’, increasing the risk of injury.   Constraining effects of padding may be due to friction rather than surface compliance, as was described in computational models of the head-neck complex [37], [38]. Compliant padding may protect the cervical spine from injury, provided the surface friction is low. Camacho et al. recommended that the coefficient of friction in an impact should be below 0.2. However, this is hard to achieve in practice, and it is likely that the padding present in typical injury prevention devices would increase the likelihood of cervical spine injuries [39].   Head constraint can result from a number of factors, including head inertia, impact orientation, and surface friction. These constraints can significantly affect injury potential, and are important to consider when developing injury prevention technologies. Given that inertia is sufficient to constrain the head in some impacts, it may be necessary to use mechanical means to actively redirect the head if the spine is to escape the loads from the following torso and avoid injury.    17  1.2.2.4 Neck Muscles There are numerous challenges associated with simulating muscles in the laboratory, due to the mechanical complexity of the cervical spine and our poor understanding of muscle activation strategies [14]. As a result, a lack of simulated musculature is the primary limitation of most in vitro and computational modeling studies.   Cadaveric specimens require an external stabilizing force to support the head and maintain postures that are representative of physiologic conditions. In vivo, stability is achieved through compression of the spine by the neck muscles; however, these compressive loads are rarely simulated experimentally [20], [26]. Some researchers have attempted to simulate physiologic loading using external fixation devices [40] or follower loads [33], [41]. However, these loads are not applied in a biofidelic manner, and control over specimen posture is limited.   The effects of neck muscles have mainly been explored using computational models [42], [43]. Dibb et al. demonstrated that muscle activation can increase the preloads on the spine by 35.8 N to 1023 N – up to 45 percent of the cervical spine tolerance – depending on whether the muscles are in a relaxed or active state [42]. This could make the cervical spine more vulnerable to failure by compressive loading [37]. In a recent volunteer study by Newell et al., the effects of neck muscle activation on cervical spine posture were investigated. In the study, participants were asked to tense their neck muscles as though preparing for a head-first impact, and the motions of the cervical vertebrae were tracked using fluoroscopy. The authors found that active muscle contractions affected the pre-impact posture of the spine [44]. Therefore, the resting lordosis,   18 commonly used in cadaver experiments, may not be an accurate representation of cervical spine posture in real world impact scenarios. Our knowledge of cervical spine injury remains limited by our poor understanding of muscle contributions, and the paucity of injury tolerance studies that account for muscle contributions.  1.2.3 Injury Tolerance of the Cervical Spine Numerous factors affect injury type and severity during axial loading; including torso interaction [30], [34], [45], cervical spine posture [26], [27], [30], head constraint [20], [28], and muscle activation [37]. Thus, it is challenging to determine a single injury tolerance for the cervical spine [30]. Attempts to characterize the strength of the cervical spine have produced widely variable results. Alem et al. impacted full-body cadavers to determine the injury tolerance of the axially loaded cervical spine.  They reported a tolerance of 4.2 kN but also noted exceptions, including one specimen that failed at 3 kN and another that appeared uninjured at 16 kN [28]. Pintar et al. also attempted to determine the dynamic tolerance of the cervical spine by impacting upright head-neck specimens [26]. Failure loads ranged from 744 N to 6.4 kN with an average of 3.3 kN at frature.   Nightingale et al. attributed the range of reported compressive tolerances to different experimental conditions, such as end constraint and orientation. They reviewed the literature for studies that followed similar experimental protocols and accounted for dynamic effects. Both pre-flexed and lordotic specimens were included. Based on these studies, they suggested that the tolerance of the young male cervical spine is 3.6 to 3.7 kN [20]. Despite these efforts, much is still unknown about the tolerance of the cervical spine, specifically when parameters such as   19 spine posture and muscle preload are altered. As a result, developing effective injury prevention devices may be challenging, as the loads that cause injury are not precisely known.  1.3 The Pro-Neck-Tor: Concept and Development The Pro-Neck-Tor (PNT) helmet was designed to protect the cervical spine from compressive injuries in head-first impacts. Several researchers have concluded that head ‘escape’ – the ability of the head to move out of the path of the torso – is necessary for injury prevention [29], [35], [37], [38]. Yoganandan et al. demonstrated that injury potential increases when the head is restrained from motion [40]. Head motion can be restricted by constraints imposed by surface friction, head pocketing, inertia and impact angle. The PNT helmet induces head rotation and translation in the sagittal plane to allow the cervical spine to escape the momentum of the following torso. The protective mechanisms of the helmet are three-fold: the cervical spine is moved away from a vulnerable, aligned posture; impact energy is dissipated through controlled motion of the head and spine; and head constraint is reduced to allow the head to escape the momentum of the following torso.    Conceptually, the helmet consists of two concentric shells that are connected by bilateral guide mechanisms (Figure 1-7). Each guide mechanism is formed by a pin, connected to the inner shell, which inserts into a slot in the outer shell. The slot is bifurcated into two ‘escape’ paths. Upon axial impact, the pin moves through one of the two escape paths in the guide mechanism, causing the inner shell and head to move relative to the outer shell. Motion through the anterior escape path results in flexion and anterior translation of the head. Motion through the posterior escape path results in extension and posterior translation of the head. These are referred to as the   20 flexion and extension deployment modes, respectively. Flexion deployment should occur in impacts that are perpendicular, or slightly posterior to the crown of the head. Extension deployment should occur in impacts slightly anterior to the crown of the head. The preferred deployment path for each impact condition was determined experimentally, by investigating which deployment mode produced the greatest load reductions on a mechanical surrogate cervical spine in head-first impacts [46]. The load threshold at which the helmet deploys is currently undefined. The helmet is intended to deploy in axial impacts that exceed the tolerance of the cervical spine; however, the head impact load must be used as a predictor of the loads on the cervical spine and may be affected by posture and torso interaction. In the future, an appropriate threshold will be determined and incorporated into new designs.                  Figure 1-7: The Pro-Neck-Tor helmet. The Pro-Neck-Tor helmet (left) consists on an inner shell and an outer shell, which are connected by bilateral guide mechanisms (right). Each mechanism contains a deployment pin (yellow) which interfaces with a slot in the outer shell. The slot is bifurcated into two ‘escape’ paths. Motion of the pin through one of the escape paths causes flexion or extension of the head and neck. © 2008 The University of British Columbia, adapted from http://www.pronecktor.com/ with permission from P. A. Cripton.    21 The helmet was designed to produce 25 mm of horizontal translation, and 19 mm of vertical displacement, resulting in 15° of head rotation. However, these parameters have not been optimized. The dimensions were chosen based on evidence that injury mechanism changes from vertical compression to compression-extension, or compression-flexion with -5 and 22 mm of eccentricity, respectively. Furthermore, increased eccentricity has been shown to reduce injury severity [32].   Originally, a cylindrical prototype was developed for proof of concept (Figure 1-8) [46]. The deployment pins were fixed directly to a surrogate head mass, and interchangeable guides were used to produce flexion or extension deployment. Each guide provided only one deployment path. The deployment mode was therefore configured prior to impact.    Figure 1-8: Early cylindrical prototype of the PNT helmet  This image shows an early cylindrical helmet prototype (left). The deployment pins were rigidly attached to a surrogate head mass (middle) and interfaced with conformal slots (right) in the helmet shell. Reproduced from Nelson et al. [47], with permission from P.A. Cripton.    22       Figure 1-9: Early deployment guide prototypes. Early prototypes had interchangeable flexion and extension guides, and the deployment mode had to be configured prior to impact. This image shows the right deployment guide configured for extension (left) and flexion (right) deployment.    Flexion deployment was found to reduce the neck injury criteria, Nij, in perpendicular and posterior-to-vertex impacts; and extension deployment reduced injury metrics in anterior-to-vertex impacts [46]. Flexion deployment reduced peak axial loads by 40% in impacts to perpendicular, rigid surfaces. Based on these findings, it was determined that the deployment mode should be selected according to the location of an impact on the head, and this premise has guided the development of the helmet to date. As our understanding of cervical spine injury mechanisms improves, the parameters used to determine the appropriate deployment mode may change.   A series of helmet prototypes have been developed since the initial cylindrical prototype, and have shown similar peak axial force reductions. These include one aluminum PNT football helmet, one carbon fiber PNT football helmet, and one PNT ski helmet (Figure 1-10). Each of these prototypes must be configured for a given deployment mode prior to impact; however, the   23 final PNT helmet must be capable of automatically selecting the deployment mode at the time of impact. Therefore, it is necessary to develop a ‘selector mechanism’ that is capable of initiating deployment according to the impact conditions. We hypothesize that the head will naturally rotate in flexion or extension under the dynamic loads present, and according to the impact angle, resulting in passive selection of the appropriate deployment mode when two deployment paths are available. This hypothesis is based partly on the motions observed in cadaveric drop tests to inclined surfaces [20], and partly on a fundamental understanding of how objects are deflected in impacts. However, this hypothesis has not yet been tested.   Figure 1-10: Examples of previous PNT helmet prototypes  1.4 Existing Devices Induced head escape has previously been explored in the context of rollover accidents. In these accidents, the cervical spine can be injured by axial loading when an inverted occupant ‘dives’ into the roof of a vehicle. Hu et al. suggested introducing a low-friction interface between the roof and layer of protective padding. The padding would dissipate impact energy while the low-  24 friction interface would reduce head constraint [37]. Halldin et al. proposed a similar system, which produces relative motion of an inner roof with respect to an outer roof in an impact [48]. Finite element analysis of the system showed a 27% reduction in T1 forces when the head contacted the roof in the neutral position. A 44% reduction was observed when the head was flexed forward 15⁰ prior to impact. Similarly, Huedorfer et al. proposed a seat-mounted airbag that deploys over-head, and moves the occupant’s head forward during a rollover [49]. This causes flexion of the cervical spine away from the following torso mass and out of an aligned posture. The roofbag reduced the compressive loads on the neck by 38% in dynamic rollover tests, and the neck injury criterion was reduced to a non-critical value.   Induced head motion appears to effectively reduce the loads on the cervical spine in head-first impacts. However, these devices are not directly applicable to sporting applications. Furthermore, they do not address the need to induce different head motions (flexion or extension) in different impacts. Technologies that have been developed specifically for sporting applications will be discussed in the following sections.  1.4.1 MIPS Helmet MIPS (Multi-directional Impact Protection System, MIPS AB, Stockholm, Sweden) is a helmet designed to induce rotation in oblique impacts, in order to reduce the angular acceleration of the head and prevent concussion. Like the PNT, this helmet consists of two concentric shells that move relative to one another in an impact. However, the MIPS helmet differs from the PNT in several important ways. First, maximum rotation between the two shells occurs in oblique impacts [50], and is on the order of millimeters. As impacts become more axial, less motion is   25 induced. The PNT is intended to induce motion an order of magnitude larger in axial loading scenarios. Furthermore, the MIPS inner helmet shell is limited to rotation; whereas the PNT inner shell and head undergo combined rotation and translation. The MIPS technology is not designed to induce enough motion to allow the head to escape the torso in a head-first impact, and has no mechanism to accomplish this. Therefore, it is not applicable to cervical spine injury prevention.  1.4.2 LEATT Brace The LEATT™ brace was developed to prevent catastrophic cervical spine injuries in head-first impacts. However, the technology is based on the assumption that most catastrophic cervical spine injuries are caused by excessive flexion, extension or lateral motion of the head. The brace is therefore intended to prevent large head motions, rather than axial loading of the cervical spine (Figure 1-11) [51]. When an impact occurs, large head motions cause the bottom of the helmet to contact the top of the brace, and the impact loads are transmitted to the torso rather than the cervical spine. Therefore, the loads on the cervical spine are affected only when the head moves far enough for helmet-brace contact to occur.   The brace reduces the loads on the cervical spine by offering an ‘alternate load path’. The inventors do not claim the brace protects against pure axial loading. It is unlikely that it would offer protection in this manner, given that the helmet only contacts the collar when large head motion occurs.      26   Figure 1-11: LEATT™ brace. The range of motion of the head (top) is reduced with the LEATT™ brace (bottom). The brace is intended to prevent excessive head excursion, and offers an alternate load path when the bottom of the helmet contacts the top of the brace. Adapted from Vaughan et al. [51], with permission from Leatt Corporation®.  1.5 Objectives & Scope The overall objective of this work was to assess the functionality of a passive PNT selector mechanism and suggest design improvements based on that assessment. Three investigations were conducted, with the following specific objectives: 1. Assess: Prototype a passive, bi-directional selector mechanism; qualitatively evaluate the mechanism functionality in a series of drop tower tests; and collect kinematic and force data for future comparison with a multibody dynamics model   27 2. Model: Develop a multibody model of the PNT helmet and simulate a series of drop tower tests in order to evaluate the dynamics governing the deployment mode 3. Design: Propose a new design for the PNT selector mechanism, based on findings from experimental drop tests and multibody modeling  The scope of this project was to evaluate the functionality of a passive PNT selector mechanism, and suggest design improvements to improve the mechanical performance of the device. No detailed attempt was made to characterize the effects of helmet deployment on neck loads or injury potential. This work satisfies a portion of a larger research project, which aims to develop a fully functional PNT helmet and evaluate how induced head motion affects the risk of injury in head-first impacts.    28 Chapter 2: Evaluating the Existing PNT Helmet Mechanism  2.1 Introduction The PNT helmet was designed to prevent cervical spine injuries, such as burst fractures and fracture-dislocations of the cervical vertebrae, that occur in head-first impacts. It consists of an inner shell and an outer shell, which are connected by an internal deployment mechanism. In an impact, the mechanism deploys and guides the head into flexion or extension. To minimize the forces on the cervical spine, the deployment direction must be selected according to the impact conditions. Extension deployment should occur in head-first impacts when the impact occurs anterior to the crown of the head, and flexion deployment should occur in impacts posterior or perpendicular to the crown of the head, as was determined from drop tests performed with a mechanical surrogate cervical spine [46].  In previous tests in our laboratory, the deployment mode has almost always been configured prior to impact. Early prototypes were modular, and interchangeable flexion and extension guides were installed to produce the desired motion on impact. Ultimately, the helmet must be able to direct the head through the appropriate path in a bi-directional (flexion and extension) guide, based on the impact conditions. To do so, the helmet must sense the location of an impact (anterior, posterior, or vertex), and select the deployment mode accordingly. The impact conditions should be sensed by mechanical means, and the helmet should not rely on electronic sensors or active actuators. The inventors, Drs. Nelson and Cripton, posited that the reaction force in anterior and posterior impacts would deflect the head away from the impact surface, resulting in passive selection of the appropriate deployment mode. It was also predicted that the   29 forward-flexed pre-impact posture would cause passive flexion deployment in horizontal impacts.  The primary objective of this study was to evaluate a passive selector mechanism with two deployment paths. A prototype was developed and tested with a Hybrid III (HIII) head in impact conditions, using a drop tower. A custom drop tower attachment was developed to control the loads applied to the head and PNT selector mechanism. The mechanism was assessed based on whether or not it induced the appropriate motion and deployed in a consistent manner in a given impact condition. Kinematic and force data were collected to characterize the helmet response quantitatively. These data were intended for tuning and validation of a multibody model of the helmet.   2.2 Methods 2.2.1 Selector Mechanism Design The aluminum prototype of the Pro-Neck-Tor helmet, which was previously developed, was used for all experimental tests. The prototype includes an inner and outer shell, and has a modular design to accommodate different guide and pin geometries (Figure 2-1). A passive selector mechanism was designed for integration with this helmet (see Appendix A  for machine drawings). The mechanism consists of a straight deployment pin that interfaces with a symmetric bifurcated slot (Figure 2-2). This slot forms the flexion and extension escape pathways, which are oriented 60⁰ from the vertical. This angle was selected to achieve 25 mm of anteroposterior excursion and 19 mm of vertical displacement, as specified in the original helmet design. Prior to deployment, the pin is oriented vertically, and is held in place by friction. On impact, the pin   30 drops out of the vertical slot and can move through either of the two escape paths. The guide was made out of Delrin to provide a low friction guide interface, and allow the aluminum pin to slide smoothly through the slot. Future designs will include a force limiter, which prevents deployment unless a force threshold is exceeded. Prototyping a force limiter was deemed unnecessary for the present study, as its presence or absence should not affect the selected deployment mode. A limiter can be incorporated into the mechanism once the selector function has been refined.     Figure 2-1: Aluminum PNT helmet prototype (left). The inner shell (top right) and outer shell (bottom right) have modular designs in order to accommodate different pin and guide geometries.    31    Figure 2-2: Passive deployment mechanism. The mechanism consists of a symmetric bifurcated slot (left) and a deployment pin (middle). Prior to deployment, the pin is oriented vertically within the guide (right). Upon impact, the pin becomes free to slide through either of the deployment paths.   2.2.2 Drop Tower Apparatus A custom, free-standing drop tower, which was previously developed in our laboratory, was used to drop the helmet onto a horizontal or sloped surface [46]. The carriage was modified to interface with the Hybrid III head, without a surrogate neck. The modifications replaced the previously used surrogate neck [52] with simple joints, so the drops could be easily reproduced in multibody dynamics software. Using basic constraints also simplified the dynamic analysis of the deployment mechanism. The helmet was attached to a Hybrid III head using a Riddell chinstrap, and inverted in the drop tower (Figure 2-3). A shaft was attached to the inferior surface of the head and inserted into an aluminum bearing housing that was rigidly fixed to the drop carriage. Thus, the head was suspended from the carriage by a simple revolute joint, at the approximate location of the occipital condyles (7 mm inferior to the occipital condyles on the Hybrid III head). This joint allowed the head to rotate in flexion or extension, while controlling the direction of the loads applied to the head through the occipital condyles. The length of the bearing housing was selected so the head could rotate without obstruction by the carriage. The 16 mm   32 drop carriage weighed 16 kg and was intended to approximate the effective mass of the following torso in a typical head-first impact [20]. The carriage was limited to vertical translation by four linear bearings. The HIII head was positioned with the Frankfort plan aligned with the horizontal. It was held in place by strings attached to the outer helmet shell, which slackened on impact. The bottom edge of the deployment guide was leveled such that the deployment pin was perpendicular to horizontal prior to impact.   A steel impact platform was covered with two sheets of thin (5 mm) padding. The angle of the impact platform was varied (0°, +15°, -15°) to generate impacts anterior, posterior and perpendicular to the crown of the head. The drop height was selected to achieve an impact velocity of approximately 2.0 m/s. Five drops were conducted for each platform angle. The resulting deployment direction was recorded in each impact.     33  Figure 2-3: Schematic of the head and PNT helmet in drop tower   2.2.3 Data Collection A single axis load cell (LC 402-5K, Omega, Laval QC, Canada) beneath the impact surface measured the head impact force. A six-axis load cell (4366J, Denton, Humanetics, Plymouth MI, USA) attached to the carriage measured the force transmitted through the occipital condyles. Load data were sampled at 50 kHz using a data acquisition system (NI cDAQ-9172, module NI 9215, National Instruments, TX, USA), and a custom LabVIEW program (National Instruments, TX, USA). The data were lowpass filtered according to SAE J211, and cross-talk in the six-axis load cell was corrected for using data provided by the manufacturer. Two high-speed cameras (Phantom V12, Vision Research, Wayne NJ, USA) recorded the drops in the frontal and sagittal z x   34 planes at 3000 frames per second. A trigger was used to synchronize the cameras and the force data collection. Fiducial markers on the outer helmet shell were tracked using 2D image analysis software (TEMA, Image Systems, Linkoping, Sweden) to determine the kinematics of the shell. Head motion was not tracked due to obstruction of the head from the camera view by the outer helmet shell.  2.3 Results The helmet deployed in a consistent manner in each impact condition (Table 2-1). In the 0⁰ and 15⁰ conditions, extension deployment occurred in every trial. In the -15⁰ impact condition, flexion deployment occurred in every trial. Thus, the desired deployment mode occurred in the +15⁰ and -15⁰ impacts. However, extension deployment occurred in the 0⁰ impacts, when flexion deployment was preferred. The deployment direction appeared to be affected by the moment acting on the head as a result of the loads from the incoming torso.          35 Table 2-1: Deployment direction observed in -15⁰, 0⁰, and 15⁰ impacts. Cells highlighted in green indicate selection of the appropriate deployment mode. Cells highlighted in red indicate inappropriate selection.  Deployment Direction Drop # 15⁰ Impact 0⁰ Impact -15⁰ Impact 1 Extension Extension Flexion 2 Extension Extension Flexion 3 Extension Extension Flexion 4 Extension Extension Flexion 5 Extension Extension Flexion  Figure 2-4 shows a typical force response for drops to a horizontal surface. Two peaks are evident in the impact force, measured by the load cell under the impact platform. The first peak occurs when the aluminum outer helmet shell contacts the impact platform. This large initial force is not transferred to the head. The second peak occurs when the deployment pin contacts the top of the guide inside the selector mechanism. At this point the impact loads are transferred to the head. Because the head is rigidly attached to the carriage mass by the bearing housing, this impact force develops in phase with the load transmitted to the upper load cell. This pattern was observed in all three impact conditions. In drop tests with cadaveric specimens the head impact force develops out of phase with the forces on the lower neck, due to the compliance of the cervical spine. As a result, the head impact force appears bimodal in these tests [20]. This bimodal response was not observed in the present experiments because the compliance of the cervical spine was not represented. The oscillations in the impact force are a caused by ringing of the aluminum outer shell.   36  Figure 2-4: Typical force response for impacts to a horizontal surface. The first impact peak is caused by the outer helmet shell contacting the platform. The second impact peak is caused by the deployment pin contacting the top of the guide. The upper load cell force develops in phase with the impact force.  The magnitude and shape of the force response was repeatable (Figure 2-5). In each impact condition, the coefficient of variation of the peak impact load was less than 7%; and the coefficient of variation of the peak load at the upper load cell was less than 4%. The average impact velocity was 2.06 m/s, 2.06 m/s and, 2.07 m/s for the -15⁰, 0⁰, and 15⁰ impacts, respectively. The average peak forces at the upper and lower load cells are provided in Table 2-2.     37   Figure 2-5: Repeatability of force response. The magnitude and shape of the force response was repeatable in all impact conditions.  Table 2-2: Peak force measurements  0⁰ Impact 15⁰ Impact -15⁰ Impact Helmet Impact Force (SD), [N] 12 829 (776) 9 238 (570) 12 170 (512) Upper Load Cell Force (SD), [N] 3 079 (63) 2 349 (73) 105 (7)    38 The angular position and angular velocity of the outer helmet shell with respect to a global coordinate system were calculated from the tracked fiducial markers. The average kinematic response is shown in Figure 2-6. In the 0° and 15° conditions, the outer shell rotated in the negative y-direction (Figure 2-7). Eventually, the outer helmet shell contacted the chin of the HIII head, causing the direction of rotation of the helmet shell to reverse. This event corresponded to the time of peak angular position and inflection of the angular velocity curve from negative to positive. Conversely, in the -15° condition, the outer helmet shell continuously rotated in the positive y-direction, as seen by the increasingly positive angular position.  Corresponding images from high-speed videos are shown in Figure 2-7. It was clear from the videos that appreciable head motion did not occur until after the peak loads developed at the upper load cell, at approximately 20 ms. Head rebound off the impact platform was evident in all impact conditions. The exact time of rebound is unclear, but force data suggest that rebound off the platform occurred at 30-40 ms in all conditions.  Deployment occurred smoothly in the +15° and 0° impact conditions. The head rotated continuously in extension upon impact, and head motion was evident by 30 ms. The +15° condition showed the greatest amount of early head motion. Deployment did not occur as smoothly in the -15° impacts. There was a distinct axial loading phase in which the head rebounded off the impact platform still in an aligned posture (30-40 ms). The impact caused tilting of the drop carriage due to play in the linear bearings. The head was then ‘whipped’ forward during the rebound phase (40-80 ms).     39 The head motions observed throughout the approximately 150 ms deployment time were large. Extension deployment caused the head to rotate until the face was within the outer helmet shell. Contact occurred between the face of the HII head and the outer shell. In flexion deployment, the head rotated until the chin contacted the bearing housing.   Figure 2-6: Average outer shell kinematic response in 0⁰, 15⁰ and -15⁰ impact conditions    40  Figure 2-7: Time lapse of head and helmet throughout (a) 0⁰ impact (b) 15⁰ impact (c) -15⁰ impact  2.4 Discussion The selector mechanism deployed in a consistent manner in each impact condition: extension deployment occurred in all of the 0° and 15° impacts, and flexion deployment occurred in every -15° impact. Thus, the behavior of the mechanism appears to be robust. However, extension deployment in the horizontal impact condition is undesirable. Evidently, the passive selector mechanism, as designed, is unable to select the appropriate deployment mode in all impact conditions.  z x   41 The factors governing deployment are not obvious from the experimental data. However, it appears that the anterior position of the carriage attachment relative to the location of the deployment pin generates a moment that biases the head to rotate in extension in the absence of anteriorly directed ground reaction forces, such as those present in the -15° impacts (Figure 2-8).  This suggests that the loads from the incoming torso may affect the deployment mode, and that changes in the pre-impact posture could alter the deployment mode in a given impact condition. A comprehensive analysis of the forces acting on the deployment mechanism is necessary to confirm this hypothesis. Ultimately, deployment selection should depend only on the location of an impact on the head, and should be unaffected by the loads applied inferiorly.   Figure 2-8: The anterior position of the carriage attachment relative to the deployment pin generates a moment that biases the head to rotate in extension  In the 0° and -15° impacts the majority of head motion occurred after the peak loads developed at the upper load cell and the head rebounded off the platform. Deployment in the 15⁰ condition occurred most smoothly, and may have caused sufficient head motion prior to rebound to reduce   42 the neck loads; however, further testing with a biofidelic neck form is necessary to confirm the effects of this motion on cervical spine loading. To effectively dissipate impact energy and guide the neck out of a vulnerable posture, head motion must occur throughout the development of the peak loads at the upper load cell, which are indicative of the loads that would be transferred to the cervical spine. The head rebound observed is an indication that the deployment action of the current mechanism is ineffective. The mechanism should redirect the impact forces to cause anterior or posterior motion of the head, rather than vertical rebound. Future designs should smoothly guide the deployment pin into the desired escape path without causing head rebound.   In contrast, the head motions following rebound were large. The head underwent approximately 60° of rotation, and as a result contacted the inside of the outer helmet shell during extension deployment, and the bearing housing during flexion deployment. Contact with the inside of the outer helmet shell could result in facial injuries, and there may be other unforeseen consequences associated with inducing large head motions. The motions induced in the present experiments were extreme and unnecessary, given that reductions in neck loads were previously achieved with only 15° of head rotation [46]. In the future, the geometry of the guide should be optimized to maximize force reduction without generating extreme head motions. Testing with biofidelic surrogate cervical spines is recommended to determine the optimum geometry.  The drop tower apparatus provides an effective means of evaluating the behavior of the selector mechanism under known loading conditions. The simple joints connecting the Hybrid III head to the drop carriage provided control over the impact conditions, and the apparatus demonstrated good repeatability. Modeling the cervical spine would have introduced complex motions and   43 loading conditions that may have confounded the analysis of the deployment mechanism. By reducing the system to only the head and helmet it was possible to isolate the effects of the selector mechanism from motions caused by torso-cervical spine-head interactions.   However, as a result of the simplified drop tower constraints the loading conditions on the head were not biofidelic. The complex articulations of the cervical vertebrae were not represented, and would be unlikely to give rise to pure cranial-caudal loads at the occipital condyles in a head-first impact. Furthermore, because the compliance of the spine is not simulated, the force-time load characteristics observed were not representative of the in vivo condition. In the current experiments the impact force developed in phase with the force at the carriage mass. However, human cadaveric specimens exhibit a bimodal response, where the peak head impact force develops prior to the loads from the following torso [20]. Therefore, the loads transmitted to the selector mechanism in these tests do not perfectly represent the conditions in real-world impacts. Moreover, the revolute joint does not accurately represent the stiffness or motion characteristics of the atlanto-occipital joint.   The aluminum prototype weighed 6.8 kg, and is not a realistic representation of the final PNT helmet. However, it is a useful design tool due to its strength and rigidity. Other plastic and carbon fiber prototypes have been made, but cannot easily accommodate new guide geometries. Additionally, these helmets deform on impact, which can complicate the assessment of the deployment mechanism and prevent repeatable testing. Thus, the aluminum shell was the most appropriate device for evaluating a new mechanism.    44 The predominant issues with the passive selector mechanism were three-fold: (1) the mechanism does not deploy in the correct direction in horizontal impacts, (2) the loads from the incoming torso may influence the deployment mode, and (3) the mechanism is unlikely to generate sufficient head motion prior to rebound to prevent injury. A new mechanism must be developed to address these shortcomings. To accomplish this, additional investigations must be conducted to fully understand the dynamics governing the deployment mode. A multibody model of the PNT helmet was developed for this purpose. The model is presented in the following chapter.    45 Chapter 3: Multi-body Modeling of the PNT Helmet  3.1 Introduction The PNT selector mechanism is intended to induce extension in anterior impacts, and flexion in posterior and vertex impacts. Dynamic drop testing of a passive PNT selector mechanism in Chapter 2 demonstrated that the mechanism was not able to automatically select the desired flexion deployment mode in impacts to horizontal surfaces. It also failed to induce appreciable head motion prior to head rebound and the development of peak loads at the occipital condyles, suggesting that the mechanism has limited ability to reduce the loads on the cervical spine in head-first impacts. The selector mechanism must be improved so it can induce appropriate head motion in all impacts.   To develop an effective mechanism, the dynamics governing the deployment mode must be better understood. During experimental testing, the impact force and the forces transmitted through the occipital condyles were measured.  However, this information is insufficient to determine how the impact loads are transferred to the selector mechanism to initiate flexion or extension deployment. Therefore, a multibody dynamics model of the experimental tests described in Chapter 2 was constructed in MSC Adams (MSC Software, Santa Ana, CA, USA), and validated against kinematic data gathered from the physical drops. A dynamic analysis was then conducted to better determine the factors governing deployment, which were used to inform the development of an improved selector mechanism. The model was also intended as a platform for testing new selector mechanism designs, to reduce the time and costs associated with physical prototyping.     46  Multibody dynamics modeling is a computationally inexpensive modeling technique. However, accuracy is sacrificed when the modeled system deviates from rigid body assumptions. It is important to understand the limitations associated with multibody modeling in order to implement the technique appropriately. The primary challenge associated with modeling impact events in Adams is the approximation of contact forces. The following section discusses how contact forces are estimated in Adams, and the limitations inherent in this approximation.   3.1.1 Modeling Contact in MSC Adams The Adams IMPACT contact force algorithm is adapted from the Hertzian Theory of Elastic Deformation for static contact between elastic bodies [53]. Hertzian theory defines a contact stiffness based on the geometry and material properties of two contacting bodies. The theoretical contact force can then be calculated based on the contact stiffness and deformation of the contacting bodies [54]. In order to extrapolate Hertz theory to dynamic situations, the software developers incorporated a damping term into the Adams contact algorithm [53]. Adams contact forces are, therefore, calculated according to the following spring-damper contact model:  𝐹 =  {0                                                                                            𝑖𝑓 𝑥 ≥ 𝑥1𝑘(𝑥1 − 𝑥)𝑒 − 𝑐𝑚𝑎𝑥?̇? ∗ 𝑆𝑇𝐸𝑃(𝑥, 𝑥1 − 𝑑, 1, 𝑥1, 0)       𝑖𝑓 𝑥 < 𝑥1  Contact occurs when the distance between two markers on opposing bodies, x, falls below a nominal distance, x1. Therefore, (x1 – x) is a measure of penetration depth, and ?̇? is a measure of the impact velocity. The remaining four parameters (stiffness, k, damping, cmax, force exponent, e, and penetration depth, d) are user-defined properties that determine the force response. ‘STEP’ indicates the use of a cubic polynomial to approximate the Heaviside step function.   47  The contact stiffness, k, is dependent on both material properties and contact geometry, and is unique for every contact. The force exponent, e, describes whether the contact has the characteristics of a stiffening (e>1), linear (e=1), or softening (e<1) spring. This parameter is mostly material dependent, and is easier to estimate using ‘rules of thumb’ than the contact stiffness. Both the contact stiffness and force exponent are based on Hertz theory.  Conversely, the maximum damping coefficient, 𝑐𝑚𝑎𝑥, has no theoretical equivalent. During contact, the damping coefficient is ramped up to 𝑐𝑚𝑎𝑥 using a cubic step function over a small penetration depth, 𝑑. This is necessary to avoid discontinuities in the contact function, which can cause integration errors during a simulation. Thus, the penetration depth does not describe a physical parameter; its only purpose is to assist integrator function.  Determining appropriate values for four independent contact parameters is not trivial. There are several methods of estimating these parameters; however, none is particularly robust. Hertz theory can provide initial estimates for contact stiffness and force exponent; however, the Hertz contact model was derived for simple geometries, linear material behavior, and static contact, and as a result theoretical estimates can be orders of magnitude different from appropriate modeling parameters. Empirical measurements tend to provide more appropriate estimates of the contact parameters, and can be used to determine the damping coefficient, the contact stiffness, and the force exponent. However, taking these measurements is not always practical. Alternatively, the parameters can be tuned until the contact force matches experimental force measurements; but the contact parameters do not act independently, and the potential   48 combinations are infinite. Without experimental data for a specific impact event, it is challenging to accurately determine the contact parameters, and modelers often rely on experience to set these values.   Once the contact parameters have been determined for a specific impact event, they are only relevant for a small range of scenarios. If the impact geometry or velocity is altered, the parameters become invalid. In addition to generating the desired force response, the selected parameters must prevent large penetrations between the contacting bodies, and allow for computational efficiency. Often there are tradeoffs associated with satisfying these requirements. Regardless of the method used to determine the contact parameters, the Adams contact algorithm dramatically simplifies real world contact behavior since the bodies are assumed to be rigid, and a single contact force is applied at the center of the contact area.   3.2 Methods A multibody model was constructed in Adams to simulate the drop tower experiments described in Chapter 2 (Figure 3-1). The geometry of the carriage mass, bearing housing, helmet, and Hybrid III head were imported directly into the model. The drop carriage, housing and helmet were assigned the density of aluminum, 2.74x10-6 kg/mm2. The mass of each body was calculated according to the imported geometry. The Hybrid III head was assigned a mass of 4.4 kg. A frictionless revolute joint was placed at the center of the shaft attaching the head to the bearing housing and carriage mass. The inner helmet shell was rigidly fixed to the center of gravity of the head, to prevent relative motion between the head and inner shell. The helmet was oriented with the deployment pins perpendicular to the ground. The anteroposterior and cranial-  49 caudal position of the helmet on the head was visually approximated to match the typical experimental setup.    Figure 3-1: Multibody model (left) of experimental drop tower apparatus (right)  The carriage mass was constrained to cranial-caudal translation, and the initial velocity of all bodies was set to 2.06 m/s to match the average experimental impact velocity. The angle of the impact platform could be varied to achieve a 15°, 0° or -15° incline. Gravity was set at 9.81 m/s2. The integrator step size, Hmax, defines the maximum time step the integrator can take, and was set at 0.1 ms to ensure that several data points were generated throughout each impact event. The integrator error was decreased until the results converged, meaning the output displacements and velocities changed by less than 1%. This occurred at an integrator error of 0.001 mm.   Contact was defined between the deployment pins and the interfacing deployment guides, causing the outer helmet shell to rest on the deployment pins. Contact was also defined between z x   50 the outer helmet shell and ground, the head and bearing housing, the inner and outer helmet shells, and the outer helmet shell and head (Table 3-1). The IMPACT function was used for all contacts.   Table 3-1: Model contact parameters.  *Head-bearing housing, inner shell-outer shell, and outer shell-head contacts  Stiffness, k [N/mm] Force Exponent, e  Damping, c [N·ms/mm] Penetration Depth, d [mm] Helmet-Ground 300 1.5 6.0 0.01 Pin-Guide 2,000 1.5 1.5 0.01 Other Contacts* 1x105 2.2 10 0.1  The ground reaction force was affected by the helmet-ground and pin-guide contact definitions. This force is ultimately transferred to the deployment mechanism to initiate deployment, and could therefore influence the deployment direction. Thus, the helmet-ground and pin-guide contact parameters were chosen to match the force response observed experimentally.   The penetration depth was set at the value recommended in the Adams documentation, 0.01 mm. The force exponent for both the helmet-ground (aluminum-padded steel) and pin-guide (aluminum-Delrin) contacts was set at the generally accepted value for aluminum-aluminum contact, 1.5. The stiffness and damping parameters were then tuned independently until the impact force generated by the model matched the experimental force response. Tuning was   51 conducted for the 0⁰ impact condition only, and the tuned parameters were then applied to the 15⁰ and -15⁰ conditions.  Hertz theory was used to provide an initial estimate of the helmet-ground and pin-guide contact stiffnesses. The theoretical values overestimated the stiffness of the system, and the helmet-ground and pin-guide contact parameters were tuned independently thereafter. First, the helmet-ground contact stiffness was decreased to match the magnitude of the first peak observed in the experimental force response within 5%. Damping was then gradually increased from zero until the impact force developed in phase with the force at the base of the carriage mass, as observed experimentally (Figure 2-4). Next, the pin-guide contact stiffness was decreased to match the magnitude of the second peak in the experimental force response as closely as possible without causing large penetrations between the pin and guide. (Penetration was considered ‘large’ when it gave rise to behaviors that would be impossible in the real world. For example, excessive penetration enabled the pin to move through the escape paths without rotating, which is a physical impossibility (Figure 3-2)). The pin-guide damping coefficient was then gradually increased in increments of 1 N·ms/mm, and fine-tuned in increments 0.1 N·ms/mm, to eliminate unrealistic spikes from the force response. This process was reiterated until the model response converged. The remaining contacts in the model were assigned the Adams default contact parameters since these contacts did not affect the initial impact forces. All contacts were assigned default friction coefficients (static coefficient = 0.3, dynamic coefficient = 0.1).    52   Figure 3-2: When the contact stiffness is low, the deployment pin is able to penetrate the guide and move through an escape path without rotating. Physically, this is not possible.  The model was validated by comparing the model kinematics to experimental measurements in three different impact conditions (0⁰, 15⁰, and -15⁰). The deployment direction was also evaluated to ensure the simulated deployment mechanism deployed in the same manner as the physical mechanism. Finally, a dynamic analysis was conducted to determine the forces that prescribe the deployment mode.  3.3 Results 3.3.1 Tuning The model response was sensitive to changes in the contact stiffness and damping parameters (Figure 3-3). In some cases, decreasing the contact stiffness by an order of magnitude decreased the impact force by more 50%.    53   Figure 3-3: Sensitivity of helmet-ground impact force to changes in contact parameters. The contact force is sensitive to changes in the contact stiffness (left) and damping (right) parameters.  However, reasonable agreement with the experimental force response was achieved with tuning (Figure 3-4). The model exhibited two impact peaks, as was observed in the experimental response, and the force magnitude of the first impact peak fell within one standard deviation of the average peak experimental value. However, the second force peak exceeded the experimental response by 55%. When the pin-guide contact stiffness was decreased to match the magnitude of the experimental response large penetrations between the pin and guide were observed, resulting in unrealistic behavior. Forces also developed more rapidly in the model, which resulted in a time lag between the model and experimental responses. ‘Reasonable’ was therefore a subjective designation that was intended to summarize the good quantitative agreement (<10% difference) 0⁰ Impact 0⁰ Impact   54 observed in the first force peak, the poor quantitative agreement (≥30% difference) observed in the second force peak, and the overall agreement in the shape of the force response.   Figure 3-4: Agreement between representative experimental response and model force response in 0⁰ impact after tuning. The oscillations observed in the experimental response are due to ringing of the aluminum helmet shell. The model response was aligned with the experimental data according to the time of the peak impact force.  Similar agreement was observed when the tuned contact parameters from the 0° impact condition were applied to the 15⁰ and -15⁰ impact conditions (Figure 3-5). Table 3-2 compares the magnitudes of the model and experimental impact peaks in each impact condition. In every condition the model showed good agreement with experimental results for one impact peak (<10% difference), and poor agreement (≥30% difference) for the other.  Peak impact force   55  Figure 3-5: Comparison of experimental and model force response in 15⁰ and -15⁰ impact conditions  Table 3-2: Force magnitudes of the first and second impact peaks in the experimental drop tests and tuned model. SD = standard deviation  Experiment Model  0⁰ 15⁰ -15⁰ 0⁰ 15⁰ -15⁰ 1st Peak  [N]  (SD) % Difference 12,829 (776) 5 9,238 (570) 30 12,170 (512) 10 12,220   13,150   13420   2nd Peak  [N]  (SD) % Difference 5,860 (454) 55 6,734 (340) 9 5,263 (267) 71 9,096   7,363   8,978    The model kinematics were less affected by changes in the contact parameters than the force response. When the contact stiffness was altered by several orders of magnitude, the peak   56 angular position fluctuated by less than 1% (Figure 3-6). Similar patterns were observed in the angular velocity measurements. Therefore, kinematics were believed to be a more robust predictor of the real-world response.   Figure 3-6: Relationship between angular position and changes in contact stiffness. The kinematic response remained constant despite altering the head-ground contact stiffness by several orders of magnitude. Angular position is shows here, but the angular velocity measurements showed similar results.  3.3.2 Validation The model kinematics were compared to experimental data to validate the model response.  The model behavior was also compared to high-speed videos of the experimental drops to observe the 0⁰ Impact   57 motion of the head and the deployment mode in each impact condition. The validation results are presented in the following sections.   3.3.2.1 Kinematics The model and experimental kinematics, indicated by the angular position and angular velocity of the outer helmet shell, are shown in Figure 3-7. The data are aligned according to the time that the deployment pin contacted the top of the deployment guide (i.e. the time of the second impact peak in the force response). This event was of primary interest, as it represents the time at which loads are transmitted to the head through the selector mechanism, and therefore is the time at which deployment is initiated. The model showed good agreement with experimental results. In the 0⁰ and 15⁰ impact conditions the model response fell close to, or within, the experimental corridors. The agreement was not as strong in the -15⁰ impact condition, as the magnitude of the response predicted by the model was smaller than that measured experimentally. However, the trends in the motion appear similar.   58   Figure 3-7: Validation of model kinematics according to the helmet angular position and angular velocity  The model response falls close to or within the impact corridors in the 0⁰ (left) and 15⁰ (middle) impact conditions. In the -15⁰ (right) condition, similar patterns of motion are observed, although the magnitude of the model response is smaller than that of the experimental response.  Small changes in the initial position of the head and helmet in the model improved the agreement between the model and experiment in the -15⁰ condition. Figure 3-8 shows the improved response when the head was flexed forward by 4⁰, prior to impact.     59  Figure 3-8: Forward flexion of the head by 4⁰ improves the model agreement with experimental results  Similar improvements were achieved by moving the helmet posteriorly on the head by 5 mm prior to impact.   3.3.2.2 High-Speed Video A qualitative comparison of the model with high-speed videos of the experiments showed good agreement in all three of the impact conditions (Figure 3-9,Figure 3-10, and Figure 3-11). In each case, the helmet rebounded off the ground when the pin contacted the top of the deployment guide. While airborne, the head and inner shell swung into either flexion or extension, causing   60 rotation of the deployment pins. When the helmet contacted the ground for the second time, the deployment pins slid partway into the corresponding flexion or extension escape path. This comparison with high-speed videos was necessary because head motion was not measured quantitatively in the experimental tests, but plays a critical role in the deployment of the selector mechanism.   The model and experiment showed good qualitative agreement in the -15⁰ impact condition, despite the quantitative differences observed in the outer helmet shell kinematics. The differences between the model and experiment appeared to result from engagement of the pin with the deployment guide. In the experiment, contact between the pin and guide causes the outer helmet shell to roll down the impact platform before it rebounds and becomes airborne. This sets the helmet in motion, causing large rotations to follow. However, rolling was not observed before rebound in the model. Nevertheless, this did not appear to affect the gross behavior of the head and deployment mechanism, and in each impact condition the deployment mode observed in the model matched the experimental deployment mode (Table 3-3). This was a critical aspect of the validation, given that the purpose of developing the model was to determine the factors that influence the deployment mode.     61  Figure 3-9: Frame-by-frame comparison of model and experiment in a 0⁰ impact   Figure 3-10: Comparison of model and experiment in a 15⁰ impact   62  Figure 3-11: Frame-by-frame comparison of model and experiment in a -15⁰ impact. Rolling of the outer shell occurs prior to rebound in the experiment, but does not occur in the model.  Table 3-3: Comparison of deployment modes observed in model and experiment  Deployment Mode  Model Experiment 0⁰ Impact Extension Extension 15⁰ Impact Extension Extension -15⁰ Impact Flexion Flexion   3.3.3 Dynamic Analysis of Deployment Mechanism A dynamic analysis was conducted to determine the forces that govern the deployment direction. This analysis revealed that the forces acting at the deployment pins and the occipital condyles   63 produce a moment about the center of gravity of the head when the deployment pins contact the top of the guide (Figure 3-12). This moment produces rotation of the head, which causes the deployment pins to rotate into the flexion or extension deployment path depending on direction of the moment. In the 0⁰ and 15⁰ conditions, the moment was positive and resulted in extension deployment. In the -15⁰ condition, the moment was predominantly negative and resulted in flexion deployment. Figure 3-13 shows the moments acting on the head in each impact condition as a function of time. Black arrows indicate the time at which the deployment pins contact the top of the deployment guide. Prior to this time, the inner and outer shells cannot move independently, and the moment acting on the head does not influence the deployment direction. Contrary to predictions from the experimental tests, the forces at the occipital condyles tend to generate a negative, flexion moment. This is due to a large, posteriorly-directed reaction force at the revolute joint. Conversely, the forces at the deployment pins tend to cause a positive, extension moment due to a large, inferiorly-directed force acting posterior to the head center of gravity. Thus, these moments oppose one another, and appear to be in precarious balance.   Figure 3-12: Forces are transmitted to the head through the occipital condyles and the deployment pins. These forces oppose one another and their relative magnitudes determine the net moment on the head.   64   Figure 3-13: The moment acting on the head (red) is the resultant of the moments due to the forces acting at the deployment pins (green) and the occipital condyles (blue). In the 0⁰ and 15⁰ impact conditions, a positive moment resulted in extension deployment. In the 15⁰ impact condition, a predominantly negative moment resulted in flexion deployment. The shaded area indicates the time during which the moment influenced the deployment mode.  The magnitude of the resultant moment is indicative of the deployment time. The largest moment was observed in the 15⁰ condition, and high-speed videos, presented in Chapter 1, revealed that head rotation occurred fastest in this configuration. In the -15⁰ condition, the effective moment did not develop immediately, as the moment was initially positive before eventually becoming negative and generating flexion deployment. Results from Chapter 2 revealed that the deployment speed was slowest in this configuration.      65 3.4 Discussion Reasonable agreement was achieved when the model force response was tuned to match the experimental response. The experimental and model force responses both showed two force peaks, and in each impact condition good agreement (<10% difference) was observed for one of those two peaks. However, the quality of the agreement was limited by the large penetrations observed with low contact stiffness.   Changes in the force response due to altering the contact parameters did not dramatically alter the kinematic behavior of the model. Decreasing the contact stiffness by several orders of magnitude altered the peak impact forces by as much as 80%, while the peak angular position was altered by less than 1%. The model was therefore validated using kinematic measures, and the model response fell close to, or within, the experimental corridors in most cases. The virtual helmet deployed in the same direction as the physical helmet in every impact scenario, and the gross behaviors of the head and helmet were the same in both the physical and virtual apparatus. Based on this validation, the model was believed to be adequate for use in a dynamic analysis of the deployment mechanism.   The dynamic analysis revealed that the deployment direction is governed by the resultant moment on the head, due to the forces at the occipital condyles and deployment pins. Additional investigations in Adams multibody dynamics software revealed that adjusting the location of the deployment mechanism, or the position of the center of gravity of the head, could alter this moment and the resulting deployment mode. This supports the conclusion that the moment is the predominant factor influencing deployment.    66  The forces at the occipital condyles arise in part from the following torso mass, and in part as a reaction to the head impact forces. These forces are challenging to control and predict, and will be influenced by the relative posture of the head, neck and torso. Postural changes that alter the position of the head center of gravity relative to the occipital condyles were shown to affect the deployment mode, and therefore the passive selector mechanism cannot be considered reliable. The deployment mechanism must be redesigned such that the deployment direction is only dependent on the location of an impact to the head.  The accuracy of this model is limited by the accuracy of the contact forces, which differed from the experimental values by as much as 71% in the worst case. It is uncertain how these discrepancies may have affected the conclusions drawn from the model. However, differences between the model and experimental results are expected, if not inevitable, given the numerous limitations associated with the Adams spring-damper contact model. The model applies a single force vector at the center of a contact. In comparison, real world contact and friction forces are distributed throughout the contact area. Furthermore, the contact algorithm is adapted from Hertz contact theory, which assumes contact between rigid, elastic bodies with simple, solid geometries. Therefore, the contact algorithm falls far short of capturing effects due to deformation of the outer helmet shell, or the layer of padding covering the impact platform. Due to these simplifications, more sophisticated modeling techniques are likely necessary if greater accuracy is desired. Conversely, tuning was performed manually and was therefore limited by the subjective nature of the tuning. A design of experiments approach may have yielded better results, and this is recommended to improve the robustness of the model in the future.   67  The contact parameters that were tuned for the horizontal impact condition were applied to the inclined conditions. This further limits the accuracy of the impact forces predicted in the model, since altering the contact geometry will alter the force response. However, the changes in contact geometry were minimal, given that the helmet shell is approximately spherical. Therefore, application of the contact parameters across impact conditions was justified in order to achieve a single model that could be applied to multiple impact scenarios. The effects of this assumption were not assessed, and should be investigated in the future.  The positions of the head and helmet were not quantified in the experiments, and it is therefore likely that the simulated drops did not precisely replicate the experimental condition. Kinematic agreement between the model and experiment was improved by adjusting the pre-impact alignment of the head and helmet in the model; however, without additional testing we cannot know with certainty how the pre-impact alignment contributed to the discrepancies observed. Furthermore, images from high-speed videos showed rotation of the carriage mass in the -15° impact condition, due to compliance in the linear bearings (Figure 2-7). This motion is not accounted for in the model, since the bearings were modeled as a single ideal translational joint. This, too, may have contributed to the differences observed in the kinematic measurements. The rotation of the carriage mass was not as pronounced in the other conditions, which may explain why the -15° condition shows greater discrepancies than the other two conditions.  Despite these limitations, the model behavior appeared to be reasonable. In each impact, the patterns of motion observed in the model were comparable to the experimental motions, and   68 distinct events occurring experimentally, such as head rebound and helmet-chin contact, could be identified. The model was constructed with the primary goal of investigating the behavior of the deployment mechanism, and the observed behavior was consistent with experimental findings. The model should not be used to predict the absolute forces that may arise in an experimental impact. However, the relative actions of the forces within the model appear to produce a response that is analogous to the real world situation, and the model proved useful in a dynamic analysis. This, combined with the results of the kinematic validation and the consistency in deployment behavior observed in the model and experiments, suggests that the model is sufficient for use as a design tool. However, caution should be used when interpreting the results of the model. The user should fully understand the model limitations, and be aware of the limited applicability of the model when used for impact conditions for which it is not validated. If the model is used for design purposes, an iterative design process is recommended, in which the user moves between the virtual and physical prototyping spaces to periodically confirm the conclusions drawn from the model.      69 Chapter 4: Conceptual Design of an Improved Selector Mechanism 4.1 Introduction A passive PNT selector mechanism was evaluated in Chapters 2 and 3 and found to be incapable of inducing the desired head motion in all impact scenarios. Therefore, there exists a need for an improved PNT selector mechanism that deploys appropriately in all impact conditions, according to location of an impact to the head. This chapter uses the analyses presented in Chapters 2 and 3 to propose a conceptual design for a new PNT selector mechanism. The design process is presented in detail in order to inform future developers of concepts that have been rejected or not developed entirely. A final design is presented, and its functionality is demonstrated using Adams multibody dynamics software. The design is intended for physical prototyping in the future.  4.1.1 Design Problems The passive selector mechanism, subject of Chapters 2 and 3 of this thesis, was intended to induce extension in anterior impacts, and flexion in posterior and vertex impacts. Physical testing revealed that the mechanism is capable of inducing the correct deployment mode in +15⁰ and -15⁰ impacts. However, in the 0⁰ impact condition the helmet deployed in extension, rather than the desired flexion mode.  Furthermore, deployment did not occur smoothly in any impact condition. In each case, head rebound occurred before the deployment pin could move through either escape path. The peak loads at the upper load cell, indicative of the peak loads on the cervical spine, were coincident with the time of pin-guide contact. Appreciable head motion occurred only after those peak loads developed and the head rebounded off the platform. As a   70 result, it is unlikely that the device would reduce the loads on the cervical spine in a head-first impact.  A dynamic analysis was performed using a computational model of the PNT helmet. The analysis showed that the deployment direction is governed by the net moment acting on the head. This moment arises from the forces applied to the head through the occipital condyles and the deployment pins, which would be affected by head-neck-torso alignment. Therefore, changes in pre-impact alignment could alter the deployment mode. Predicting the resultant moment on the head for any given impact condition is non-trivial, and therefore this method of initiating deployment is not reliable. Ultimately, the deployment mode should be dependent only on the impact direction.   Furthermore, the forces at the deployment pins and occipital condyles oppose one another. If the moment arising from the deployment pins exceeds that of the occipital condyles, extension deployment occurs. Similarly, if the moment arising from the occipital condyles is greater than that of the deployment pins, flexion deployment occurs. As the magnitude of the net moment increases, the deployment speed increases as well. In the 15⁰ and 0⁰ conditions, the extension moment from the deployment pins dominates. However, as the angle of the impact platform becomes increasingly negative, the flexion moment from the occipital condyles gradually overcomes the moment from the deployment pins. Naturally, there must be an impact configuration in which these are perfectly balanced. As this configuration is approached the net moment will decrease, it will become harder to produce either flexion or extension escape, and head rebound will become more pronounced. When the net moment is exactly zero neither   71 escape can occur. According to the multibody model, these conditions arise when the impact platform is inclined at -13⁰ (Figure 4-1).    Figure 4-1: When the impact platform is inclined at -13⁰ the moments arising from the forces at the occipital condyles and deployment pins are balanced. When the net moment is zero, the deployment pin lacks the rotational force necessary to overcome the bifurcation in the guide. Therefore the head will rebound without moving through either escape path.   4.1.2 Design Objective There exists a need for an improved deployment mechanism that generates appropriate head motion in every impact condition. The new mechanism must meet the following criteria: 1. The mechanism must induce head extension in anterior impacts, and head flexion in posterior and vertex impacts 2. The deployment mode must be affected only by the location of the impact vector on the head 3. The deployment pin must move smoothly through the intended escape path in order to dissipate energy prior to head rebound   72  It should be noted that these criteria reflect our present knowledge of cervical spine injury mechanisms, and may be subject to change. Specifically, future investigations may reveal that, in addition to the location of the impact vector on the head, additional parameters should be considered when determining the deployment mode. However, we are aware of no such parameters at this time.  4.2 Design Methodology: Parameter Analysis The ‘parameter analysis’ design methodology was used to generate a conceptual design for a new selector mechanism. The methodology describes an approach for generating conceptual designs after a need has been identified and the design requirements have been defined [55].   The preliminary phase of the conceptual design process is termed ‘technology identification’. Here, the designer identifies several possible means of addressing the design problem, which provides numerous starting points for the design process. Technology identification is followed by the parameter analysis process, which consists of three stages: parameter identification, creative synthesis, and evaluation (Figure 4-2).    73  Figure 4-2: Conceptual design process using parameter analysis. PI = Parameter Identification, CS = Creative Synthesis, E = Evaluation. Reproduced from Kroll et al. [55], with permission from Cambridge University Press.  During the parameter identification stage, a critical design parameter is identified. This parameter is not a quantified metric. Rather, it is a conceptual issue with the existing design that inhibits its function. By identifying the single most important design parameter the problem is simplified and the designer is relieved of trying to address all the design requirements at once. The critical parameter changes as the design evolves. Those parameters deemed ‘less important’ in the early design stages, may become critical parameters in later stages.   Conceptual design configurations are developed in the creative synthesis stage. Here, a potential solution is proposed and then tested using theory, sketches, and ‘back-of-the-envelope’ calculations as representations of a physical system. In the present work, the multibody model developed in Chapter 3 was used to test conceptual designs as they were developed.   Finally, the conceptual design is assessed in the evaluation phase. The design’s performance is compared to the design requirements, weaknesses are identified, and a new critical parameter is   74 defined. Thus, the parameter analysis cycle begins anew. This process is repeated until all the design requirements have been met.   4.3 Design Process 4.3.1 Need Identification A need statement was developed to direct the design process. To avoid unnecessarily constraining the design, a need statement should be solution independent. The PNT helmet is intended to address a need to reduce the loads on the cervical spine in head-first impacts. However, a range of injury prevention devices might satisfy this need, and the need is therefore too broad to direct the continued development of the PNT selector mechanism, specifically. The need statement was, therefore, defined as follows:  There exists a need to develop a helmet-mounted selector mechanism that induces head motion (flexion or extension) according to the location of an impact to the head (posterior, vertex, or anterior), in order to reduce the loads on the cervical spine in head-first impacts.  This need statement is not solution-independent. It prescribes the method of reducing neck loads (head motion), as well as the technology used to accomplish this (helmet and selector mechanism), and therefore constrains the potential design solutions. This was intentional, and intended to limit the scope of solutions to those related to the continued development of the existing PNT helmet. It should be noted that the term ‘selector mechanism’ is not intended to refer to the existing selector mechanism configuration, which consists of a deployment pin and   75 guide. Rather, it refers to any internal mechanism that is capable of inducing head motion on impact, and future designs are not constrained to modifying the existing design.   4.3.2 Design Requirements The design requirements were intended for the evaluation of preliminary prototypes, and most were evaluated according to simple pass/fail acceptance criteria (Table 4-1). The ‘size’ criterion was not defined quantitatively, because the acceptable dimensions will change depending on the location of the mechanism within the helmet. The head motion criterion states that the head must rotate at least 8⁰ prior to head rebound for a design to be acceptable. This threshold was chosen based on the 8⁰ of head motion observed in computer simulations of the anterior impact condition. Deployment occurred most smoothly in this condition, and therefore the induced head motion was selected as the baseline performance criteria that future mechanisms must meet.  Table 4-1: Design requirements for a new PNT selector mechanism Category Requirement Acceptance Criteria Performance Must induce the correct deployment mode in every impact.  Must initiate head motion prior to, or without, head rebound Yes    >8⁰ of head rotation prior to head rebound Size Must fit within the existing aluminum helmet prototype Yes Other Must not include electrical components Yes    76 The passive selector mechanism failed to meet the performance requirements. According to the computational model presented in Chapter 3, the mechanism induced only 3.3⁰ and 4.0⁰ of rotation in the posterior and vertex impacts, respectively; and a configuration was identified in which 0⁰ of rotation occurred. Furthermore, the mechanism did not deploy in the desired direction in the horizontal impact condition.  4.3.3 Technology Identification Based on the need defined in section 4.3.1, two categories of potential technologies were defined: technologies that modify the existing selector mechanism, and technologies that are novel. For efficiency, design efforts were focused on adapting the existing selector mechanism. The aluminum helmet prototype has a modular design, and can easily accommodate modified versions of the existing selector mechanism. Additionally, the passive selector mechanism offers a simple platform for testing new design concepts. An effective selector mechanism will have three primary functions: sense the location of an impact on the head, select the appropriate deployment mode based on feedback from the sensor, and move the head into flexion or extension. The passive selector mechanism primarily addressed the ‘move’ function, which is the simplest of the three functions to achieve. With minor modifications, new ‘sensing’ and ‘selecting’ elements can be tested with minimal design effort. Once developed, these elements can be adapted if an alternate method of generating motion is desired. For example, the designer may wish to eliminate the redundancy that exists due to having bilateral guide mechanisms.    77 4.3.4 Parameter Identification I Improvements must be made in order to (1) sense the location of an impact, and (2) redirect the head out of an axial alignment. There are numerous mechanical means for identifying the direction or location of an impact. However, smoothly redirecting the head is less trivial and is essential for the PNT mechanism to be effective. ‘Induce head motion in axial impacts’ was therefore designated as the critical parameter.  4.3.5 Critical Synthesis I In certain loading conditions, the net moment on the head is zero and the deployment pin is unable to rotate into either of the escape paths (Figure 4-1). To overcome this, the concept of a default deployment mode was devised. Prior to impact, the helmet would be configured to deploy in flexion, the ‘default’ mode. With only one escape path available, flexion deployment would occur smoothly even under pure axial loading conditions. If an anterior impact was sensed, an active configuration change in the selector mechanism would be induced. This change would eliminate the flexion deployment path, and force the inner shell and head to move into extension. Introducing a default deployment mode has the added advantage of simplifying the design problem. Initially, it seemed necessary to independently identify posterior, vertex, and anterior impact conditions. However, with a default deployment mode, it is only necessary to sense anterior impact conditions. Table 4-2 shows some of the design configurations that were conceived to create a selector mechanism with a default deployment mode.    78  Table 4-2: Possible design configurations for a selector mechanism with a default deployment mode Concept Sketch Description Advantages  Disadvantages Movable Pin  By default, the deployment pin is angled anteriorly, forcing the pin through the flexion path. If an anterior impact is sensed, the pin switches to a posterior-facing configuration, to cause extension deployment. i. Only one moving piece ii. Simple to prototype i. Pin is small and prone to breaking  ii.  Impact signal must be transmitted from outer to inner shell iii. Pin may still rebound off the top of the bifurcation Movable Guide   By default, a switch at the apex of the guide is angled posteriorly, forcing flexion deployment. If an anterior impact is sensed the tip is switched to an anterior-facing position, forcing extension deployment i. Impact signal sensed on outer shell can be directly transmitted to guide ii. Simple to prototype  iii. Only one moving piece  i. Switch is small and prone to breaking ii. Switch may not move fast enough to change deployment path in an impact  Railway Switch  By default, the inner shell is guided through a flexion escape path on a set of rails. In an anterior impact, a railway switch reconfigures the tracks for extension. i. Switch does not need to move far, which makes changing configurations fast i. Miniature rails & wheels would be hard to manufacture ii. Several components    79 4.3.6 Evaluation I If successfully implemented, a selector mechanism with a default deployment mode could induce smooth head motion by constraining the head and inner shell to a single, continuous path. The mechanism would then behave similar to previous prototypes where the deployment direction was configured prior to impact (Figure 1-9). With a continuous deployment path, head rebound resulting from the pin contacting the top of the guide would be reduced and head motion would occur throughout the development of neck loads. The default deployment mode would also force the helmet to deploy in flexion during impacts to the vertex of the head, despite any physical tendency for the head to move in extension. With only one deployment path available at a time, the potential for deployment errors would be greatly diminished.  Therefore, this design concept has the potential to satisfy all of the design requirements; however, a method of automatically triggering the necessary configuration change must be developed.   The advantages and disadvantages of specific design configurations are shown in Table 4-2. Of these designs, the railway switch is the hardest to implement. It requires multiple moving parts, and would be hard to manufacture within the size constraint of the aluminum helmet shell. The movable deployment pin would also be challenging to implement, because the impact signal sensed on the outer shell would need to be mechanically transmitted to the inner shell, without the two shells being rigidly coupled. Therefore, the movable guide was selected as the most feasible design concept. To be effectively implemented, a sensor must be developed to detect anterior impacts and trigger the switch to change positions.     80 4.3.7 Parameter Identification II, Creative Synthesis II and Evaluation II A new critical parameter was identified for the movable guide design, according to the weaknesses identified in Evaluation I. This parameter was ‘sense the occurrence of an anterior impact.’ With flexion as the default deployment mode, only anterior impacts must be detected. However, these impacts must be detected in time for the switch to change configurations before the deployment pin makes contact with it.  Three types of sensors were identified as potential solutions to the design problem: inertial sensors, deformation sensors and position sensors. Examples of each are compared in Table 4-3. A spring sensor that detects relative motion between the inner and outer helmet shells was believed to be the most feasible, for reasons outlined in the table. In an impact, the reaction force exerted on the outer helmet shell will act in different directions depending on whether the impact is anterior, posterior or perpendicular to the vertex of the head (Figure 4-3). Anterior impacts will produce a reaction force with a posteriorly directed component, while posterior and vertex impacts will not. A properly positioned spring would therefore experience compression in anterior impacts, but not in other impact conditions.   Figure 4-3: Anterior impacts can be identified by posteriorly directed reaction force (shown in red)   81  Figure 4-4 shows one possible configuration for a spring sensor. A spring is placed between the outer helmet shell and the deployment guide, allowing a small amount of relative motion between the two components. The deployment guide is coupled to the inner helmet shell through contact with the deployment pin. In an anterior impact, the outer shell will experience a posteriorly directed force, and therefore will accelerate posteriorly. The guide and pin are isolated from the impact force by the spring, and thus do not accelerate at the same rate as the outer helmet shell. The result is posterior motion of the outer helmet shell relative to the inner shell and deployment guide, which generates compression of the spring. Contact between the posterior edge of the guide and helmet shell prevents the opposite motion from occurring in posterior impacts. Therefore, the spring becomes compressed in anterior impacts, but is unchanged in both posterior and vertex impacts. This concept was demonstrated in Adams.   Figure 4-4: Possible configuration of spring sensor. Spring is placed between outer shell and guide. Guide is coupled to inner shell through contact with the deployment pin. A posteriorly directed force on the outer shell will generate relative motion between the outer shell and guide, and cause compression of the spring.   82  Once a spring sensor was selected, it was necessary to harness the spring deformation to trigger a conformational change in the selector mechanism. For brevity, the detailed design process for the trigger is omitted from this document. The final design is described in the following section.   Table 4-3: Potential sensors for use in an active guide mechanism  Description Advantages Disadvantages Pendulum  (inertial sensor) A pendulum is free to swing within the outer helmet shell. The pendulum will swing in the opposite direction of a force on the outer helmet shell.   If the head is not oriented vertically prior to impact, gravity will alter the pendulum position, rendering it ineffective Spring  (deformation sensor) Springs can be used to detect relative motion between the inner and outer shell. Depending on the direction of motion, the direction of an impact can be deduced.  Can pretension the spring so motion due to gravity or random head motion does not trigger deployment Horizontal forces in near-axial impacts are small and therefore challenging to detect Buttons  (position sensor) A button is located anteriorly on the helmet shell. Depression of the button signals anterior impact. i. Easy to design ii. Easy to differentiate impact from other non-impact head accelerations   i. Can be accidentally depressed in non-injurious impacts ii. Signal is sensed in midsagittal plane and must be transmitted to laterally positioned guides     83 4.4 Final Conceptual Design The final selector mechanism design is shown in Figure 4-5 (see Appendix B  for machine drawings). The mechanism has four components: a guide, a switch, a locking cylinder, and a sliding block. The sliding block is rigidly fixed to the outer helmet shell. It interfaces with a horizontal slot in the guide, which houses a compression spring. The locking cylinder rests above the sliding block. It interfaces with a semi-cylindrical groove in the switch. The switch is cocked with a torsional spring, but cannot rotate when the locking cylinder is in place. Figure 4-5 represents the resting configuration of the right deployment mechanism. In this configuration, the mechanism is primed for flexion deployment.   Figure 4-5: Final selector mechanism design. The mechanism shown in the figure is the right deployment mechanism. A = anterior, P = posterior  The guide is attached to the outer helmet shell by slotted holes. These slots allow relative translation between the outer shell and guide. At rest, the compression spring forces the guide to A P P A   84 the end of its travel on the sliding block. When an anterior impact occurs, the spring is compressed as the outer helmet shell and sliding block move posteriorly relative to the guide. This motion causes the locking cylinder to fall out of the groove in the switch (Figure 4-6). The switch is then released, and the torsion spring causes rotation of the switch into the extension configuration.    Figure 4-6: Action of the new selector mechanism in anterior impact. The sliding block is rigidly fixed to the outer shell, which moves posteriorly with respect to the guide in an anterior impact (left). The locking pin then drops out of the slot in the switch (middle), and the switch flips into the extension configuration (right). The mechanism shown in the figure is the right deployment mechanism. A = anterior, P = posterior  4.5 Analysis Proof of concept was achieved in Adams using a first-generation design. This design is not identical to the design described above; however, it functions according to the same principle. Therefore, the updated design was not simulated.   A P   85 The model described in Chapter 3 was adapted to incorporate the new selector mechanism. It was necessary to alter the pin-guide contact parameters because the contact geometry was altered. As a result, there is some uncertainty in the model behavior.   Figure 4-7 shows the behavior of the proposed mechanism in each impact condition. In the 0⁰ and -15⁰ impacts the switch remained in the default position, resulting in flexion deployment. In the 15⁰ condition the mechanism correctly identified an anterior impact, and was released to generate extension deployment. The switch motion during anterior impact is highlighted in Figure 4-7. The helmet deployed appropriately in every impact condition, satisfying the first design objective. Additionally, the mechanism has only mechanical components, as stipulated by the design requirements.     86      Figure 4-7: Deployment of the PNT helmet with an active selector mechanism. In a 15⁰ impact (top) the mechanism responded to an anterior impact condition by flipping the switch into the extension deployment mode. In the 0⁰ (middle) and -15⁰ (bottom) conditions, the switch remained in the default flexion configuration.   87 The deployment pin was shortened from its original length in order to allow time for the switch to change configurations. By decreasing the pin length by 8 mm, an additional 4 ms of switch time was gained. The angle between the vertical slot and the escape paths was also decreased from 60⁰ to 16⁰. This enabled the pin to move more smoothly into the escape paths, and reduced head rotation to prevent contact between the face and outer helmet shell. Furthermore, the deployment mechanism was placed in line with the occipital condyles. In its original position, the deployment mechanism was located posterior to the occipital condyles, which tended to produce an extension moment on the head. When the new mechanism was simulated in this position, the extension moment on the head opposed the flexion action of the mechanism. This caused the pin to jam in the guide, rather than slide smoothly through the flexion escape path. Moving the deployment mechanism anteriorly decreased the extension moment on the head, and allowed the deployment pin to move smoothly through the escape path in each impact.  The deployment pin moved smoothly through the guide, without substantial head rebound. This suggests that the loads on the spine would be reduced in an impact; however, the current apparatus is too simple to confirm this. Physical testing with a surrogate cervical spine is therefore necessary to understand how the existing mechanism alters the loads on the spine. It should be noted that some rebound did occur in each impact condition. In the anterior, vertex and posterior impact conditions 7.5⁰, 7.8⁰, and 7.3⁰ of head rotation were observed prior to rebound. Thus, the new mechanism shows improvement over the initial, passive selector, but does not satisfy the design criteria, which specifies a minimum of 8⁰ of head rotation. However, it is likely that small alterations in the guide geometry could improve the motion of the deployment pin and enable this mechanism to satisfy the design criteria.   88  The simulated compression spring had a spring stiffness of 10 N/mm. This stiffness is possible to achieve with standard springs that fit within the dimensions of the current guide mechanism, satisfying the size criterion. However, the simulated spring was specified for the loads that developed in the model. While this provides an initial estimate of the required spring stiffness, experimental testing is required to determine the spring stiffness necessary to achieve a functional physical prototype. Obtaining a torsional spring with enough strength to flip the deployment switch prior to pin contact is the greatest foreseeable challenge. Assuming the pin travels 1.5 cm at 3 m/s prior to contact with the switch, the switch would have to deploy within 5 ms of contact between the outer shell and ground. Physical prototyping is necessary to determine the maximum deployment speed that can be achieved with a torsional spring. It may be necessary to slow the travel of the deployment pin prior to contact with the switch, or otherwise increase the time allowed for the switch to move.  4.6 Discussion Analysis of the previous, passive deployment mechanism led to the conclusion that a ‘default’ deployment mode is necessary to enable smooth deployment. This concept was a critical leap in the development of the PNT deployment mechanism, and virtual prototyping in Adams showed that the proposed mechanism has the potential to meet all the design requirements outlined in Table 4-1. However, physical prototyping is required to ensure that the current design can be manufactured within the prescribed size constraints, and is capable of inducing sufficient head motion to reduce the loads on the cervical spine.     89 Despite the encouraging results from the computational model, there are challenges associated with the continued development of this mechanism. Foremost, the helmet must be able to detect anterior impacts even when the impact angle is small. As the angle of the impact surface approaches 0⁰, the forces directed posteriorly on the head will approach 0 N, and it will become more difficult to trigger the switch. The simulated mechanism requires a spring force of only 30 N to release the switch. However, investigations of head accelerations in everyday and vigorous activities have reported head loads as large as 100-300 N [56]. Thus, the switch could easily be activated in non-impact loading conditions. Continued development of this mechanism is encouraged; however, the design presented in this chapter is only one of many possible design configurations. In light of the associated challenges, other design configurations should also be investigated.  Additional development is also necessary to ensure the mechanism does not deploy in oblique impacts, or impacts below the injury threshold of the cervical spine. Future designs must include a limiter that enables deployment only during axial impacts that exceed a force threshold. This is non-trivial given that head loads are not strongly correlated with the loads on the cervical spine [28]. Furthermore, having bilateral deployment mechanisms is redundant, and it is probable that asymmetrical loading would cause the mechanisms to deploy in different directions during an impact. It is strongly encouraged that future designs seek to limit this redundancy.   The final PNT mechanism will be required to meet more design requirements than those presented in Table 4-1. The force threshold required to initiate deployment must be defined; and the optimum head trajectory during deployment should be determined. The size and weight of   90 the helmet should also be close to those of existing helmets. Acceptance criteria for these requirements must be defined quantitatively in the future, but were considered irrelevant for the development of a preliminary prototype. The objective of the present design project was to develop a mechanism that meets the critical functional requirements. The engineering adage, ‘first make it work, then make it better’ was the guiding philosophy in this process.      91 Chapter 5: Conclusion 5.1 Summary  A passive PNT selector mechanism was evaluated using physical drop tower testing and computational modeling. The drop tower tests were used to observe the deployment behavior of the mechanism in different impact scenarios. The computational model was then used to determine how impact forces were transmitted to the selector mechanism in order to initiate flexion or extension deployment. The mechanism was intended to induce extension in impacts anterior to the vertex of the head, and flexion in impacts perpendicular or posterior to the vertex of the head. The functionality of the mechanism was assessed according to its ability to induce the appropriate deployment mode in different impact conditions. An improved PNT selector mechanism was then designed based on the findings of the physical and virtual analyses.  Prior to this work, a passive selector mechanism with two escape paths had never been tested. All previous prototypes were configured prior to impact to deploy in the desired manner. It was hypothesized that if two escape paths were available, the impact reaction forces would direct the head through the necessary path without intervention. In other words, posterior impacts would produce anteriorly directed reaction forces that would cause head and neck flexion, and anterior impacts would produce posteriorly directed reaction forces that would cause head and neck extension. It was further hypothesized that the initial position of the head (i.e. Frankfort plane aligned with the horizontal) would cause flexion deployment in vertex impacts. Physical testing revealed that the passive selector mechanism was unable to induce the desired deployment mode in all impact conditions. The mechanism deployed appropriately in 15⁰ and -15⁰ impacts; however, extension deployment occurred in 0⁰ impacts where flexion deployment was preferred.   92 Additionally, the head rebounded off the impact platform before it could rotate into flexion or extension. This was most pronounced in the -15⁰ and 0⁰ conditions, and it is unlikely that the head motion observed prior to rebound was sufficient to reduce the loads on the cervical spine. Deployment occurred more smoothly in the 15⁰ impacts. In this condition, the head rotated eight degrees prior to rebound.   Computational modeling showed that the deployment mode was governed by the resultant moment acting on the head. This moment arose from the forces applied to the head through the occipital condyles and the deployment pins. Therefore, the net moment and subsequent deployment mode were affected by loading from the torso. However, deployment should depend only on the location of an impact to the head. Thus, the passive deployment mechanism was unreliable and was redesigned.   A new selector mechanism design was proposed. By default, the mechanism was configured to deploy in flexion. When an anterior impact was sensed, it underwent a configurational change which produced extension deployment. Specifically, relative motion between the inner and outer helmet shells triggered the release of a switch which changed the deployment mode from flexion to extension. Anterior impacts are mechanically identifiable because a component of the reaction force on the head is directed posteriorly, whereas posterior and vertex impacts do not result in a posterior reaction force. This concept simplified the mechanism requirements because it was necessary to identify only one impact condition, rather than three. Computational modeling demonstrated that the mechanism was capable of inducing the desired head motion in anterior,   93 posterior and vertex impacts. Furthermore, the mechanism produced more head rotation prior to rebound than the passive selector mechanism.  In summary, a new selector mechanism was designed based on evidence that the head and neck will not passively move into flexion or extension according to the impact conditions. The design presented in this thesis is a step forward in the ongoing development of the PNT helmet. Although additional design work is necessary to achieve a fully functional device, this work has brought us closer to the ultimate goal of commercialization and injury prevention. Physical prototyping is recommended in order to thoroughly assess the capabilities of the proposed mechanism.  5.2 Strengths and Limitations 5.2.1 Strengths This is the first time bi-directional PNT deployment mechanisms have been tested and evaluated. This study provided valuable insight into the design requirements necessary to develop an effective PNT selector mechanism.   During experimental testing, the Hybrid III head was rigidly attached to the drop tower carriage by a revolute joint. The apparatus allowed for good repeatability between tests and provided control over the loading conditions on the head and selector mechanism. This control could not be achieved with mechanical neck surrogates or cadaveric cervical spines, which behave chaotically by comparison. As a result, the helmet behavior was isolated from the effects of cervical spine motion and varying torso interaction, in order to precisely understand the   94 functionality of the selector mechanism. Furthermore, the test apparatus was composed only of simple joints and could, therefore, be precisely replicated in a multibody model. It was possible to directly compare the model to experimental results, which minimized uncertainty in the model as well as development time. The apparatus was modeled in multibody dynamics software to minimize simulation times. Therefore more design iterations were conducted than would have been possible with more computationally expensive modeling techniques or physical prototyping.   5.2.2  Limitations Although the drop tower apparatus provided control over the experimental conditions, it also simplified the anatomical constraints on the head. The atlanto-occipital joint was represented as a simple pin joint, and the motion and compliance of the cervical spine were not simulated. Typically, the loading interactions between the torso and head are complex. In drop tests with cadaveric specimens, a bimodal loading response has been observed. Due to compliance of the cervical spine, the peak head impact loads arise before the head is loaded from the torso [20]. It is uncertain how these interactions would alter the loads on the selector mechanism, and the subsequent deployment behavior; however, a small pilot study was conducted to investigate these effects. A mechanical surrogate cervical spine, previously developed in our lab, was used to simulate the cervical spine in head-first impacts [52]. The surrogate was inverted in the drop tower with the Hybrid III head and PNT helmet attached. The deployment behavior of the passive selector mechanism was then observed in anterior impact conditions (5 drops, 15⁰ inline) and posterior impact conditions (3 drops, -15⁰ incline). The deployment behavior was the same as that observed with the simplified constraint, and similar head motions were produced.   95 Therefore, the simplified test apparatus was deemed suitable for initial assessment of a deployment mechanism.  The computational model was verified by comparing the model kinematics to experimental measurements. The quality of this comparison was limited by a number of factors. First, head motion was not tracked in the high-speed videos, and therefore quantitative kinematic comparison was based on the motion of the outer helmet shell.  This comparison is less meaningful, since head rotation was the primary indicator of deployment behavior; analyses of head motion were limited to qualitative comparisons. Second, the orientation of the head and helmet prior to impact were not quantified, and therefore the pre-impact alignment could not be precisely simulated in the model. Furthermore, image distortion was not accounted for during analysis of the high-speed videos. The errors associated with these limitations are unknown.   Computational modeling of the drop tower apparatus was performed using multibody dynamics software. As a result, all contacts were represented as basic spring-damper systems. This is a dramatic simplification of real-world contact behavior, and is likely the primary reason for the discrepancies observed between the experimental and computational force responses. The force response might have been improved by using a design of experiments approach to determine the contact parameters, or by empirically measuring the contact response. The error in the model may have affected the conclusions drawn from the dynamic analysis of the selector mechanism. However, the finding that deployment mode is governed by the moment on the head was consistent with experimental observations, and there has been no evidence to contradict this. Therefore, further development of the contact model was not warranted.   96  Determination of whether the model showed ‘sufficient’ agreement with experimental data was subjective. A quantitative bound could have been set for the minimum acceptable difference between the model and experiment; however, this bound would have been arbitrary. Rather, the model was evaluated based on whether it addressed the need for which it was developed. In this case, the model was used to observe how changes in the impact conditions altered the gross behavior (flexion or extension) of the mechanism. We gained sufficient insight from the model, and therefore, continued development was not warranted. However, in light of the limitations associated with the model, an iterative design process was recommended, where new mechanism designs are ultimately physically prototyped. This limits the repercussions associated with erroneous model results, allowing development time to be focused on design rather than model refinement. Qualitative measures of adequacy are far from ideal; however, sound judgment can be employed when meaningful, quantitative measures are lacking.  5.3 Recommendations and Future Work The functionality of the proposed selector mechanism was demonstrated theoretically using computational modeling. I recommend building a physical prototype to test and compare with the model results, and to identify design challenges that are not immediately apparent from within the virtual environment. I also recommend additional design work to develop a more robust method of detecting anterior impacts. While the concept of a mechanism with a default deployment mode may be a critical step in the PNT selector mechanism design, there are still challenges to be addressed. For example, the current mechanism is at risk of deploying under head accelerations that arise from everyday activities. Additional work must be done to either   97 safeguard the mechanism from inappropriate deployment, or determine another method of detecting anterior impacts. Furthermore, the bilateral guide mechanisms that are currently present should be replaced with a single mechanism, to reduce the possibility of deployment errors. Out-of-plane impacts should be performed to assess the deployment behavior in more complex loading conditions.  In the future, the design requirements for the final PNT selector mechanism must be fully defined. For example, the threshold force at which deployment is initiated must be determined, and the guide geometry should be optimized to minimize neck loads without causing extreme head motions. Additional testing with mechanical neck surrogates or cadaveric specimens will likely be necessary to determine these parameters.  Once a final, or near-final, design is achieved, prototypes of appropriate scale, mass and material should be developed. Tests with mechanical neck surrogates and cadaveric spine specimens must then be conducted to confirm that the helmet can reduce the loads on the cervical spine in head-first impacts without causing unanticipated, secondary injuries.  5.4 Conclusion The PNT helmet is intended to reduce the loads on the cervical spine in head-first impacts by guiding the head out of the path of the following torso. If successfully developed, the PNT helmet may reduce the incidence of cervical spine injuries in sport-related impacts. Inducing head motion to escape torso loading has been previously investigated in automotive applications.   98 To the best of our knowledge, no device to induce head motion has been developed for sporting applications. As such, this work has considerable practical implications.   In this work a passive, bi-directional PNT selector mechanism was evaluated using a combination of experimental testing and computational modeling. The mechanism was found to be inadequate, as it was unable to induce the necessary head motion in all impact conditions. An improved selector mechanism design, with a default deployment mode, was proposed. This design is a critical step towards developing a fully functional selector mechanism, but additional design work is necessary. In the future, the computational model developed here can be used to perform rapid design evaluations; however, caution should be used when interpreting the model results, due to known limitations. Ultimately, all designs intended for long-term development should be physically prototyped and tested.    99 Bibliography [1] B. T. Benevento and M. L. Sipski, “Neurogenic Bladder, Neurogenic Bowel, and Sexual Dysfunction in People With Spinal Cord Injury,” Phys. Ther., vol. 82, no. 6, pp. 601–612, Jun. 2002. [2] R. Levi, C. Hultling, M. S. Nash, and A. Seiger, “The Stockholm spinal cord injury study: 1. Medical problems in a regional SCI population,” Paraplegia, vol. 33, no. 6, pp. 308–315, 1995. [3] The National Spinal Cord Injury Statistical Center, “Spinal Cord Injury (SCI) Facts and Figures at a Glance.” Aug-2014. [4] R. C. Cantu and F. O. Mueller, “Catastrophic Spine Injuries in American Football, 1977–2001:,” Neurosurgery, vol. 53, no. 2, pp. 358–363, Aug. 2003. [5] E. R. Dodwell, B. K. Kwon, B. Hughes, D. Koo, A. Townson, A. Aludino, R. K. Simons, C. G. Fisher, M. F. Dvorak, and V. K. Noonan, “Spinal Column and Spinal Cord Injuries in Mountain Bikers A 13-Year Review,” Am. J. Sports Med., vol. 38, no. 8, pp. 1647–1652, Aug. 2010. [6] F. Tarazi, M. F. S. Dvorak, and P. C. Wing, “Spinal Injuries in Skiers and Snowboarders,” Am. J. Sports Med., vol. 27, no. 2, pp. 177–180, Mar. 1999. [7] A. Gobbi, B. Tuy, and I. Panuncialman, “The incidence of motocross injuries: a 12-year investigation,” Knee Surg. Sports Traumatol. Arthrosc., vol. 12, no. 6, pp. 574–580, Nov. 2004. [8] C. H. Tator and V. E. Edmonds, “National survey of spinal injuries in hockey players.,” Can. Med. Assoc. J., vol. 130, no. 7, pp. 875–880, Apr. 1984. [9] J. S. Torg, J. J. Vegso, M. J. O’Neill, and B. Sennett, “The epidemiologic, pathologic, biomechanical, and cinematographic analysis of football-induced cervical spine trauma,” Am. J. Sports Med., vol. 18, no. 1, pp. 50–57, Jan. 1990. [10] H. H. Bohlman, “Acute fractures and dislocations of the cervical spine. An analysis of three hundred hospitalized patients and review of the literature.,” J. Bone Jt. Surg., vol. 61, no. 8, pp. 1119–1142, 1979. [11] The National Spinal Cord Injury Statistical Center, “Spinal Cord Injury (SCI) Facts and Figures at a Glance.” Aug-2015. [12] R. M. Johnson, D. L. Hart, E. F. Simmons, G. R. Ramsby, and W. O. Southwick, “Cervical orthoses. A study comparing their effectiveness in restricting cervical motion in normal subjects,” J. Bone Jt. Surg., vol. 59, no. 3, pp. 332–339, 1977.   100 [13] M. M. Panjabi, J. Cholewicki, K. Nibu, J. Grauer, L. B. Babat, and J. Dvorak, “Critical load of the human cervical spine: an in vitro experimental study,” Clin. Biomech., vol. 13, no. 1, pp. 11–17, 1998. [14] J. M. Winters and J. D. Peles, “Neck muscle activity and 3-D head kinematics during quasi-static and dynamic tracking movements,” in Multiple Muscle Systems, Springer, 1990, pp. 461–480. [15] R. Braakman and P. J. Vinken, “Unilateral facet interlocking in the lower cervical spine,” J. Bone Joint Surg. Br., vol. 49, no. 2, pp. 249–257, 1967. [16] R. Roaf, “A study of the mechanics of spinal injuries,” J. Bone Joint Surg. Br., vol. 42, no. 4, pp. 810–823, 1960. [17] R. J. Bauze and G. M. Ardran, “Experimental production of forward dislocation in the human cervical spine,” J. Bone Joint Surg. Br., vol. 60, no. 2, pp. 239–245, 1978. [18] R. W. Nightingale, J. H. McElhaney, W. J. Richardson, T. M. Best, and B. S. Myers, “Experimental Impact Injury to the Cervical Spine: Relating Motion of the Head and the Mechanism of Injury*,” J. Bone Jt. Surg., vol. 78, no. 3, pp. 412–21, 1996. [19] S. P. Broglio, E. E. Swartz, J. J. Crisco, and R. C. Cantu, “In Vivo Biomechanical Measurements of a Football Player’s C6 Spine Fracture,” N. Engl. J. Med., vol. 365, no. 3, pp. 279–281, Jul. 2011. [20] R. W. Nightingale, J. H. McElhaney, D. L. Camacho, M. Kleinberger, B. A. Winkelstein, and B. S. Myers, “The dynamic responses of the cervical spine: buckling, end conditions, and tolerance in compressive impacts,” SAE Technical Paper, 1997. [21] A. H. Burstein, “The response of the cervical spine to axial loading: feasibility for intervention,” 1994. [22] Y. K. Liu and Q. G. Dai, “The second stiffest axis of a beam-column: implications for cervical spine trauma,” J. Biomech. Eng., vol. 111, no. 2, pp. 122–127, 1989. [23] L. Voo and Y. K. Liu, “Segmented column model to predict human cervical spine buckling,” in Frontiers in head and neck trauma: Clinical and biomechanical, vol. 21, IOS Press, 1998, p. 442. [24] L. H. Sekhon and M. G. Fehlings, “Epidemiology, demographics, and pathophysiology of acute spinal cord injury,” Spine, vol. 26, no. 24S, pp. S2–S12, 2001. [25] R. W. Nightingale, D. L. Camacho, A. J. Armstrong, J. J. Robinette, and B. S. Myers, “Inertial properties and loading rates affect buckling modes and injury mechanisms in the cervical spine,” J. Biomech., vol. 33, no. 2, pp. 191–197, 2000.   101 [26] F. A. Pintar, N. Yoganandan, L. Voo, J. F. Cusick, D. J. Maiman, and A. Sances, “Dynamic characteristics of the human cervical spine,” SAE Technical Paper, 1995. [27] D. J. Maiman, A. Sances Jr, J. B. Myklebust, S. J. Larson, C. Houterman, M. Chilbert, and A. Z. El-Ghatit, “Compression injuries of the cervical spine: a biomechanical analysis,” Neurosurgery, vol. 13, no. 3, pp. 254–260, 1983. [28] N. M. Alem, G. S. Nusholtz, and J. W. Melvin, “Head and neck response to axial impacts,” SAE Technical Paper, 1984. [29] R. W. Nightingale, B. J. Doherty, B. S. Myers, J. H. McElhaney, and W. J. Richardson, “The influence of end condition on human cervical spine injury mechanisms,” SAE Technical Paper, 1991. [30] G. S. Nusholtz, J. W. Melvin, D. F. Huelke, N. M. Alem, and J. G. Blank, “Response of the cervical spine to superior-inferior head impact,” SAE Technical Paper, 1981. [31] N. Yoganandan, M. Haffner, D. J. Maiman, H. Nichols, F. A. Pintar, J. Jentzen, S. S. Weinshel, S. J. Larson, and A. Sances, “Epidemiology and injury biomechanics of motor vehicle related trauma to the human spine,” SAE Technical Paper, 1989. [32] D. J. Maiman, N. Yoganandan, and F. A. Pintar, “Preinjury cervical alignment affecting spinal trauma,” J. Neurosurg. Spine, vol. 97, no. 1, pp. 57–62, 2002. [33] A. Saari, E. Itshayek, and P. A. Cripton, “Cervical spinal cord deformation during simulated head-first impact injuries,” J. Biomech., vol. 44, no. 14, pp. 2565–2571, 2011. [34] G. S. Nusholtz, D. E. Huelke, P. Lux, N. M. Alem, and F. Montalvo, “Cervical spine injury mechanisms,” SAE Technical Paper, 1983. [35] R. W. Nightingale, J. H. McElhaney, W. J. Richardson, and B. S. Myers, “Dynamic responses of the head and cervical spine to axial impact loading,” J. Biomech., vol. 29, no. 3, pp. 307–318, 1996. [36] G. S. Nusholtz, J. W. Melvin, D. E. Huelke, N. M. Alem, and J. G. Blank, “Response of the cervical spine to superior-inferior head impact,” 1996, pp. 197–237. [37] J. Hu, K. H. Yang, C. C. Chou, and A. I. King, “A numerical investigation of factors affecting cervical spine injuries during rollover crashes,” Spine, vol. 33, no. 23, pp. 2529–2535, 2008. [38] D. L. Camacho, R. W. Nightingale, and B. S. Myers, “The influence of surface padding properties on head and neck injury risk,” J. Biomech. Eng., vol. 123, no. 5, pp. 432–439, 2001. [39] D. L. Camacho, R. W. Nightingale, and B. S. Myers, “Surface friction in near-vertex head and neck impact increases risk of injury,” J. Biomech., vol. 32, no. 3, pp. 293–301, 1999.   102 [40] N. Yoganandan, A. Sances Jr, D. J. Maiman, J. B. Myklebust, P. Pech, and S. J. Larson, “Experimental spinal injuries with vertical impact.,” Spine, vol. 11, no. 9, pp. 855–860, 1986. [41] M. M. Panjabi, T. Miura, P. A. Cripton, J.-L. Wang, A. S. Nain, and C. DuBois, “Development of a system for in vitro neck muscle force replication in whole cervical spine experiments,” Spine, vol. 26, no. 20, pp. 2214–2219, 2001. [42] A. T. Dibb, C. A. Cox, R. W. Nightingale, J. F. Luck, H. C. Cutcliffe, B. S. Myers, K. B. Arbogast, T. Seacrist, and C. R. Bass, “Importance of muscle activations for biofidelic pediatric neck response in computational models,” Traffic Inj. Prev., vol. 14, no. sup1, pp. S116–S127, 2013. [43] K. Brolin, S. Hedenstierna, P. Halldin, C. Bass, and N. Alem, “The importance of muscle tension on the outcome of impacts with a major vertical component,” Int. J. Crashworthiness, vol. 13, no. 5, pp. 487–498, 2008. [44] R. S. Newell, G. P. Siegmund, J.-S. Blouin, J. Street, and P. A. Cripton, “Cervical Vertebral Realignment When Voluntarily Adopting a Protective Neck Posture,” Spine, vol. 39, no. 15, pp. E885–E893, 2014. [45] V. R. Hodgson and L. M. Thomas, “Mechanisms of cervical spine injury during impact to the protected head,” SAE Technical Paper, 1980. [46] T. S. Nelson, “Towards a neck injury prevention helmet for head-first impacts: a mechanical investigation,” University of British Columbia (Vancouver), 2011. [47] T. S. Nelson and P. A. Cripton, “White Paper: A helmet for prevention and mitigation of spinal column and spinal cord injuries in Head First Impacts,” 2008. [48] P. H. Halldin, K. Brolin, S. Kleiven, H. von Holst, L. Jakobsson, and C. Palmertz, “Investigation of conditions that affect neck compression-flexion injuries using numerical techniques,” in Proc. 44th Stapp Car Crash Conference, Atlanta, GA, Society of Automotive Engineers, Inc., SAE Technical Paper, 2000, pp. 127–138. [49] B. Heudorfer, M. Breuninger, U. Karlbauer, M. Kraft, and J. Maidel, “Roofbag—a concept study to provide enhanced protection for head and neck in case of rollover,” in Proceedings of the 19th International Technical Conference on the Enhanced Safety of Vehicles, Washington DC, June, 2005, pp. 6–9. [50] M. Aare and P. Halldin, “A new laboratory rig for evaluating helmets subject to oblique impacts,” Traffic Inj. Prev., vol. 4, no. 3, pp. 240–248, 2003. [51] K. Vaughan and S. Peldschus, “Declaration by independent reviewers.” [52] T. S. Nelson and P. A. Cripton, “A new biofidelic sagittal plane surrogate neck for head-first impacts,” Traffic Inj. Prev., vol. 11, no. 3, pp. 309–319, 2010.   103 [53] W. Daniel, Adams - contacts overview, best practices, and tips. 2011. [54] A. C. Fischer-Cripps and I. Mustafaev, Introduction to contact mechanics. Springer, 2000. [55] E. Kroll, S. S. Condoor, and D. G. Jansson, Innovative conceptual design: theory and application of parameter analysis. Cambridge University Press, 2001. [56] J. R. Funk, J. M. Cormier, C. E. Bain, H. Guzman, E. Bonugli, and S. J. Manoogian, “Head and neck loading in everyday and vigorous activities,” Ann. Biomed. Eng., vol. 39, no. 2, pp. 766–776, 2011.    104 Appendices  Appendix A  Machine Drawings for Passive Deployment Mechanism    Figure A-1: Passive Mechanism Deployment Guide   105   Figure A-2: Passive Mechanism Deployment Pin   106 Appendix B  Machine Drawings for New Deployment Mechanism   Figure B-1: Locking cylinder in new mechanism   107  Figure B-2: Switch    Figure B-3: Sliding block   108  Figure B-4: New deployment guide     

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0165740/manifest

Comment

Related Items