Understanding Astronaut Shoulder Injury

Home , Astronautics, Constellation Space Suit, Hard Upper Torso, Mark III (space suit), Space suit

Human-Spacesuit Interaction: Understanding Astronaut Shoulder Injury

ALEXANDRA MARIE HILBERT

B.S. Mechanical Engineering Cornell University, 2013

Submitted to the Department of Aeronautics and Astronautics in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE IN AERONAUTICS AND ASTRONAUTICS at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY

June 2015

Signature of Author Department of Aeronautics and Astronautics May 21, 2015

Certified by Dava J. Newman, Ph.D. Apollo Professor of Astronautics and Engineering Systems Director of Technology and Policy Program Thesis Supervisor

Accepted by Paulo C. Lozano, Ph.D. Associate Professor of Aeronautics and Astronautics Chair, Graduate Program Committee

Human-Spacesuit Interaction: Understanding Astronaut Shoulder Injury

ALEXANDRA MARIE HILBERT

Submitted to the Department of Aeronautics and Astronautics on May 21, 2015 in Partial Fulfillment of the Requirements for the Degree of

Master of Science in Aeronautics and Astronautics

ABSTRACT

Extravehicular activities (EVA), or space walks, are a critical and complex aspect of human spaceflight missions. To prepare for safe and successful execution of the required tasks, astronauts undergo extensive training in the Neutral Buoyancy Lab (NBL), which involves many hours of performing repetitive motions at various orientations, all while wearing a pressurized spacesuit. The current U.S. spacesuit—the Extravehicular Mobility Unit (EMU)—is pressurized to 29.6 kPa (4.3 psi) and requires astronauts to exert a substantial amount of energy in order to move the suit into a desired position. The pressurization of the suit therefore limits human mobility, causes discomfort, and leads to a variety of contact and strain injuries. Shoulder injuries are one of the most severe injuries that astronauts contend with, and are mainly attributed to the EMU’s hard upper torso (HUT). While suit-related injuries have been observed for many years and some basic countermeasures have been implemented, there is still a lack of understanding of how humans move inside the spacesuit. The objective of this research is therefore to gain a greater understanding of this human-spacesuit interaction and potential for shoulder injury through two approaches: quantifying and analyzing the suit-induced pressures that arise in the shoulder region, and comparing the shoulder muscle forces that arise in the unsuited and suited conditions by modeling human-spacesuit interaction.

The first approach provides an “inside look” of the pressure distributions and pressure profiles that arise at the interface between the human shoulder and the torso of the spacesuit, thereby suggesting which areas of the shoulder might be prone to contact injury. A commercially produced pressure sensing system is used to collect shoulder pressure data during a human subject experiment that involves three experienced subjects performing a series of upper body motions in both unsuited and suited conditions. Pressure distributions reveal that: 1) the least experienced subject generates the highest pressures, 2) for the majority of movements for all subjects, pressure is concentrated just above the clavicle over the soft musculature at the top of the shoulder, 3) the top of the shoulder is one of the regions in which maximum pressure is located most frequently, and 4) the shoulder blade is a secondary region of concern with regards to frequency of experiencing maximum pressure. Pressure profile analysis reveals that 1) for most subjects, general profile trends vary in shape across movement groups, 2) repetitions within each

3 movement group are consistent in shape, and for most subjects also in magnitude, 3) the highest pressures are typically found near the top of the shoulder, and 4) the shoulder blade area is of concern for at least one subject. As these results are primarily observational in nature, a statistical analysis is performed to assess the effects of motion type and anthropometric region on peak pressure magnitudes. This analysis shows that results cannot be generalized across subjects as they are likely affected by individual anthropometry, suit fit, and the biomechanics of how each subject performs the motion. However, a number of interesting trends regarding which motions or regions yield higher pressures are found for each of the individual subjects. The results are specific to the subjects, suit sizes, and experimental conditions used in this particular experiment; however, the application of these quantitative and repeatable techniques during future experiments, suit fit sessions, or NBL runs would lead to a more complete understanding of human-spacesuit interaction at the shoulder interface.

The second approach analyzes the effects of spacesuits on muscle forces in the shoulder region. Data regarding spacesuit joint torques and the joint angles of a suited subject are integrated into an upper-extremity musculoskeletal model in OpenSim to evaluate which muscles are most affected by the spacesuit. Looking specifically at a shoulder abduction/adduction motion, shoulder abductors, adductors, and stabilizer muscle groups are evaluated for significant changes in force from the unsuited to suited condition, and individual muscles within the shoulder region are also evaluated for significant changes from the unsuited to suited conditions. From a statistical analysis of the musculoskeletal simulation results, it is found that of the three investigated muscle groups—shoulder abductors, adductors, and stabilizers—only the abductors experience a statistically significant change in total muscle force between the unsuited and suited conditions. Looking specifically at the individual muscles that constitute the abductors and stabilizers, we find that only the middle deltoid experienced a statistically significant change in force from the unsuited to suited condition. A number of explanations are provided for the observed force profiles and the statistical results. The presented results are specific to the subject’s motion data, suit torque data, and the musculoskeletal model that are used; however expanding this analysis to more subjects, other body joints, and a more complex musculoskeletal model would provide useful results for industry experts. Valuable information could be provided to EVA operations teams, flight doctors, and spacesuit designers regarding which movements or tasks should be avoided or performed minimally to prevent injury. The resulting muscle forces could also be used to set limits on the joint torques that are engineered in future spacesuits.

Each of the approaches implemented in this thesis provides a different avenue for addressing the issue of shoulder injury in the spacesuit. While the pressure analysis contributes to the understanding of human-spacesuit interaction by informing on the anthropometric regions that might be most susceptible to contact injury, the musculoskeletal analysis provides insight as to which individual muscles are most susceptible to strain injury. Both of these quantitative, evidence-based approaches contribute to an increased understanding of the potential for shoulder injury in the spacesuit.

Thesis Supervisor: Dava J. Newman, Ph.D. Title: Apollo Professor of Astronautics and Engineering Systems Director of Technology and Policy Program Massachusetts Institute of Technology

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my wonderful advisor, Professor Dava Newman, for providing me with numerous once-in-a-lifetime opportunities during my short grad school experience. I feel extremely fortunate to have had you as an advisor as your energy and passion for space exploration are truly inspiring. My experience at MIT would not have been the same without your support, guidance, and encouragement. I wish you the best in your own new adventure over the next few years!

To Professor Leia Stirling and Dr. Aleksandra Stankovic, thank you for helping me with the design of my statistical analyses. To Gaurav, Alex, Ana, and Dustin, thank you for taking the time to help me with all of the intricacies of OpenSim. To all of our collaborators at David Clark and NASA—particularly Shane Jacobs, Shane McFarland, Lindsay Aitchison, and Amy Ross—thank you for allowing us to perform our experiments in your facilities and for taking the time to give us invaluable feedback on our research.

To Allie and Ana, thanks for leaving me some work to do! But really, thank you for paving the way for me and getting me up to speed on the whole astronaut injury project. I really appreciate that both of you were always taking the time to explain things to me and help me come up with good ideas for my Master’s thesis. Your previous work made my life a lot easier!

To Ana and Raquel…I know it sounds super cliché, but you guys know it’s true…thank you for being the best officemates ever! You were both so welcoming from my very first day at MIT, and I could not imagine lovelier ladies to share an office with. Thanks to both of you for never being “too cool” to hang out with the new students. Ana - from our coffee outings when we were first getting to know each other to our seven hour road trips to upstate New York (and the romantic McDonald’s dinners at Lee along the way), it has really been a fun two years. I’m so excited for you to graduate and to see all of the amazing things that you do in the future. Raquel – we took a little bit longer to get to know each other well, but before long I really felt like you were a sister to me. Thank you for always being there to listen to me (whether I was having a hard time or just chatting away to avoid doing work), and for bringing a little piece of Texas to my life in Boston. Also, don’t forget to look for jobs in Colorado! I know I will see both of you before too long, whether it’s in Colorado, Texas, or Ithaca, but the three of us should plan an exotic reunion trip sometime within the next couple of years!

To Pierre….where to begin?! Thank you for being my partner in crime over the past two years. It really wouldn’t have been the same without you…no terrible jokes, no epic movies, nobody to make fun of how much I eat, nobody for me to pick on, nobody to take Wednesday selfies with….it certainly would have been a boring two years without you. I know it will be hard, but try not to miss me too much. There will be some tears, but just remember -- if you are missing me, just eat a donut! In all seriousness, thank you for all of the crazy times over the past two years. Don’t forget to visit me (Texas AND Colorado)!

To Eddie, thank you not only for being my thesis buddy over these last few weeks, but also for being such a great friend over the past two years. You are honestly one of the nicest people I’ve ever met – you bring such positive energy with you wherever you go and are always up for an

5 adventure (like going sledding the day before quals). We’ve had some really fun times and I hope you come visit me in Colorado so we can continue the fun! Keep doing big things!

To all of the other MVLers, thank you for making our lab such a fun and welcoming place. My time at MIT would have been entirely different (and much less fun) if it weren’t for all of you.

To Tio Tin, Gloria, Veronica, Natalia, and Isabel – I know I was only able to visit once or twice each year, but you have no idea what a difference those visits made to me. Just spending a few days with you all was always a refreshing break from the stressful environment of college. Tio Tin and Gloria - thank you for always picking me up at far-away bus stations, welcoming me into your home, and feeding me delicious home-cooked meals. To the girls – you really have no idea how much I have enjoyed being able to see you ever since you moved to New York. You are like sisters to me. I love you all very much and can’t wait to see what is coming in your futures!

To John, thank you for always being there for me, through the best and worst of times. These two years have been an experience of a lifetime for me, so thank you for always supporting me, visiting me, and being amazingly patient. It’s crazy to think of where we were at the beginning of my time at grad school, and where we are now. Whether in Boston, Owego, New Jersey, Texas, or elsewhere, we have had so many memorable adventures together – and that’s only our beginning. I love you and I can’t wait for us to finally be together in Denver. It will be here sooner than we realize….just a few more months until we are done with goodbyes!

And finally, to my family – I could not have survived two years at MIT, much less four years at Cornell, without your inspiration, encouragement, prayers, and love. While I can’t quite say that I am finally escaping the cold to come work close to home, I think Colorado will be slightly less cold and am thankful that I will be only a short flight away from home. And of course I’m looking forward to spending some relaxing time at home this summer! Mami and Daddy – thank you for fostering my passion for science and for supporting me throughout my college career. Now you can officially say that both of your daughters are done with school! To Caroline – I know you think I have just been drawing goofy pictures of aliens for the past two years, so I’m sorry that this thesis won’t live up to those expectations…But really, thank you for being the best sister I could have asked for. And hats off to you for graduating from 3 more years of school than me. I love you all and can’t wait to see you very soon!

TABLE OF CONTENTS

Abstract…………………………………………………………………………………………………… 3

Acknowledgements…………………………………………………………………………………….... 5

Index of Figures…………………………………………………………………………………………... 9

Index of Tables………………………………………………………………………………………...... 10

CHAPTER 1: INTRODUCTION 1.1 Motivation ...... 11 1.2 Objectives, Specific Aims, and Contributions ...... 12 1.3 Thesis Outline ...... 13

CHAPTER 2: LITERATURE REVIEW 2.1 Extravehicular Activity ...... 15 2.2 Astronaut Injury ...... 17 2.3 Shoulder Injury in Astronauts ...... 19 2.4 Countermeasures for Shoulder Injury ...... 22 2.5 Development of a Quantitative Understanding of Human-Spacesuit Interaction ...... 23

CHAPTER 3: HUMAN-SPACESUIT INTERACTION QUANTIFICATION 3.1 Overview ...... 25 3.2 Experimental Methods ...... 26 3.2.1 Sensor Systems...... 26 3.2.2 High-Pressure Shoulder Sensor ...... 26 3.2.3 Sensor Calibration ...... 27 3.2.4 Experimental Design ...... 28 3.2.5 Sensor Orientation ...... 29 3.2.6 Limitations ...... 31 3.3 Graphical and Numerical Analysis ...... 31 3.3.1 Methods ...... 31 3.3.2 Results ...... 32

3.3.2.1 Pressure Distributions...... 32 3.3.2.2 Maximum Pressure Locations ...... 34 3.3.2.3. Pressure Profiles ...... 36 3.3.3 Discussion...... 41 3.4 Statistical Analysis: Effect of Motion and Anthropometric Region on Peak Pressure Magnitudes ...... 42 3.4.1 Methods ...... 42 3.4.2 Results ...... 43 3.4.3 Discussion...... 48 3.5 Conclusions and Future Work ...... 48

CHAPTER 4: MUSCULOSKELETAL MODELING 4.1 Overview ...... 51 4.2 Modeling Methods ...... 51 4.2.1 Human Modeling ...... 52 4.2.2 Spacesuit Modeling ...... 53 4.2.3 Human-Spacesuit Interaction Modeling ...... 55 4.3 Analytical Methods ...... 57 4.4 Results ...... 58 4.4.1 Muscle Groups ...... 58 4.4.2 Individual Muscles ...... 61 4.5 Discussion ...... 65 4.6 Limitations ...... 67 4.7 Conclusion ...... 68

CHAPTER 5: CONCLUSION……………………………………………………..…………………………………….71

Appendices..…………………………………………………………………………………………..… 79

INDEX OF FIGURES

Figure 2.1: Extravehicular Mobility Unit and exploded view diagram…………………….………16

Figure 2.2: Pivoted HUT and Planar HUT………………………………………………………….…16

Figure 2.3: The Mark III planetary exploration suit……………………………………………….….17

Figure 2.4: Astronaut training in the NBL in an inverted position………………………………….20

Figure 2.5: Restriction of scapulothoracic motion in the HUT……………………………………….20

Figure 2.6: Diagram of the shoulder harness inside the planar HUT………………………………22

Figure 2.7: Most commonly used removable shoulder padding in the EMU………………………23

Figure 3.1: In-suit sensor systems as placed on the subject……………………………………….....26

Figure 3.2: Experimental sensor systems………………………………………………………………26

Figure 3.3: Calibration device…………………………………………………………………………..27

Figure 3.4: Movement tasks performed by each subject……………………………………..……….28

Figure 3.5: Example test protocol for one subject…………………………………………………….29

Figure 3.6: Orientation of Novel sensor……………………………………………………………..…30

Figure 3.7: Subject-specific orientation of Novel sensor relative to anthropometric landmarks…30

Figure 3.8: In-suit harness of the Mark III with attached shoulder pads……………………….…..31

Figure 3.9: Pressure distributions over the shoulder at the point of maximum total pressure….32

Figure 3.10: Shoulder locations that experience max. pressure during the movement……………35

Figure 3.11: Subject 1: Pressure profiles (kPa) plotted for selected sensors………………..………36

Figure 3.12: Subject 2: Pressure profiles (kPa) plotted for selected sensors…………………..……38

Figure 3.13: Subject 3: Pressure profiles (kPa) plotted for selected sensors…………………..…….39

Figure 3.14: Maximum pressure profile (kPa) of the unsuited condition…………………….…….41

Figure 3.15: Subject 1: Post-hoc pairwise comparisons within anthropometric regions…….……43

Figure 3.16: Subject 1: Post-hoc pairwise comparisons within motion types………………….…..44

Figure 3.17: Subject 2: Post-hoc pairwise comparisons within anthropometric regions…….……45

Figure 3.18: Subject 2: Post-hoc pairwise comparisons within motion types………………………45

Figure 3.19: Subject 3: Post-hoc pairwise comparisons within anthropometric regions…………..46

Figure 3.20: Subject 3: Post-hoc pairwise comparisons within motion types…………………..….47

Figure 4.1: Adjusted upper extremity model used for musculoskeletal analysis………………….52

Figure 4.2: Modified fish-scale test setup for shoulder abduction/adduction of the Mark III…..53

Figure 4.3: Torque-angle curves resulting from testing of the shoulder abduction/adduction motion of the Mark III at 4.3 psi………………………………………………………………………..54

Figure 4.4: Extracted conservative values for the torque-angle relationships in Figure 4.3……...54

Figure 4.5: IMU placement on the body inside the suit………………………………………….…..55

Figure 4.6: Conceptual depiction of suited shoulder abduction/adduction…………………….....55

Figure 4.7: Modeling methodology depicting the steps for analyzing muscle dynamics…….…..56

Figure 4.8: Rotator cuff muscles…………………………………………………………………….…..57

Figure 4.9: Shoulder abductors and adductors of the chosen model……………………………..…58

Figure 4.10: Total force by muscle group………………………………………………………………60

Figure 4.11: Boxplots comparing unsuited and suited cases, by muscle group……………….…..61

Figure 4.12: Individual abductor muscle forces………………………………………………………62

Figure 4.13: Individual stabilizer muscle forces………………………………………………………62

Figure 4.14: Boxplots comparing unsuited and suited cases, for individual muscles…….………64

INDEX OF TABLES

Table 3.1: Peak pressures of each motion for the three subjects……………………………………...40

Table 3.2: Summarized results of statistical analysis…………………………………………...……..47

Table 4.1: Relevant upper body anthropometric measurements of Subject 3………………………56

Table 4.2: Peak forces by muscle group………………………………………………………..………60

Table 4.3: Peak forces for individual muscles……………………………………………………..…..63

Table 4.4: Summarized results – peak forces for muscle groups and muscles………………………..…..66

Table 4.5: Maximum isometric forces of considered muscles……………………………………….67

INTRODUCTION

1.1 MOTIVATION As we begin the next chapter of human spaceflight, we must consider the incredible challenges involved in the exploration of other planetary bodies. When considering astronaut health, these challenges can be divided into two categories: maintaining healthy biological levels during transit to and from the destination, and protecting and aiding the astronaut while performing mission critical tasks on the surface of the planetary body. Whether returning to the Moon, landing on Mars, or rendezvousing with a near-Earth asteroid, astronauts performing extravehicular activity (EVA) need protection from harsh environments and require high levels of mobility and endurance to allow for successful execution of manual tasks. The number of hours spent performing EVAs during these future missions will present a substantial increase from any mission that has ever occurred. As a result, the time astronauts spend physically training for such missions will also increase significantly, thereby exacerbating any current issues that arise as a result of EVA training. Since 1999, evidence of astronaut injury due to EVA training in the current gas-pressurized spacesuit has grown. While studies investigating the prevalence and likely mechanisms of these injuries have led to appropriate countermeasures and operational changes, more recent literature has not shown a decrease in astronaut injury. If we intend to train astronauts for exploration missions involving extensive EVAs, it is imperative that we develop a complete understanding of the issues at hand. Creating quantitative and repeatable methods of identifying these issues is a crucial step toward mitigating astronaut injury.

1.2 OBJECTIVES, SPECIFIC AIMS, AND CONTRIBUTIONS The primary research objective of this work is to develop an increased understanding of the potential for shoulder injury in pressurized spacesuits. This is accomplished through two specific aims.

The first specific aim is to quantify and analyze the pressure distributions and pressure profiles that arise at the interface between the human shoulder and the torso of the spacesuit. This provides insight as to which areas of the shoulder are prone to contact injury. A commercially- produced pressure-sensing system (Novel GmbH, Munich, Germany) is used to collect shoulder pressure data during a series of suited and unsuited motions. The following research questions are evaluated using a range of graphical, numerical, and statistical analyses:

. Do anthropometric regions of concern vary between subjects? . Does experience in the suit have an effect on the pressure magnitudes experienced? . Does anthropometric region have an effect on the pressure magnitudes experienced? . Does motion type have an effect on the pressure magnitudes experienced?

The second specific aim is to assess muscle forces in the upper arm and shoulder regions during unsuited and suited simulations of shoulder motions using an adjusted methodology for modeling human-spacesuit interaction. Data regarding spacesuit joint torques and the body joint angles of a suited subject are integrated into a musculoskeletal model in OpenSim (Stanford, CA). The differences between the unsuited and suited muscle forces during a shoulder abduction/adduction motion are evaluated and discussed. The results address the following research questions:

. Which muscle groups experience a significant change in muscle force from the unsuited to suited condition? . Within the muscle groups of interest, which individual muscles experience a significant change in muscle force from the unsuited to suited condition?

The five primary contributions of this research are as follows:

1. Presenting the first quantitative “look” inside the spacesuit at the shoulder interface 2. Developing quantitative methods for analyzing pressure distributions and pressure profiles that arise during dynamic shoulder motions 3. Quantitatively evaluating the effects of anthropometric region and motion type on pressures experienced at the shoulder 4. Expanding an existing methodology for musculoskeletal modeling of human-spacesuit interaction to a complex shoulder joint motion 5. Improving an existing methodology for musculoskeletal modeling of human-spacesuit interaction through the use of suited body joint angles obtained from inertial measurement units

All of these contributions provide evaluative methods through which the likelihood of astronaut shoulder injury can be identified and therefore mitigated. These methods are transferrable for use in any spacesuit or other shoulder interface of interest, and can also be implemented at other body locations.

1.3 THESIS OUTLINE

Chapter 1 introduces the issue of astronaut injury in the spacesuit during extravehicular activity (EVA), and presents the primary objectives and contributions of this research.

Chapter 2 reviews relevant literature on EVA training, astronaut injury, shoulder-specific injury, and previous studies and countermeasures regarding these issues.

Chapter 3 details the methods, results, and discussion of the shoulder pressure data that were gathered during a human subjects experiment inside the spacesuit. A combination of graphical, numerical, and statistical analyses are implemented to examine the data. Results regarding the pressure distributions, locations of maximum pressure, and which motions and anthropometric regions have an effect on the pressure magnitudes reached are presented.

Chapter 4 presents an adjusted methodology for integrating spacesuit mechanics, human biomechanics, and human-spacesuit interaction into a musculoskeletal model. Results and discussion are also presented on the force generations of muscle groups and individual muscles in the shoulder region, and how these vary between the unsuited and suited conditions.

Finally, Chapter 5 provides a summary and conclusion of all results of the thesis. Recommendations for expanding upon this work are made as well.

LITERATURE REVIEW

2.1 EXTRAVEHICULAR ACTIVITY Considering that we have performed a number of extravehicular activity (EVA) expeditions on the Moon and in the vacuum of space, it is common to assume that we are well prepared for future EVAs on the Moon and Mars. However, when we look at the facts of past missions and those planned for the future, we see that there are still large challenges to overcome. While less than 20 lunar EVAs were performed in the entire Apollo Program and no previous astronaut or spacesuit has performed more than three lunar EVAs, the Constellation Lunar Program missions planned for up to 24 hours of EVA per crew member per week, and up to 76 lunar EVAs during a 6-month mission (Gernhardt et al., 2009). This huge increase in EVA hours would only grow for Mars missions as we will be staying for longer durations of up to 500 days. Such an increase in EVA hours is of particular concern, not only because it will be an unprecedented amount of time in the spacesuit on the surface of another planetary body, but also because it will correlate to an even larger increase in training hours in the suit while still on Earth.

Currently, extravehicular activity training focuses on preparing for the microgravity environment of space, as astronauts are only sent to the International Space Station (ISS). Training for this weightless environment is performed primarily in the Neutral Buoyancy Laboratory (NBL), a 23.5 million liter pool located at NASA Johnson Space Center (Houston, TX) that contains full-size mock-ups of the International Space Station (ISS). Training in this neutrally buoyant

15 environment provides astronauts with an analog to the weightlessness they will experience in low Earth orbit (LEO). Though hours spent in the NBL are mission-specific, astronauts spend an average of 11.6 hours of training in the NBL per each hour of planned in-flight EVA (Strauss et al., 2005). In the NBL, astronauts wear training versions of the currently operational spacesuit (NASA, 1998; NASA NBL Standards, 2002; NASA Hazard Analysis, 2002), and are outfitted with weights in order to make them neutrally buoyant (Johnson et al., 2004). The current U.S. spacesuit—the Extravehicular Mobility Unit (EMU)—is a 14-layer suit weighing 64 kg (140 lb) FIGURE 2.1: EXTRAVEHICULAR MOBILITY UNIT (AT LEFT) that is pressurized to 29.6kPa (4.3 psi). AND EXPLODED VIEW DIAGRAM (AT RIGHT). (IMAGE With the addition of the portable life SOURCES: NASA, HAMILTON SUNDSTRAND) support system (PLSS) backpack, the total suit weight comes to approximately 115 kg (254 lb). One of the central components is the hard upper torso (HUT), a fiberglass shell that connects to the arm, helmet, and lower torso assemblies (Strauss et al., 2005). Two HUT designs, the pivoted and the planar HUTs, are available for use in the NBL, although only the planar HUT is used currently in spaceflight. The planar HUT has planar scye bearings in fixed planes at the arm openings, whereas the FIGURE 2.2: PIVOTED HUT (AT LEFT) AND PLANAR HUT (AT pivoted HUT has a shoulder gimbal with a two-point pivot, thereby aiding the RIGHT). (IMAGE SOURCE: NASA) range of motion of the shoulder joint (Strauss et al., 2005).

Given the operational exploration requirements of future missions to the Moon or Mars where mobility, agility and dexterity are high priorities, use of the Extravehicular Mobility Unit (EMU) is entirely infeasible. The EMU severely limits human mobility and requires astronauts to exert a substantial amount of energy in order to move the suit into a desired position. Lower body mobility is particularly restricted as the EMU was originally designed to satisfy minimal mobility and operational requirements for performing EVAs during the Space Transportation System, or Shuttle era (Jordan et al., 2005).

Over the years, a number of prototype suits have been developed for planetary exploration. One of particular focus in this thesis is the Mark III, first built in 1987 by both NASA and ILC Dover and having undergone a number of iterations in the years since then. Shown in Figure 2.3, the Mark III is the best characterized prototype spacesuit. The suit presents an increase in mobility

16 over the EMU using several rotating bearings, but its structure and programming cause astronauts to use different biomechanical strategies than they would typically use while unsuited. The torso and hip assemblies are hard shells, while soft goods cover the arms and legs.

While planetary exploration prototype suits have been designed to enhance astronauts’ range of motion and improve mobility, the interface between the suit and the human remains an issue of concern. Most spacesuit evaluations consider the human-spacesuit system as a whole (Morgan et al., 1996; Jaramillo et al., 2008; Matty and Aitchison, 2009; Norcross et al., 2010; Aitchison, 2012; Valish and Eversley, 2012); however, the kinematics of the human inside the suit do not typically match the motion of the suit itself. For example, the user must first move her arm to make contact with the suit and then move the arm and spacesuit joint after contact to perform a useful motion. This difference in kinematics not only requires the astronaut to exert extra energy, but also has the potential to lead to a variety of injuries. Suit pressurization and the weight of the suit itself can also present potential sources of injury. The prevalence and mechanisms of astronaut injury are discussed in the following sections.

FIGURE 2.3: THE MARK III PLANETARY EXPLORATION SUIT. (IMAGE SOURCE: NASA)

2.2 ASTRONAUT INJURY Since the start of NASA’s human spaceflight programs, anecdotal reports from astronauts have mentioned the occurrence of in-flight musculoskeletal injuries (Scheuring et al., 2009). However, it was not until 1999, in parallel with the steep increase in EVA training in preparation for assembly of the International Space Station (ISS), that injury frequency began to be viewed as a major mission concern (Johnson et al., 2004). Soon after, the Lifetime Surveillance of Astronaut Health (LSAH) published the results of an investigation of musculoskeletal injury rates of Shuttle astronauts from STS-1 through STS-89 (Wear, 1999). The results suggested that astronauts had a greater in-flight injury rate than comparison control subjects, and that during their mission period of one year pre-flight to one year post-flight, astronauts experience an injury rate three times higher than outside their mission period (Wear, 1999). In the years following, a number of investigations were launched to further characterize astronaut injury and to understand whether the injuries were a result of pre-flight training, in-flight injury, or post-flight injury due to physiological deconditioning (Scheuring et al., 2009).

During a one-year span from 2002 to 2003, one study investigated 83 astronauts who trained in the NBL (Viegas et al., 2004). Twenty-eight astronauts were found to have hand symptoms, while 19 had shoulder symptoms and 9 experienced elbow symptoms. The authors suggested that training until astronauts’ launch dates could cause their injuries to persist during spaceflight.

Another study, performed over an 18-month timeframe from 2002 to 2004, specifically considered NBL EVA training and determined the frequency and incidence rates of injury symptoms by body location and also characterized the mechanisms of injury and suggested countermeasures (Strauss, 2004; Strauss et al., 2005). Of 770 spacesuit symptom questionnaires from 86 subjects, it was found that 24.6% of tests yielded symptoms. Of the total number of reported symptoms, 47.6% were in the hands, 20.7% were in the shoulders, and 11.4% were in the feet. The most severe symptoms were found in the shoulders, hands and feet, in decreasing order of severity. It was stated that causal mechanisms included hard contact with the spacesuit components and muscle strains for the shoulder, moisture and hard contact leading to fingernail injuries in the hands, and hard contact with the boots for the feet.

At the end of 2005, the Space Medicine Division at NASA Johnson Space Center (JSC) began planning the Apollo Medical Operations Project—a study that would eventually interview 14 of the surviving Apollo astronauts in an effort to gather experience-based recommendations for improving crew health and performance for future exploration missions (Scheuring et al., 2008). Recommendations relevant to the EVA suit focused primarily on improving suit functionality and the human factors integration. Frequent recommendations were made for improvement to glove dexterity and a general increase in suit functionality by increasing its flexibility while decreasing its mass and pressure level and lowering the center of gravity. It was believed that these changes would also aid in decreasing astronaut fatigue. Participants stated that potential risks included falling onto an outstretched arm, falling from a height, performing rover operations, and traversing slopes greater than 20-26 degrees inclination, while actual injury occurrences included one shoulder strain resulting from a malfunctioning drilling tool, one wrist laceration from the wrist ring of the EVA suit, and frequent forearm soreness due to the EVA glove.

In 2009, a study was published on the analysis of a newly created database of in-flight musculoskeletal injuries in the U.S. space program, comprehensive of missions from the Mercury program all the way through the conclusion of ISS Expedition 13 in September 2006 (Scheuring et al., 2009). For each injury, the data collected included the location, type, and mechanism of injury, as well as the type of in-flight exercise performed by the crewmember, previous history of injury, treatment, and the post-flight outcome. Analysis of the 219 identified injuries showed that the most common locations of injury were the hand, back, and shoulder, while the most common injury types were abrasions, contusions, strains, and lacerations. Leading mechanisms of injury included crew activities, EVA suit components, and exercise. Looking specifically at the 50 injuries that resulted from the EVA suit, the most common areas of injury were the hand, foot, and shoulder. There were also five additional injuries that occurred during EVA but did not result from suit components – four of which were muscular strains resulting from the EVA activities. Additionally, there were nine recorded incidences related to EVA on the lunar surface, of which five were in the hand and the others included wrist pain, shoulder strain, and lower extremity muscle fatigue.

The studies described above all represent overarching investigations of astronaut injury. Many of these findings sparked further investigations of injury in particular areas of the body. As the work presented in this thesis focuses on the human-spacesuit interaction at the shoulder, the following section provides a literature review of shoulder-specific injury studies.

2.3 SHOULDER INJURY IN ASTRONAUTS From 1999 to 2002, suspicions began to grow regarding a potential connection between EVA training and astronaut shoulder injury as a number of astronauts were being treated for musculoskeletal shoulder issues (Williams and Johnson, 2003; Scheuring et al., 2009). In 2002, a review of the records of every astronaut who performed an EVA from STS-82 to STS-111 (1997- 2002) revealed that four of the 38 astronauts experienced NBL-related injuries, no cases of NBL- related shoulder injuries were recorded from 1997 to 2002, but that 11 of the 16 astronauts on the astronaut strength, conditioning, and rehabilitation (ASCR) list were being treated for shoulder issues (McCluskey, 2002). As a result, a shoulder injury tiger team was created to evaluate the prevalence and causes of shoulder injury in EVA astronauts (Williams and Johnson, 2003).

The Tiger Team determined that NBL EVA training was indeed directly linked to a number of shoulder injuries (Johnson et al., 2004). Findings regarding the prevalence of shoulder symptoms included the following:

. 64% of surveyed EVA astronauts experienced some level of shoulder pain that was deemed attributable to EVA training in the EMU (Johnson et al., 2004) . 45% of subjects had preexisting shoulder injuries, 30% of which were reported to have been worsened by NBL training (Williams and Johnson, 2003) . Two of the astronauts—neither of which had any previous history of shoulder injury— had surgery for EVA training-related shoulder injuries (Johnson et al., 2004) . The percentage of crew reporting shoulder pain increased from astronaut candidates (0%) to contingency (11%), development runs (37%), EVA skills (45%), and finally to mission assigned training (56%), suggesting that the physical activity required for EVA training causes experienced EVA astronauts to have an increased risk of shoulder injury (Johnson et al., 2004) . Common reports of shoulder pain were made during or within 24 hours of training runs (Johnson et al., 2004)

Information was also gathered on the type of shoulder symptoms that were experienced. The Tiger Team state that “the typical episode of shoulder pain associated with EVA training at the NBL is described as a moderate dull ache over the top of the shoulder or within the shoulder joint that started either during the NBL run or within 24 hours following the run…the pain usually lasts less than a week” (Williams and Johnson, 2003). In 68% of cases, both shoulders are affected, while in 31% of cases it is solely an issue in the dominant shoulder. About half of those experiencing pain had tried adjusting suit fit or shoulder padding in order to prevent it. In addition, 55% of astronauts reported nocturnal pain that resulted from inflammation or rotator cuff tendon or muscle damage. While none of these conditions sound overly severe, only 55% of subjects claimed that the pain would disappear before their next NBL run (within 48 hours). The authors suggest that continued training before resolving such issues would increase the risk of overuse injury.

Data was also gathered on the suspected mechanisms of shoulder pain. This investigation revealed that the most common injury mechanisms were training in an inverted position, training in the planar HUT, performing repetitive motions, and using heavy tools (Johnson et al., 2004). Inversion, defined as a position in which the body is at a head-down angle of more than 45 degrees, essentially loads the shoulders with the astronaut’s full body weight. Although inversion during training typically lasts no more than a few minutes at a time, the lack of a restraint within the suit resulted in shoulder bruising for nearly every case. As a result of its interface with the human shoulder, the HUT was the hardware component that received the most attention. It was noted that some astronauts in fact prefer FIGURE 2.3: ASTRONAUT TRAINING IN training in the pivoted HUT as a means of optimizing suit fit THE NBL IN AN INVERTED POSITION. and reducing the risk of shoulder injury. Another hardware (IMAGE SOURCE: NASA) component that received substantial attention was EVA training tools. It was found that several of the most frequently used tools had not been adjusted to be neutrally buoyant in water, and were therefore a weight of between 5 and 11 pounds that the astronaut had to use underwater. The authors state that use of such a tool with fully extended arms would put the crewmember’s shoulder under a high level of stress. Another operational issue of note that was discovered in this investigation was the insufficient review of worksite design or changes to the worksite, which caused astronauts to have to work outside of a comfortable work envelope. It was suggested that doing so in addition to fighting the combined torques of their arm weight, tool weight, and the suit pressurization provides an avenue toward overuse and repetitive stress injuries.

A number of findings were also made regarding anthropometrics, biomechanics, and suit design and fit. The following represent a selection of the points most relevant to this thesis:

. None of the dimensions that were controlled in the design and production of the HUT were directly related to astronaut measurements being recorded at the time (Williams and Johnson, 2003) . EMU sizing techniques were based on linear measurements as opposed to volumetric measurements that take into account how the human fills the suit components (Johnson et al., 2004) . The suit sizing algorithm that was being used for the planar HUT was originally developed for the pivoted HUT, and did not include clinically relevant dimensions such as bi-acromial breadth and shoulder circumference (Johnson et al., 2004) FIGURE 2.4: RESTRICTION OF . The planar HUT restricts scapulothoracic motion SCAPULOTHORACIC MOTION IN THE HUT. (IMAGE SOURCE: WILLIAMS AND JOHNSON, of the shoulder joint due to the placement of the 2003) scye bearing, which can result in rotator cuff

impingement (Williams and Johnson, 2003; Johnson et al., 2004) . The size of the planar HUT’s body seal closure and scye bearings can increase the risk of shoulder injuries due to donning/doffing issues and mobility impairment (Johnson et al., 2004) . Astronauts at the time were not aware of the option to use a harness within the HUT, and no formal review of padding design was performed with the introduction of the planar HUT (Williams and Johnson, 2003) . Use of load alleviating devices and padding did not appear to significantly impair mobility or affect training (Williams and Johnson, 2003)

Considering these findings and the many more included in the Tiger Team report, the authors suggested that elimination of the risk of shoulder injury would require a combination of suit redesign along with optimized tools, tasks, and crew conditioning. They also recommended the following:

“Laser anthropometric studies of male and female astronauts, biomechanical analysis of shoulder joint motion in both genders, and use of CAD models of shoulder joint motion and EMU shoulder joint design should all be incorporated into the development of the next-generation space suit…Sustained emphasis on avoiding inverted body orientations, developing neutrally buoyant high-fidelity tools, working within the design envelope of the EMU, and crew conditioning are also critical in reducing the risk of injury.” (Williams and Johnson, 2003)

One of the more general astronaut injury studies mentioned previously (Strauss et al., 2005) also took a deeper look at shoulder injuries, leading to many of the same conclusions as the Tiger Team investigation. Forty-seven percent of injuries were caused by contact with the HUT, 46% were strain, sprain, tear, overuse, or impingement injuries, and 7% were attributed to harness contact. The shoulder symptoms that resulted from direct contact with the planar HUT occurred primarily at the scye joint (Graziosi et al., 2000), and were found to worsen with head down and shoulder and arms abducted (Strauss et al., 2005). Activities performed while head down, rotated laterally, or in a supine position tended to cause shoulder rotator cuff stress and strain injuries, which were also aggravated by the design of the planar HUT (Graziosi et al., 2000; Newman et al., 2000; Gonzalez et al., 2002). The authors noted that the mismatch between the HUT openings and the normal range of motion of the human shoulder put limits on the safe work envelope (Stankiewicz et al., 1993; Shields et al., 1998; Williams and Johnson, 2003). It was found that comfort pads, shoulder harnesses, planning for the use of heavy tools, working inside a safe work envelope, and avoiding shoulder-stressing maneuvers were all effective ways of decreasing the frequency of shoulder complaints (Strauss et al., 2005).

After these comprehensive studies were published and awareness of the issue of shoulder injury spread through the community, a large majority of the operational recommendations were taken into account and implemented. However, as of 2012, two crucial issues remained unaddressed: the planar HUT of the EMU was not redesigned, and there was limited ability to accurately track astronaut training since training occurs at several different locations (Scheuring et al., 2012).

In an effort to evaluate astronaut shoulder injury including more recent data, another study was performed using NBL database information from 1995 through 2011 (Scheuring et al., 2012). This new data showed that 23 astronauts have had shoulder surgery, two of whom required surgery

21 on both shoulders. Of these cases, 12 astronauts claimed that the direct cause was related to training in the planar HUT. Of these 12, 50% of cases occurred after 2004. No subjects who trained only in the pivoted hut attributed their shoulder surgery to the suit, while almost 67% of those who trained only in the planar HUT did attribute their need for surgery to the suit. Additionally, astronauts who performed at least 5 EVAs were more than 2 times as likely to have shoulder surgery compared to those who only performed one EVA. In terms of training, astronauts with more than 92 training runs were almost 6 times more likely to have shoulder surgery than those with 0-9 runs. Amongst a number of conclusions, this study showed that despite the implementation of the Tiger Team recommendations, the incidence of shoulder injuries and shoulder surgeries has remained an issue.

A study published in 2014 was the first to quantitatively evaluate the causes of astronaut shoulder injury (Anderson, 2014; Anderson et al., 2015). Using a database of 278 astronauts that included anthropometric measurements, training record, and injury record, this study used statistical methods to determine which variables contribute to an increased risk of astronaut injury. Both for subjects whose injuries were directly attributable to EVA training and for those whose shoulder symptoms began during their mission, it was found that percent of training performed in the planar HUT was the strongest predictor variable for injury, while training frequency and recovery between sessions were also important variables. Looking specifically at subject anthropometry, it was found that the most relevant measurements for predicting injury were bi- deltoid breadth, expanded chest depth, and shoulder circumference. While the statistical models presented in this paper do not accurately identify every astronaut that may experience shoulder injury, it is a large improvement over the previous qualitative evaluations of the issue.

2.4 COUNTERMEASURES FOR SHOULDER INJURY At the time of the Tiger Team investigation, the primary countermeasures used to address shoulder injury in the NBL include padding and a shoulder harness for the HUT. While these had been available for many years, astronauts were not necessarily aware of their existence (Williams and Johnson, 2003). The two most commonly used pads, shown in Figure 2.7, are primarily for protection from the scye bearing and HUT shell. The rarely used shoulder harness, shown in Figure 2.6, is essentially a pair of suspenders inside the HUT that have a pad assembly at the shoulders meant to absorb the load of the suit at the point of contact. In 2003, results of tests run in the NBL to evaluate these countermeasures suggested that their combination could reduce the load of the scye bearing joint on the astronaut’s shoulders, but that no single FIGURE 2.5: DIAGRAM OF THE SHOULDER configuration would work for every subject. HARNESS INSIDE THE PLANAR HUT. (IMAGE SOURCE: WILLIAMS AND JOHNSON, 2003) While countermeasures and operational adjustments were an important step toward preventing shoulder injury, the general understanding of how the human interacts with the spacesuit still remained primarily qualitative, and suit fit remained somewhat of an art. Thus while the above-mentioned investigations provided valuable insights

22 for the potential for injury in the EMU, not all of these lessons would be translatable to future suit designs. There still remained a need for a quantitative understanding of human-spacesuit interaction that would involve methods transferrable to future suit designs.

FIGURE 2.7: MOST COMMONLY USED REMOVABLE SHOULDER PADDING IN THE EMU. (IMAGE SOURCE: STRAUSS ET AL., 2005)

2.5 DEVELOPMENT OF A QUANTITATIVE UNDERSTANDING OF HUMAN-SPACESUIT INTERACTION In November 2011, NASA Grant NNX12AC09G, “Spacesuit Trauma Countermeasure System for Intravehicular and Extravehicular Activities” was created, and a team of researchers at the Massachusetts Institute of Technology (MIT) along with collaborators at Trotti and Associates, Inc. (Cambridge, MA) began to work on a variety of related projects. Over the past three years, this research has addressed several main objectives: 1) analyzing data for correlations between anthropometry, space suit components, and injury, 2) modeling human-spacesuit interaction, 3) designing and developing modular protective devices to mitigate injury, and 4) quantifying and evaluating human-spacesuit interaction using a suite of sensors (Newman et al., 2014).

The first objective was addressed by performing a meta-analysis investigating these injury trends, proposing an injury classification system, and creating predictive statistical shoulder injury models (Anderson, 2014; Anderson et al., 2015), described previously in Section 2.3. The second objective—modeling human-spacesuit interaction—has been addressed for the lower body by performing a biomechanical analysis using OpenSim (Stanford, CA) in which motion capture data and EMU joint torque data were used to understand the effect of the space suit on muscle activation and force generation in the knee (Diaz and Newman, 2014). The third objective was addressed by the development of injury protection concepts that evolved into ergonomic, engineered protective device prototypes. A number of materials were evaluated for their offloading capabilities, and both passive and inflatable prototypes were created (Anderson and Diaz et al., 2012). The fourth and final objective of evaluating human-spacesuit interaction has been partially addressed. Once it was established that a pressure sensing capability would allow for quantification of the interaction between a suited subject and the spacesuit, a pressure sensing system was built to evaluate pressures over the arm, and an additional commercially produced pressure sensor was purchased for evaluation of pressures at the shoulder. Using these two systems in conjunction with kinematic inertial measurement units, a human subjects experiment was performed at David Clark Company and at NASA Johnson Space Center. Analysis of results

23 from the in-house pressure sensing system are discussed elsewhere (Anderson, 2014) and a portion of results from the kinematic data has also been published (Bertrand et al., 2014). The results presented in the following chapters of this thesis are the first publication of the quantitative analysis of the human-suit interaction at the shoulder.

In summary, given the history of astronaut injury in the spacesuit, it has become increasingly apparent that quantitative techniques are needed to develop a more complete understanding of human-spacesuit interaction. Numerous studies have reported both the incidence and mechanisms of injury, yet very few have attempted to use quantitative techniques to assess the human-suit interface in such a way that could influence suit fit techniques or even future suit design. The following chapters present two different approaches for quantitatively evaluating the effect of the spacesuit on the human: 1) quantifying and analyzing the suit-induced pressures that arise in the shoulder region, and 2) using a musculoskeletal model to compare the muscle forces that arise in the unsuited and suited conditions. Each of these methods and their subsequent results are presented in Chapters 3 and 4.

HUMAN-SPACESUIT INTERACTION QUANTIFICATION

3.1 OVERVIEW The work presented in this chapter addresses the first specific aim of quantifying and analyzing the pressure distributions and pressure profiles that arise at the interface between the human shoulder and the torso of the spacesuit. We explore whether anthropometric regions of concern vary across subjects, whether experience in the suit have an effect on the pressure magnitudes that arise, whether subjects are consistent among repetitions of the same motion, and whether anthropometric region or motion type have an effect on the pressure magnitudes experienced.

A variety of analytical methods, described in more detail in the sections that follow, are used to address this specific aim. Analyzing the pressure distributions aids in determining which areas of the shoulder are experiencing the highest pressures and are potentially more susceptible to injury, as well as which areas are experiencing the lowest pressures. Analyzing the pressure profiles provides information on how the pressures vary over the course of a particular movement, while also allowing us to determine if there is any time effect present. Analysis of pressure profiles can also show how the responses vary between different locations on the sensor. Finally, statistical analyses are used to explore subjects’ consistency of movement and whether or not there is any effect of motion type or anthropometric region on the pressure magnitudes that arise at the suit-shoulder interface.

3.2 EXPERIMENTAL METHODS

3.2.1 SENSOR SYSTEMS As mentioned previously, it was determined that pressure sensing capabilities and kinematic sensors would provide an informative characterization of the human-suit interface. As a result, a suite of sensors were selected and developed for use in a human subject experiment. The custom-built pressure sensing system placed over the arm was designed to target a low pressure-sensing regime (0-100kPa) whereas the commercially purchased pressure sensor (Novel GmbH, Munich, Germany) placed over the shoulder was specified for a higher pressure range (20- 600kPa). As shown in Figure 3.1, the two FIGURE 3.1: IN-SUIT SENSOR SYSTEMS AS PLACED ON THE pressure sensing systems were integrated into SUBJECT. (IMAGE SOURCE: ANDERSON, 2014) a conformal garment while the inertial measurement units were secured directly to the subject’s body. Images of each of the separate hardware systems are shown below in Figure 3.2.

Low Pressure High Pressure Novel APDM Inertial “Polipo” sensors sensor and hardware Measurement Unit

A) B) C)

FIGURE 3.2: EXPERIMENTAL SENSOR SYSTEMS. A) LOW-PRESSURE POLIPO SENSORS. B) HIGH-PRESSURE NOVEL SENSOR FOR THE SHOULDER-HUT INTERFACE. C) APDM OPAL INERTIAL MEASUREMENT UNIT FOR MEASUREMENT OF SUBJECT KINEMATICS. (IMAGE SOURCE: ANDERSON, 2014)

As the following analysis and discussion will focus on the results from the high pressure sensor placed at the shoulder interface, it is important to focus on an understanding of this system.

3.2.2 HIGH-PRESSURE SHOULDER SENSOR The technology used to quantify the human-spacesuit interaction at the shoulder is the Pliance® sensing system developed by Novel GmbH, a German company that specializes in dynamic pressure distribution measurement technology. The Pliance® system is used for accurate measurement of pressure and load distribution on both hard and soft surfaces (Pliance Sensing

System, 2013). The system was connected to a flexible, elastic sensor made from capacitive transducers with high-tech elastomers, which are calibrated through pre-determined loading sequences so as to create a baseline for future measurements, guarantee accuracy, and generate reproducible data (Pliance Sensing System, 2013). Relevant aspects of the Pliance® software include acquisition and storage of pressure distribution data, viewing absolute pressure values in each sensor, playback of measurements, and viewing maximum pressure, force and contact area.

This sensor was deemed appropriate for use in this application since a similar Novel pressure sensing mat was used in unpublished studies on the Extravehicular Mobility Unit (EMU) hard upper torso by the Anthropometry and Biomechanics Facility (ABF) at NASA Johnson Space Center. The particular sensor used in our experiment is a modified S2073 sensor mat with 128 individual sensiles arranged in a grid of 16 by 8. The entire sensor mat is approximately 22.4 cm x 11.2 cm. Each sensile is 1.4 cm in length and width and measures in the pressure range of 20- 600 kPa. The Pliance® system uses ten 1.2 V nickel metal hydride batteries with 2000 mAh, and the sensor is run at 330 mA. While the data collection rate can be adjusted, for the purposes of our experiment, measurements were recorded at a rate of 50 Hz (once every 0.02 s). Appendix B describes how to use the on-board storage capability of the Novel Pliance® system, as was implemented in this experiment.

In order to keep the Novel sensor mat in place during the experiment, subjects wore a conformal shirt garment with a rectangular pocket interface into which the Novel sensor mat was inserted. An additional cover shirt was worn over this as well, although its function was primarily to keep the low-pressure sensing system in place.

3.2.3 SENSOR CALIBRATION Prior to performing the experiment, the Novel sensor was calibrated to ensure accurate data collection during the official measurement trials. The calibration device used (shown below in Figure 3.3) was also provided by Novel GmbH and was developed specifically for use with the

FIGURE 3.3: CALIBRATION DEVICE. NOVEL PRESSURE SENSING MAT WAS LOADED WITH VARIOUS PRESSURE LEVELS, INCREMENTED THROUGHOUT ITS SPECIFIED PRESSURE RANGE. (IMAGE SOURCE: PLIANCE SENSING SYSTEM, 2013)

Pliance® sensing system. It consists of an inflatable rubber bladder that is housed by secure rigid plates. The sensor being calibrated is placed on the calibration board and centered within the apparatus. Compressed air is then fed into the device, thereby exerting pressure on the inserted sensor mat. By following the calibration steps provided in the Novel software, the sensor is loaded at various known pressures, and the resulting calibration curves are created within the software. These files are then stored such that subsequent testing can make use of the calibration.

3.2.4 EXPERIMENTAL DESIGN This experiment was performed using three subjects in the Mark III planetary exploration suit. The suit was pressurized to 4.3 psi and used a liquid cooling system. Each subject wore different sized suit parts based on their anthropometry, and subjects were allowed to use various padding and comfort aids to ensure optimal suit fit and comfort. Of particular note for this study is that all subjects used the built-in shoulder pads on the shoulder harness of the Mark III (see Figure 3.8). The three subjects were right-handed males with extensive experienced working in the pressurized suit.

FIGURE 3.4: MOVEMENT TASKS PERFORMED BY EACH SUBJECT. THREE ISOLATED JOINT MOVEMENTS AND TWO FUNCTIONAL TASKS WERE PERFORMED BY EACH OF THE THREE SUBJECTS, IN THE SUITED AND UNSUITED CONDITIONS. (IMAGE SOURCE: ANDERSON, 2014)

Each subject performed a series of upper-body motions aimed at characterizing this human-suit interaction. Three isolated joint movements were evaluated: elbow flexion/extension, shoulder flexion/extension, and shoulder abduction/adduction. These motions were chosen as they are standard one-joint motions that are typically found in biomechanics literature. However, in order to assess motions that are more similar to those that an astronaut would encounter during extravehicular activity, two additional functional tasks were evaluated: a cross-body reach and an overhead hammering task. These five motions are depicted and described in Figure 3.4.

With regards to the experimental procedure, each of the five motions was repeated 12 times, and these repetitions were divided into three groups of four repetitions to allow for assessment of fatigue or changes in biomechanical strategies. Motions were divided into movement groups such that the order was counterbalanced within the group and randomized between subjects (Anderson et al., 2014). During the experiment, subjects were first trained on each of the movements while wearing the spacesuit. They then proceeded to perform the motions in the prescribed order, with 5-minute breaks between each movement group to allow for the collection of subjective feedback and to allow the subject to rest. After all movement groups were completed, subjects doffed the suit and were then asked to perform the same five motions unsuited, with only four repetitions each, attempting to mimic the same speed and range of motion as when they were suited. Throughout this entire experiment, pressure profiles and joint angles were recorded. An example of one representative test procedure is shown below in Figure 3.5. For a more detailed explanation of the experimental design, refer to Anderson et al., 2014. Appendix A also includes the full experiment test plan described above.

FIGURE 3.5: EXAMPLE TEST PROTOCOL FOR ONE SUBJECT. EACH SUBJECT UNDERWENT A TRAINING SESSION, FOLLOWED BY THREE MOVEMENT GROUPS CONSISTING OF A COUNTERBALANCED ORDERING OF THE MOVEMENTS. ORDER WAS ALSO RANDOMIZED BETWEEN SUBJECTS. SUBJECTIVE DATA WAS COLLECTED IN BEFORE AND AFTER EACH OF THE MOVEMENT GROUPS. (IMAGE SOURCE: ANDERSON, 2014)

3.2.5 SENSOR ORIENTATION Before presenting results, it is important to understand how the Novel pressure sensor was oriented with respect to the subjects’ shoulders. For a general understanding of the sensor’s placement, refer to the diagram in Figure 3.6. This diagram shows that what we will refer to as the lower portion of the sensor (the blue region) corresponds to the anthropometric region toward the clavicle and front of the body, whereas the upper portion (the green region) overlays the back of the shoulder, toward the shoulder blade. The colors depicted are solely for aiding in

29 understanding the orientation of the sensor, and are in no way related to the color scheme of any pressure distribution maps that follow.

FIGURE 3.6: ORIENTATION OF NOVEL SENSOR. THIS ORIENTATION CORRESPONDS TO THE INFORMATION IN EACH PRESSURE DISTRIBUTION FIGURE. COLORING IS FOR ORIENTATION AND IS NOT RELATED TO PRESSURE SCALES.

For a subject-specific diagram of how the sensor was oriented with respect to subjects’ anthropometric landmarks, see Figure 3.7. As labeled in the image, the depicted bony landmarks include the acromion and a guiding point along the clavicle, which allowed us to approximate the orientation of the clavicle. This data was recorded after the suit was doffed. From Figure 3.7, we also see that the sensor placement varied substantially between subjects, which follows logically from the fact that the three subjects had very different anthropometry. These individual differences in sensor placement are taken into account in the interpretation of results.

FIGURE 3.7: SUBJECT-SPECIFIC ORIENTATION OF NOVEL SENSOR RELATIVE TO ANTHROPOMETRIC LANDMARKS.

3.2.6 LIMITATIONS

There are a few limitations associated with the methods by which the shoulder pressure data was collected. First of all, all three subjects used the built-in shoulder pads of the Mark III. While this affects the interface between the hard upper torso and the subject’s shoulder, its use is not a significant concern since it still accurately represents the shoulder loading an astronaut would experience during EVA. However, what may be a limitation is that we have no way of identifying a shift in the shoulder pad placement. Out of our three subjects, only one had a visibly significant change in shoulder pad placement between donning and doffing the suit. While we have reason to believe that this shift likely occurred during additional functional tasks that were performed after the official experimental protocol and therefore did not affect the relevant data, there is no way to know for sure.

However, randomization of the motions prevents a shoulder pad shift from falsely suggesting FIGURE 3.8: IN-SUIT HARNESS OF THE MARK differences in pressure between the motions. III WITH ATTACHED SHOULDER PADS.

A second potential limitation is that the location of anthropometric landmarks with respect to the sensor mat was only recorded after the suit was doffed; thus it is possible that the sensor shifted slightly from its initial position at the start of the experiment. However, as the sensor was placed in a pocket of the conformal garment that matched its exact size (no room to shift), it is unlikely that the sensor shifted significantly with respect to the subject’s bony landmarks.

3.3 GRAPHICAL AND NUMERICAL ANALYSIS In the following sections we present a variety of analysis intended to provide a holistic understanding of the human-suit shoulder interface. First, graphical and numerical analyses are presented to instill a visual and conceptual understanding of the pressure distributions and profiles.

3.3.1 METHODS In the graphical and numerical analysis of the measured pressure values in the shoulder region, we present two main categories of data: 1) the overall pressure distributions and 2) the pressure profiles seen in each of the motions. Analyzing the pressure distribution aids in determining which areas of the shoulder are experiencing the highest pressures during a series of upper body motions and are therefore more susceptible to injury. It also provides a visual understanding of what is happening at this interface. Analyzing the pressure profiles provides information on how the pressures vary over the course of a particular movement, while also allowing us to determine if there is any time effect present (steadily increasing or decreasing pressure profiles within a set of 4 repetitions could suggest a movement strategy adjustment or movement inconsistency). Analysis of pressure profiles can also show us how the responses vary between different locations

31 on the sensor. Appendix D refers the reader to the MATLAB code used to process the data and perform these analyses.

3.3.2 RESULTS It should be noted that for all of the following results, data from the elbow flexion/extension motion is excluded as it was deemed less relevant to the shoulder human-suit interface.

3.3.2.1 PRESSURE DISTRIBUTIONS

We first consider the pressure distribution maps shown below in Figure 3.9. For each of the motions, the depicted pressure map represents the pressure distribution at the “peak” of the movement. In other words, it shows how pressure is distributed over the shoulder region at the moment in the motion when the highest pressure appeared. This point in the motion was chosen for depiction since the main concern is the potential for injury, which is most likely to occur where localized peak pressures occur. Note that all subjects’ data are depicted with the same measurement scale.

FIGURE 3.9: PRESSURE DISTRIBUTIONS OVER THE SHOULDER AT THE POINT OF MAXIMUM TOTAL PRESSURE. ELBOW FLEXION NOT SHOWN.

Looking first at Subject 1’s measurements, we see that shoulder flexion/extension has the highest overall pressures, with a peak pressure of over 100 kPa and a relatively large area that experiences pressures over 40 kPa. Pressure is concentrated on the proximal side of the sensor, closest to the subject’s neck, and the peak values appear near the upper edge of the sensor, toward the subject’s shoulder blade. In the case of shoulder abduction/adduction, we see that while the pressure values are slightly lower (peak of ~90 kPa), pressure is now concentrated in a region centered just above the clavicle, likely coinciding with a number of shoulder muscles. In the crossbody reach motion, the peak pressure is again slightly lower (~85 kPa), but the location and shape of the pressure distribution is nearly identical to that of shoulder abduction/adduction. In the overhead hammering task, the location and shape of the pressure distribution are again similar to those of the previous two motions. For this task, however, the distal edge of the sensor is much less loaded, and the pressure magnitudes are lower than in the other motions (~75 kPa).

Moving now to Subject 2’s measurements, we see that in shoulder flexion/extension, pressure is concentrated along a line just above the clavicle, again likely over soft musculature near the top of the shoulder. This area of pressure concentration (peak of ~75 kPa) is not widely distributed, but a secondary area of pressure concentration seems to appear at the most distal end of the lower edge of the sensor. In the case of shoulder abduction/adduction, only the most proximal third of the sensor is loaded (the edge closest to the neck), with the peak pressure (~60 kPa) arising toward the shoulder blade. For the crossbody reach motion, again only the proximal portion of the sensor is loaded; however, the peak pressure (~70 kPa) arises near the top of the shoulder. Finally, for the overhead hammering task, the pressure distribution looks similar to that of shoulder flexion/extension in that pressure is concentrated along a line just above the clavicle. However the peak pressure for this task (~80 kPa, which is also the maximum pressure seen across all movements for this subject) is seen in a secondary area of pressure concentration, again located at the most distal end of the lower edge of the sensor.

Looking at Subject 3’s measurements, we see that shoulder flexion/extension gives rise to the broadest distribution of pressure of all of the movements. Pressure is mainly concentrated on the proximal half of the sensor and toward the shoulder blade; however, the location that experiences the peak pressure (~80 kPa) is along the lowest edge of the sensor. In the case of shoulder abduction/adduction, pressure is concentrated in the center of the sensor (peak of ~55 kPa), corresponding to the muscular region on top of the shoulder. In the crossbody reach motion, the same region is loaded (with a slight shift toward the neck), this time with a slightly lower peak pressure (~45 kPa). In the overhead hammering task, pressure is still concentrated centrally, although it is shifted slightly toward the distal end of the shoulder (with a peak of ~45 kPa).

Comparing subjects, we see that Subject 1 experiences higher pressures than Subjects 2 or 3 in all of the movements except the overhead hammering task, for which Subject 2 has a slightly higher peak pressure magnitude. On the other hand, Subject 3 experiences the lowest pressures for all of the movements. While Subjects 1 and 3 experience their highest pressure in shoulder flexion/extension, Subject 2 has the highest pressure in overhead hammering. Additionally, it is interesting to note that for each subject, pressure was concentrated in a consistent location for 3 of the 4 movements: just above the clavicle, over the soft musculature at the top of the shoulder.

3.3.2.2 MAXIMUM PRESSURE LOCATIONS

The pressure distribution maps being compared in the previous section are only representative “snapshots” of each motion for each subject. It is possible that observations and comparisons would be different if we were to compare the distributions of the same motions but taken from different movement groups. For this reason, the diagram below is provided to show the locations of peak pressure throughout the entirety of the experiment. In other words, Figure 3.10 shows every location that experiences the maximum pressure at any moment in the movement, across all three motion groups. In these diagrams, the colors indicate frequency: green spots experience the maximum pressure for 0-5% of the movement duration, yellow spots for 5-10%, and red spots for more than 10% of the movement duration. Thus while all of the colored spots indicate a location on the shoulder that experiences the maximum pressure at some moment in the motion, we are most concerned with the red spots, since they are experiencing the maximum pressure the most often (which would make these locations more susceptible to injury). Note that the diagram for Subject 2’s shoulder abduction/adduction movements is not included, as it was impossible to identify the beginning and end of each set of movement repetitions.

Looking first at Subject 1, we see that during all of the movements, the maximum pressures are generally located in a few main regions of the shoulder: 1) a concentrated area near the shoulder blade on the proximal side of the sensor, 2) a broader region over the clavicle and extending upward to the top of the shoulder, and 3) scattered locations below the clavicle. In shoulder flexion/extension, all three of these regions see maximum pressures with high frequency (denoted by red). In both shoulder abduction/adduction and crossbody reach, only the region over and above the clavicle experiences maximum pressures with high frequency. However, in the case of shoulder abduction/adduction, the other two regions still see maximum pressure with mid-level frequency (denoted by yellow). In the crossbody reach, only the shoulder blade region experiences maximum pressure with mid-level frequency. Finally, in the overhead hammering task, both the shoulder blade and clavicle regions experience maximum pressures with high frequency. Comparing across movements, it appears that shoulder flexion/extension has the most scatter in locations of maximum pressure.

Looking now at Subject 2, we see that in all three of the included motions, the maximum pressures are located in two general regions of the shoulder: 1) a broad region over the top of the shoulder, on the proximal half of the sensor, and 2) a concentrated area at the bottom left corner of the sensor, corresponding to a region below the clavicle. In shoulder flexion/extension, both of these regions see maximum pressures with high frequency, and there is an additional region on the lowest fourth of the proximal side of the sensor (partially over the clavicle) that experiences maximum pressure with low frequency. In crossbody reach, only the region over the top of the shoulder sees maximum pressure with high frequency, but the region below the clavicle sees maximum pressure with mid-frequency. In overhead hammering, maximum pressures arise with high frequency in both of the main regions described above. Comparing across movements, overhead hammering again has the smallest amount of scatter.

For Subject 3, we see that in all of the motions, the maximum pressures are concentrated in the central area of the sensor, which corresponds to the region over the top of the shoulder. In addition, for all of the motions except overhead hammering, there is a secondary region further toward the shoulder blade on the proximal edge of the sensor that experiences maximum

34 pressure with mid-to-high frequency. For this subject, all motions appear to have the same relatively low amount of scatter.

FIGURE 3.10: LOCATIONS OF THE SHOULDER THAT EXPERIENCE MAXIMUM PRESSURE AT SOME POINT IN THE MOVEMENT. ELBOW FLEXION NOT SHOWN. SHOULDER ABDUCTION/ADDUCTION FOR SUBJECT 2 EXCLUDED. COLORS INDICATE RELATIVE FREQUENCY AT WHICH EACH LOCATION EXPERIENCES MAXIMUM PRESSURE.

Comparing across subjects, we see that all subjects have the top of the shoulder as one of the regions in which maximum pressure is located the most frequently. Additionally, Subjects 1 and 3 usually have a secondary region toward the shoulder blade that experiences maximum pressure with mid-to-high frequency.

3.3.2.3. PRESSURE PROFILES

SUITED PRESSURIZED DATA

We now consider pressure profiles, shown below in Figures 3.11 – 3.13 for Subjects 1-3, respectively. For each motion, two or three selected pressure profiles are plotted for each of the three movement groups. The profiles plotted for each case are the highest magnitude profiles of each general trend of sensor response. In other words, since many sensors “move together”— meaning they peak and flatten out at the same time—we have chosen to show only one profile for each of these general trends (the one with the highest magnitude). To illustrate the locations of the individual sensiles that correspond to the chosen pressure profiles, corresponding sensor diagrams are depicted to the right of each plot. It is important to note that profiles seen in one sensile are not necessarily similar to those of neighboring sensiles. Therefore it cannot be assumed that sensiles in a particular region exhibit similar responses.

FIGURE 3.11: SUBJECT 1: PRESSURE PROFILES (KPA) PLOTTED FOR SELECTED SENSORS. LOCATIONS OF SELECTED SENSORS ARE INDICATED IN CORRESPONDING DIAGRAMS. PROFILES PLOTTED FOR EACH CASE ARE THE HIGHEST MAGNITUDE PROFILES OF EACH GENERAL TREND OF SENSOR RESPONSE.

Note that all of the plots have the same scales, the y-axis being pressure in kPa and the x-axis being a normalized time axis. Normalizing the x-axis and plotting each of the profiles on the same scale allows for easier comparison between the different cases. For Subject 2, shoulder abduction/adduction was again excluded since in most repetitions it was impossible to identify the profile of the movement.

Starting with Subject 1, we first make a number of observations for shoulder flexion/extension. While the shapes of general profile trends are relatively consistent between the first two movement groups, these differ from the trends seen in the third movement group. However, the profile shapes typically remain consistent for each of the 4 repetitions within one movement group. Additionally, the highest peaks (75-95 kPa) are reached at a point near the shoulder blade, whereas the peaks that arise at other instances in the motion (30-60 kPa) are located along and below the clavicle. In shoulder abduction/adduction, most of the profile shapes are different between the 3 movement groups. However, the profile shapes again remain consistent within each set of 4 repetitions, as do the peak pressures reached (70-80 kPa). The region that sees the peak pressures is consistently along or slightly above the clavicle, whereas the peaks that arise at other instances in the motion (30-60 kPa) are located either near the shoulder blade or below the clavicle. For the crossbody reach, the profile shapes are relatively consistent across the 3 movement groups and within each set of 4 repetitions. The highest peaks (60-80 kPa) are experienced along and slightly above the clavicle, as are the lower magnitude peaks (30-60 kPa). Finally for the overhead hammering task, the general profile trends are a bit more varied between movement groups, and we also see some differences in the profile shapes of individual repetitions within the same set. The highest peaks (50-70 kPa) are seen along and slightly above the clavicle. In this case, however, all of the general profile trends rise and fall at approximately the same times. Thus overall for Subject 1, we can make a few general observations: for 3 of the 4 motions, 1) general profile trends varied in shape, 2) each of the 4 repetitions within a movement group had visually consistent shape and peak pressure magnitudes, and 3) the highest pressures were recorded along and slightly above the clavicle.

Now looking at Subject 2 (below in Figure 3.12), we see for shoulder flexion/extension that the shapes of the general profile trends vary substantially across the 3 movement groups. The shape and magnitudes of the 4 repetitions within each movement group are not consistent either. Even the location of the highest peaks varies between movement groups, as it on top of the shoulder for Movement Groups 1 and 3, and both below the clavicle and at the shoulder blade for Movement Group 2. In general, peak pressures range from 30-70 kPa. For crossbody reach, most profile shapes are not consistent across movement groups, although they are relatively consistent in shape within each set of 4 repetitions. The highest peaks (40-60 kPa) are experienced both on top of the shoulder and below the clavicle. Finally for overhead hammering, the general profile shapes vary between movement groups, but are consistent within each movement group. In this case, all of the locations of interest are at the top of the shoulder or below the clavicle, regardless of whether they experience the highest peaks (30-70 kPa) or the lower peaks (20-40 kPa). Overall for Subject 2, we can say that: 1) for all 3 of the included motions, general profile trends varied between movement groups, 2) for 2 of the 3 motions, profiles were visually consistent in shape— but not in magnitude—within each set of 4 repetitions, and 3) in all 3 motions, regions on the top of the shoulder and below the clavicle experience the highest pressures.

FIGURE 3.12: SUBJECT 2: PRESSURE PROFILES (KPA) PLOTTED FOR SELECTED SENSORS. LOCATIONS OF SELECTED SENSORS ARE INDICATED IN CORRESPONDING DIAGRAMS. PROFILES PLOTTED FOR EACH CASE ARE THE HIGHEST MAGNITUDE PROFILES OF EACH GENERAL TREND OF SENSOR RESPONSE.

Now considering Subject 3, we see for shoulder flexion/extension that the shapes of the general profile trends are consistent in Movement Groups 2 and 3, but differ for Movement Group 1. However, within each movement group, the shape of the pressure profiles remains consistent across all 4 repetitions, and the magnitudes remain visually consistent for Movement Groups 2 and 3 but not for all cases in Movement Group 1. The location experiencing the highest pressures (45-70 kPa) is nearly always by the shoulder blade, although in the case of Movement Group 1, similar magnitudes of pressure were reached in 3 separate regions: near the shoulder blade, on top of the shoulder, and below the clavicle. The top of the shoulder appears to be a common place for other general profile trends to peak (30-40 kPa). In shoulder abduction/adduction, general profile trend shapes are visually consistent across movement groups, as are the majority of the repetitions within movement groups. The highest peaks (40-50 kPa) are seen consistently in the region on top of the shoulder, whereas other general profile trends (with peaks at 20-40 kPa) are found either on top of the shoulder or toward the shoulder blade. For the crossbody reach, all measured pressures are below ~35 kPa and located either over the top of the shoulder or at the shoulder blade. General profile shapes appear relatively consistent in both shape and magnitude. Finally, for overhead hammering, profile shapes are somewhat consistent across movement groups, although a few differ. Repetitions within a movement group are visually consistent in both shape and magnitude. All general profile trends are located in the top of the shoulder region, with peaks ranging from ~10-35 kPa. Thus overall for Subject 3, we can state that 1) for all motions, general profile trends appear relatively consistent in shape across movement groups, 2)

38 for all motions, most repetitions within a movement group were visually consistent in shape and magnitude, 3) for all motions, the highest pressures are sometimes experienced at the top of the shoulder, 4) for 3 of the 4 motions, the shoulder blade region is experiencing the peak pressure at some instance in the movement.

FIGURE 3.13: SUBJECT 3: PRESSURE PROFILES (KPA) PLOTTED FOR SELECTED SENSORS. LOCATIONS OF SELECTED SENSORS ARE INDICATED IN CORRESPONDING DIAGRAMS. PROFILES PLOTTED FOR EACH CASE ARE THE HIGHEST MAGNITUDE PROFILES OF EACH GENERAL TREND OF SENSOR RESPONSE.

Comparing across subjects, we can summarize by stating that 1) for 2 subjects, general profile trends varied in shape across movement groups, 2) for all 3 subjects, repetitions within each movement group were consistent in shape, 3) for 2 subjects, repetitions within each movement group were consistent in magnitude, 4) for all 3 subjects, the highest pressures were—at least sometimes—experienced near the top of the shoulder, and 5) for 1 subject, the shoulder blade was a secondary region of concern.

The following table provides a summary of the peak shoulder pressures seen for each of the motions of each subject. Values are in kPa (mean + SD).

TABLE 3.1: PEAK PRESSURES OF EACH MOTION FOR THE THREE SUBJECTS, IN KILOPASCALS (KPA). LISTED AS MEAN + SD. ALL VALUES FOR SUITED PRESSURIZED CONDITION. Sh. Flex./Ext. Sh. Abd./Add. Crossbody Hammering Subject 1 85.4 + 5.0 75.1 + 5.4 75.1 + 5.4 54.7 + 5.7 Subject 2 49.4 + 7.8 N/A 49.2 + 5.0 56.5 + 12.4 Subject 3 59.5 + 6.2 43.6 + 3.7 25.2 + 5.0 33.0 + 2.8

UNSUITED DATA CONSIDERATION

Ideally, the subjects’ unsuited motions would perfectly mimic those performed inside the pressurized suit. In this way, we could subtract any unsuited motion artifacts from the suited profiles and obtain the pressure profiles that are attributable to the suit itself. Despite having subjects perform the unsuited motions with a pace and range of motion similar to the suited condition, we see from the data that the motions were not necessarily replicated. In a few cases, the unsuited profiles reach higher pressures than the suited profiles, suggesting that the motion was being performed differently. Thus, there is no utility currently in directly comparing suited and unsuited data. However, as future work will integrate pressure data with kinematic data, determining the pressure-angle relationship for each motion will allow for determination of how the suited and unsuited pressures compare for particular arm angles.

While we cannot directly compare suited and unsuited data, it is still interesting to consider the unsuited pressure profiles on their own, shown in Figure 3.14 for all subjects. In this case only the maximum pressure profile is plotted. As before, all of the plots have the same scales, where the y-axis is pressure in kPa and the x-axis is a normalized time axis.

Examining these profiles graphically, it appears that for nearly all of the movements of each subject, the peak pressures reached by subjects were consistent in the 4 repetitions. The only possible exception is Subject 1’s shoulder abduction/adduction motion, which has peaks varying between 55-75 kPa.

FIGURE 3.14: MAXIMUM PRESSURE PROFILE (KPA) OF THE UNSUITED CONDITION. PROFILES PLOTTED FOR EACH CASE ARE THE HIGHEST MAGNITUDE PROFILES REGARDLESS OF LOCATION ON THE SENSOR MAT.

3.3.3 DISCUSSION In the pressure distribution portion of the graphical and numerical results presented in Section 3.3.2.1, it was established that Subject 1 experiences higher pressures than Subjects 2 or 3 in all of the movements except the overhead hammering task, for which Subject 2 has a slightly higher peak pressure magnitude. As Subject 1 was the least experienced of the 3 subjects, the higher pressure readings are likely a reflection of this relative inexperience in the suit. Subject 1 may not yet have determined his optimal suit fit, and could also still be learning to work with the suit as opposed to fighting against it. On the other hand, Subject 3 experienced the lowest pressures for all of the movements. While Subjects 1 and 3 experienced their highest pressure in shoulder flexion/extension, Subject 2 had the highest pressure in overhead hammering. This could be a result of the difference in placement of the sensor on each of the subjects’ bodies, a difference in

41 suit fit, or a difference in biomechanical movement strategy. Additionally, it was noted that for each subject, pressure was concentrated in a consistent location for 3 of the 4 movements: just above the clavicle, over the soft musculature at the top of the shoulder. This points to this musculature area as being a region of particular concern.

In the examination of anthropometric locations that experience maximum pressure, it was observed that for Subject 1, shoulder flexion/extension exhibited the most scatter in locations of maximum pressure. This could be a result of one of two issues: 1) the subject could be performing the movement consistently, but the location of maximum pressure shifts over the course of one individual movement, or 2) the subject changed his motion strategy across repetitions, therefore affecting where the pressures were localized. As these diagrams combine the subject’s data across all of the 12 repetitions, we cannot differentiate between these two possibilities. Conversely, the small amount of scatter in overhead hammering suggests that either 1) the subject was extremely consistent in how the movement was performed, or 2) the movement itself does not cause the location of maximum pressure to shift over the course of an individual movement. Similarly for Subject 2, overhead hammering again had the smallest amount of scatter. This could be attributed to either of the same two options suggested for Subject 1. Comparing across subjects, it was observed that all subjects have the top of the shoulder as one of the regions in which maximum pressure is located the most frequently. Additionally, Subjects 1 and 3 usually have a secondary region toward the shoulder blade that experiences maximum pressure with mid-to-high frequency. As suggested previously, differences between subjects could result from a slight difference in sensor placement with respect to anthropometric landmarks, a difference in suit fit, or a difference in movement strategy.

3.4 STATISTICAL ANALYSIS: EFFECT OF MOTION AND ANTHROPOMETRIC REGION ON PEAK PRESSURE MAGNITUDES While graphical and numerical analysis presents us with a holistic understanding of the data, we can further our comprehension by performing statistical analyses. The statistical tests presented in this section evaluate the effects of motion type and anthropometric region on the pressure magnitudes that arise at the suit-shoulder interface.

3.4.1 METHODS An experimental study was made of the effects of motion type and anthropometric region on pressure magnitudes measured at the suit-shoulder interface. A two-factor ANOVA (Factor A - Motion: shoulder flexion/extension, shoulder abduction/adduction, crossbody reach, overhead hammering; Factor B - Region: clavicle, top of shoulder, shoulder blade) was run on each of the individual subjects, as inter-subject variability makes combined data unrepresentative of any particular individual. The pressure values being compared are the mean peak pressures of each combination of motion and anthropometric region (i.e. 12 mean peak pressures are being compared – 4 motions x 3 regions – each of which is the average of the 12 repetitions). As mentioned in the explanation of the experimental design, motion order was randomized and counter-balanced and no fatigue was noted across the duration of the experiment; thus we can state that for each subject, there are a total of twelve independent repetitions of each of the 12 treatments. Prior to undertaking formal inference procedures, the appropriateness of the two- factor ANOVA model was evaluated for normality, constancy of error variance, and

42 independence of error terms. No issues were found, so proceeding with the statistical analysis was deemed appropriate. For all tests, an alpha value of 0.05 was used to determine significance.

3.4.2 RESULTS For Subject 1, interaction effects (p<0.0005) and both main effects (p<0.0005 for both) were found. Post-hoc Tukey’s HSD tests at the 0.05 significance level revealed that, for the clavicle region, overhead hammering led to significantly lower pressures than the other three motions. In the top of the shoulder musculature, overhead hammering was again significantly lower than the other three motions, while shoulder flexion/extension also led to significantly higher pressures than shoulder abduction/adduction. In the shoulder blade region, shoulder flexion/extension led to significantly higher pressures than the rest of the motions, while crossbody reach was significantly higher than overhead hammering. These results are depicted in Figure 3.15. Comparing anthropometric regions within individual motions, it was found that for shoulder flexion/extension, the shoulder blade region experienced significantly higher pressures than the other two regions. For shoulder abduction/adduction, crossbody reach and overhead hammering, the clavicle experienced significantly higher pressures than the other two regions. These results are shown in Figure 3.16. Note that the two graphs in Figures 3.15 and 3.16 depict the same data, just grouped differently in order to highlight pairwise comparisons both within region and within motion. All other relevant comparisons were not significant.

FIGURE 3.15: SUBJECT 1: POST-HOC PAIRWISE COMPARISONS WITHIN ANTHROPOMETRIC REGIONS. COMPARISONS LABELED WITH A STAR INDICATE STATISTICALLY SIGNIFICANT DIFFERENCES.

FIGURE 3.16: SUBJECT 1: POST-HOC PAIRWISE COMPARISONS WITHIN MOTION TYPES. COMPARISONS LABELED WITH A STAR INDICATE STATISTICALLY SIGNIFICANT DIFFERENCES.

For Subject 2, interaction effects were not significant (p=0.07), but it was found that there were main effects of both motion type (p<0.05) and anthropometric region (p<0.0005), suggesting that some effects associated with motion type and anthropometric region do exist. Post-hoc Tukey’s HSD tests at the 0.05 significance level revealed that in the top of the shoulder region, overhead hammering led to significantly higher pressures than shoulder abduction/adduction. These results are depicted in Figure 3.17. Comparing anthropometric regions within individual motions, it was found that for shoulder flexion/extension and crossbody reach, the clavicle experienced significantly lower pressures than the top of the shoulder. For shoulder abduction/adduction, the clavicle experienced significantly lower pressures than the other two regions. For overhead hammering, the top of the shoulder experienced significantly higher pressures than the other two regions. These results are shown in Figure 3.18. Again, note that the two graphs in Figures 3.17 and 3.18 depict the same data, just grouped differently in order to highlight pairwise comparisons both within region and within motion. All other relevant comparisons were insignificant.

FIGURE 3.17: SUBJECT 2: POST-HOC PAIRWISE COMPARISONS WITHIN ANTHROPOMETRIC REGIONS. COMPARISONS LABELED WITH A STAR INDICATE STATISTICALLY SIGNIFICANT DIFFERENCES.

FIGURE 3.18: SUBJECT 2: POST-HOC PAIRWISE COMPARISONS WITHIN MOTION TYPES. COMPARISONS LABELED WITH A STAR INDICATE STATISTICALLY SIGNIFICANT DIFFERENCES.

For Subject 3, interaction effects and both main effects were found to be significant (p<0.0005 for all). Post-hoc Tukey’s HSD tests at the 0.05 significance level revealed that, in the clavicle region, shoulder flexion/extension was significantly higher than the other three motions, while crossbody reach was significantly lower than all other motions. In the top of the shoulder region, crossbody reach was again significantly lower than all other motions, and overhead hammering was significantly lower than shoulder flexion/extension and shoulder abduction/adduction. In the shoulder blade region, shoulder flexion/extension was significantly higher than any other motion, and overhead hammering was significantly lower than all other motions. These results are depicted in Figure 3.19. Comparing anthropometric regions within individual motions, it was found that for shoulder flexion/extension, the shoulder blade region was significantly higher than the other two regions, and the top of the shoulder was significantly higher than the clavicle. For shoulder abduction/adduction, the top of the shoulder was significantly higher than the other two regions. For crossbody reach, the clavicle had significantly lower pressures than the other two regions. For overhead hammering, the top of the shoulder was significantly higher than the other two regions, and the clavicle was significantly higher than the shoulder blade. These results are shown in Figure 3.20. Again, note that the two graphs in Figures 3.19 and 3.20 depict the same data, just grouped differently in order to highlight pairwise comparisons both within region and within motion. All other comparisons were insignificant.

Motion Type x Anthropometric Region: Subject 3

FIGURE 3.19: SUBJECT 3: POST-HOC PAIRWISE COMPARISONS WITHIN ANTHROPOMETRIC REGIONS. COMPARISONS LABELED WITH A STAR INDICATE STATISTICALLY SIGNIFICANT DIFFERENCES.

FIGURE 3.20: SUBJECT 3: POST-HOC PAIRWISE COMPARISONS WITHIN MOTION TYPES. COMPARISONS LABELED WITH A STAR INDICATE STATISTICALLY SIGNIFICANT DIFFERENCES.

Table 3.2 summarizes the results presented above, including general trends found for each subject from the post-hoc Tukey tests.

TABLE 3.2: SUMMARIZED RESULTS OF STATISTICAL ANALYSIS. AT LEFT: RESULTS OF TWO-FACTOR ANOVA. AT RIGHT: GENERAL TRENDS FOUND FROM POST-HOC TUKEY TESTS.

3.4.3 DISCUSSION While useful on an individualized basis, these results yield no common conclusions across subjects. This suggests that the effects of motion type and anthropometric region on pressure at the shoulder-suit interface cannot be generalized across subjects as they are likely affected by individual anthropometry, suit fit, and the biomechanics of how each subject performs the motion (particularly for the functional tasks).

Although the results yielded by these three subjects cannot be generalized, there are a number of ways in which this information could be used to decrease the risk of astronaut injury. One option would be to perform this analysis on more subjects, keeping track of subject anthropometry and suit sizing such that generalizations could eventually be made for subjects of a given size and suit fit. Generalizations regarding which regions are more susceptible to injury for different body types could then inform the need for protective devices in astronauts’ initial suit fit sessions. This would be a preliminary preventative technique to avoid injury since many inexperienced astronauts may not yet know how a proper suit fit should feel and therefore may not request sufficient in-suit protection.

In the case that generalizations cannot be made for various body types, this type of analysis could still be used on an individualized basis. For example, in the case of Subject 2, the top of the shoulder constantly experienced the highest pressures. Thus, in the 1-g environment and potentially other partial-gravity environments, Subject 2 could use protection for this region as a means of preventing possible injury.

While this analysis provides an idea of which regions may be more susceptible to injury, we cannot directly correlate a certain pressure reading with a definitive statement as to whether the subject will be injured or not. The spacesuit-induced contact injuries seen in astronauts accumulate over longer time scales than what was tested in our experiment, and typically only include minor injuries such as skin irritation, redness, or bruising of the underlying soft tissue (Williams and Johnson, 2003). Further testing could be done to develop a better understanding of the load distributions and timescales that lead to injury under different conditions.

3.5 CONCLUSIONS AND FUTURE WORK In summary, the results and discussion of the graphical and numerical analyses provide us with an “inside look” of how the Mark III spacesuit affects the pressure distributions and pressure profiles experienced at the shoulder. From the pressure distribution analysis, we came to a few general observations: 1) the least experienced subject generated the highest pressures, 2) the region just above the clavicle over the soft musculature at the top of the shoulder is of particular concern, as pressure was concentrated in this location for the majority of movements for all subjects, and it is also one of the regions in which maximum pressure is located most frequently, and 3) the top of the shoulder blade is a secondary region of concern for some subjects, as it experiences maximum pressure with mid-to-high frequency. We also made a number of detailed observations on each of the four movements for each subject and how these individual distributions are affected by the suit. From the pressure profile analysis, we determined that 1) for most subjects, general profile trends vary in shape across movement groups, 2) repetitions within each movement group are consistent in shape, and for most subjects also in magnitude, 3) the highest pressures are typically found near the top of the shoulder, and 4) the shoulder blade

48 area is of concern for at least one subject. Again we also made a number of detailed observations on the profiles of individual motions.

While the graphical and numerical analyses are quantitative in nature, the presented results are still primarily observational. For this reason, a statistical analysis was performed that considered the effects of motion type and anthropometric region on peak pressure magnitudes, showing that results cannot be generalized across subjects as they are likely affected by individual anthropometry, suit fit, and the biomechanics of how each subject performs the motion. It nevertheless provided an effective technique for evaluating each of the three subjects and determining which motions or regions of the shoulder might be most provocative of higher pressures on an individual basis.

Taking the combined results of the graphical, numerical, and statistical analyses, we can directly address the primary research questions posed in Chapter 1 of this thesis, restated here.

. Do anthropometric regions of concern vary between subjects?

This question was best addressed by the graphical and numerical analyses. The patterns that emerged with regards to the pressure distribution shapes and which regions experienced maximum pressure the most frequently certainly show very different results across subjects. However, if we consider the general shoulder regions, the primary region of concern for all three subjects was the over the soft musculature at the top of the shoulder.

. Does experience in the suit have an effect on the pressure magnitudes experienced?

Due to the small number of subjects investigated, we cannot say that experience in the suit has a statistically significant effect on the pressure magnitudes reached. However, it was found that the least experienced subject generated the highest pressures in nearly all of the motions.

. Does anthropometric region have an effect on the pressure magnitudes reached? . Does motion type have an effect on the pressure magnitudes experienced?

Both of these questions are addressed through the statistical analysis presented above. Results show that the answers depend heavily on the particular subject, due to differences in anthropometry, suit fit, and biomechanics. Some subjects may show an effect of both motion and anthropometric region, while others experience only one or the other.

As suggested previously, it is possible that these types of analyses could be used to aid in the suit fit process, whether on a generalized basis by body type or on an individualized basis. Results of such analyses could inform appropriate sizing of the HUT or padding placement to improve suit fit and reduce the risk of injury. Of course, we cannot expect all astronauts to wear padding in a particular body area just because it tends to experience higher pressures than the surrounding regions, especially since we have not quantified the pressure threshold for injury and injuries typically accumulate over longer time scales than what was tested in the previously described experiment. Neither would we want to use this in-suit sensor system to test subjects to injury in an attempt to determine that threshold. Perhaps an alternative option would be to gather experienced astronauts who have not undergone any major injury, and collect data on the

49 pressures that are measured when they are in their preferred suit fit. Thus by building a database of what a good suit fit “looks like” in terms of the pressure distribution and peak pressures reached during certain motions, suit engineers and technicians could aid inexperienced astronauts in determining an appropriate suit fit, based on this experiential data.

Additionally, further studies should integrate pressure data with joint angle information, so that we can determine at exactly what point in the motion these peak pressures are arising, how different movement strategies affect the pressure profiles and distributions of pressure, and how suited motions compare to unsuited ones. All of this information would allow us to more accurately determine when injury is most probable, and therefore aid in preventing such issues.

The conclusions presented in this chapter are the first known publication of a quantitative analysis of the contact pressures observed at the human-suit shoulder interface. For a more complete understanding of the effect of the spacesuit on shoulder injury potential, we must not only consider the contact pressures, but also the muscle forces required to function in the suit. Chapter 4 further investigates the effects of spacesuits on muscle forces in the shoulder region.

MUSCULOSKELETAL MODELING

4.1 OVERVIEW The work presented in this chapter addresses the second specific aim of developing a musculoskeletal model and simulation methodology for analysis of the effects of spacesuits on muscle forces in the shoulder region. Data regarding spacesuit joint torques and the joint angles of a suited subject are integrated into an upper-extremity musculoskeletal model in OpenSim to evaluate which muscles are most affected by the spacesuit. Looking specifically at a shoulder abduction/adduction motion, we explore whether shoulder abductors, adductors, and stabilizer muscle groups experience significant changes in force from the unsuited to suited condition, as well as whether individual muscles within the shoulder region experience significant changes from the unsuited to suited conditions.

4.2 MODELING METHODS The work presented in this chapter is an expansion of a musculoskeletal modeling framework that was developed to analyze human-spacesuit interaction and musculoskeletal performance during extravehicular activity (Diaz and Newman, 2014). For the purposes of this study, we focus on the shoulder abduction/adduction motion of one subject. While the ultimate goal of assessing this methodology is to analyze various motions and subjects, it was determined that focusing on one motion of one subject would provide a good methodological basis that could then be

51 expanded upon. Shoulder abduction/adduction was chosen as the motion of study since lateral and overhead motions are of greatest interest in EVA training-related shoulder injuries (Williams and Johnson, 2003). The chosen subject was selected since he was experienced working in the suit and performed the shoulder abduction/adduction motion without falsely increasing his range of motion through manipulation of the shoulder bearings (Bertrand et al., 2014).

For the particular application discussed in this chapter, adjustments were made to the original framework developed by Diaz and Newman in order to 1) assess the feasibility of using kinematics data from inertial measurement units as opposed to motion capture, and 2) make the framework compatible with the chosen upper-extremity musculoskeletal model. Both of these changes in methodology are discussed in further detail in the following sections. Nevertheless, as in the original methodology, there are three main modeling capabilities needed to understand how the human interacts with the spacesuit: human modeling, spacesuit modeling, and human- spacesuit interaction modeling.

4.2.1 HUMAN MODELING The human-spacesuit interaction model was created using OpenSim, an open-source platform for the development and analysis of musculoskeletal models and dynamic simulations of movement (Delp et al., 2007). A variety of models and simulations are available for use, or users can create their own models. Depending on the application, users can make use of a number of OpenSim’s tools that perform inverse kinematics, inverse dynamics, forward dynamics, or other analyses.

As this study focuses on the shoulder region, it was necessary to select an upper-extremity model. The chosen model (see Figure 4.1) is an adjusted form of the VA Upper Extremity Model, which was developed by a team at Stanford University (Holzbaur et al., 2005). The original model had 15 degrees of freedom for the shoulder, elbow, forearm, wrist, thumb, and index finger, and included 50 muscles across these joints. Kinematics and force-generating parameters were derived from experimental data. This model was adjusted by Hamner and Steele to simplify the shoulder joint and remove the degrees of freedom and muscles associated with the thumb and index finger. The adjusted model thus has 7 degrees of freedom for the shoulder, elbow, forearm, and wrist, and includes 29 muscles. However, this model was intended primarily for kinematic analysis, so it was necessary to add inertial properties before using the model to analyze a dynamic motion. Hence inertial properties were extracted from a more complex shoulder model (the MoBL- FIGURE 4.1FIGURE: ADJUSTED 6: UPPER EXTREMITY MODEL USED FOR ARMS Upper Extremity Dynamic Model – Saul et al., 2015) that is MUSCULOSKETAL ANALYSIS. currently being developed and applied to the simplified model, to (IMAGE SOURCE: OPENSIM) make it appropriate for dynamic analysis. These inertial properties were based on previously published descriptions of the body segments (McConville et al., 1980; Reich and Daunicht, 2000). The MoBL-ARMS Upper Extremity Dynamic Model was not used as its complexity and unstable nature in attempted simulations made its use infeasible for this particular study. Future work should involve a deeper investigation of this complex shoulder model, as stable simulations using this model could lead to more accurate results.

4.2.2 SPACESUIT MODELING As in the original framework developed by Diaz and Newman, the spacesuit’s influence on the human is modeled through the application of external torques to the human model joint. This is a justified method since astronauts must work against the suit to bend the pressurized suit joint.

The torque-angle relationships for the Mark III shoulder joint were obtained experimentally at Johnson Space Center’s Advanced Space Suit Lab, using a modified fish scale method (NASA, 2010). In this method, the pressurized unmanned suit was first secured to a restraint table such that the shoulder joint was parallel to the floor, in an attempt to reduce the effects of gravity. Then, a 3-axis accelerometer was attached to the upper arm bearing, while a load cell was attached closer to the wrist disconnect, perpendicular to the joint axis and parallel to the floor. An image of this test configuration is shown in Figure 4.2. Then, as the test conductor pulled the joint through its range of motion, the load cell measured the force needed to pull the suit joint while the accelerometer recorded the joint angles. It is important to note that the spacesuit was at a nominal suit pressure of 4.3 psi, and was tested without the Thermal Micrometeoroid Garment or the outer layer.

FIGURE 4.2: MODIFIED FISH-SCALE TEST SETUP FOR SHOULDER ABDUCTION/ADDUCTION OF THE MARK III. AT LEFT: OVERALL TEST SETUP, TOP RIGHT: ACCELEROMETER, BOTTOM RIGHT: LOAD CELL. (IMAGES ADAPTED FROM NASA, 2010) To obtain the torque values, the measured forces were simply multiplied by the distance to the point of rotation. The same test conductor pulled the shoulder through its range of motion numerous times, resulting in the collection of torque-angle curves shown in Figure 4.3. The rectangle overlaid on this graph is meant to show the requirements for the range of motion during lunar EVA (35 degrees adduction through 160 degrees abduction). From the curves shown in Figure 4.3, a conservative torque-angle curve was extracted for use in modeling the spacesuit. In other words, the highest torque values for each angle were extracted, such that simulations using this data would consider the worst case scenario. The adapted torque-angle curve is shown in Figure 4.4, where the torque is converted into SI units, and the sign convention is flipped such that an increase in positive degrees indicates abduction. The depicted relationship in Figure 4.4 only goes up to the operational requirements limit of 160 degrees as any angles above that are nearly impossible to achieve with a person inside the suit, whereas the data from the NASA document was an empty-suit test. It should also be noted that the torque-angle data provided in

53 the NASA document has a lower limit of 27 degrees abduction, presumably a result of the fact that the Mark III shoulder components are naturally at a rest position around this value. For shoulder abduction/adduction motions with a maximum angle of less than the maximum of 160 degrees, as was the case for the data used in the study (the subject consistently had a maximum angle that was less than 100 degrees), corresponding torque-angle curves were created for the range of motion that was achieved.

FIGURE 4.3: TORQUE-ANGLE CURVES RESULTING FROM TESTING OF THE SHOULDER ABDUCTION/ADDUCTION MOTION OF THE MARK III AT 4.3 PSI. TORQUES EXPRESSED IN INCH-POUNDS. (IMAGE SOURCE: NASA, 2010)

FIGURE 4.4: EXTRACTED CONSERVATIVE VALUES FOR THE TORQUE- ANGLE RELATIONSHIPS SHOWN IN FIGURE 4.3. TORQUES EXPRESSED IN UNITS OF NM.

4.2.3 HUMAN-SPACESUIT INTERACTION MODELING Kinematic data from inertial measurement units (IMUs) were obtained from the same experiment described in Chapter 3 of this thesis. As a reminder to the reader, subjects performed each of five motions a total of 12 times, with the repetitions divided into three groups of four repetitions. Motions were divided into movement groups such that the order was counterbalanced within the group and randomized between subjects (Anderson et al., 2014). Both internal body joint angles and external suit angles were collected for all motions. Based on both quantitative measures and subjective feedback, it was determined that no fatigue was noted across the duration of the experiment.

As mentioned in the introduction to this section, this study focuses on the shoulder abduction/adduction motion of one subject: Subject 3 from the previously presented experiment. Subject 3 was selected since he was more experienced working in the suit than Subject 1, and he performed the shoulder abduction/adduction motion without falsely increasing his range of motion through FIGURE 4.5: IMU PLACEMENT ON THE BODY manipulation of the shoulder bearings, as Subject 2 did INSIDE THE SUIT. (IMAGE ADAPTED FROM (Bertrand et al., 2014). While joint angles were collected BERTRAND ET AL., 2014) for all 12 of Subject 3’s repetitions of shoulder abduction/adduction, potentially unreliable IMU readings for 4 of these repetitions left 8 repetitions for consideration in this study. The included repetitions were all four repetitions of Movement Group 1, and the first two repetitions of each of Movement Groups 2 and 3. For this motion, the subject began with both arms at his side, and then bent the arms at the shoulder joint through the coronal plane, through his maximum range of motion (shoulder abduction). After reaching the extent of his range of motion, the subject the lowered his arms to the relaxed position (shoulder adduction). A depiction of an example shoulder abduction/adduction motion is shown below in Figure 4.6. The maximum shoulder angle reached in each of the repetitions ranged between 72 and 97 degrees.

FIGURE 4.6: CONCEPTUAL DEPICTION OF SUITED SHOULDER ABDUCTION/ADDUCTION. (IMAGE SOURCE: P. BERTRAND)

The internal joint angles were used to drive the kinematics of the human musculoskeletal model and for determination of the corresponding spacesuit joint-torques to be applied throughout the motion. For a detailed explanation of how joint angles were obtained using the IMUs, see Bertrand et al., 2014. Figure 4.5 shows the placement of the IMUs on the person inside the suit, as well as on the exterior of the suit itself. Relevant measurements of the subject are shown in Table 4.1.

TABLE 4.1: RELEVANT UPPER BODY ANTHROPOMETRIC MEASUREMENTS OF SUBJECT 3 (CM) Measurement Length (cm) Shoulder - Elbow 22 Elbow - Wrist 25

Integrating the human modeling and spacesuit modeling in OpenSim follows the process shown in Figure 4.7. First, using OpenSim’s scaling tool, the musculoskeletal model was scaled to match the subject’s anthropometry. This scales the body’s dimensions as well as the inertial properties of the different body segments. Then, a motion file—which includes shoulder angles as a function of time—was created using the internal joint angles from the IMUs. This data is loaded into OpenSim to drive the kinematics of the scaled musculoskeletal model. Finally, using the scaled model and the newly created motion file as inputs to OpenSim’s Static Optimization tool, we can compute the muscle activations and forces that drive the dynamic model through the prescribed motion. Having not applied any external torques, these results correspond to the unsuited condition. For the suited condition, an external torques file that includes shoulder torques as a function of time can be included as an additional input. This external torques file is created using both the experimental IMU joint angles and the torque-angle relationships described previously.

FIGURE 4.7: MODELING METHODOLOGY DEPICTING THE STEPS FOR ANALYZING MUSCLE DYNAMICS.

Static Optimization is not the only tool that can be used to determine muscle forces in OpenSim. An alternative option is Computed Muscle Control (CMC), which was used in the original methodology developed by Diaz and Newman. Either tool can be used to analyze dynamic motions, the difference being the methods by which the muscle forces are calculated. Static Optimization uses an inverse dynamics approach while CMC uses a forward dynamics approach. Inverse dynamics uses the known kinematics that describe a model’s motion and any external loads applied to the model to solve the equations of motion for the net forces and torques at each joint that produce the movement (Kuo, 1998). The Static Optimization tool then takes this analysis a step further and resolves the net joint moments into individual muscle forces at each moment in time, using an optimization process that minimizes the sum of squared muscle activations

56 amongst all of the muscles and considers the force-length-velocity properties of muscle. On the other hand, CMC uses a combination of proportional-derivative (PD) control and static optimization to compute the muscle excitation levels that will drive the joint angles of the model toward the specified kinematic trajectories (Thelen and Anderson, 2006). As in Static Optimization, at each step of the motion the equations of motion are solved for the net forces and torques at each joint; however the difference is that forward dynamics treats muscle forces as time-dependent variables, which may more accurately represent the physiological properties of the muscles (Anderson et al., 2001) and is less likely to lead to discontinuities in muscle forces that may be seen with the inverse dynamics technique. However, the disadvantage of forward dynamics is that it is much more computationally expensive, especially considering that some studies have shown that these two methods can lead to very similar results (Anderson et al., 2001).

For the purposes of this thesis, both Static Optimization and Computed Muscle Control were tested for one example motion. While Static Optimization took a few minutes to run and CMC took just under one hour, the resulting patterns and magnitudes of muscle forces were similar between the two methods (both of which were verified against published data of the individual muscle forces that arise during a shoulder abduction/adduction motion (Escamilla et al., 2012)). Additionally, no discontinuities in muscle forces were found in the results of Static Optimization, but there were discontinuities found in the CMC results. In general, throughout the preliminary testing of the final chosen model as well as others that were being considered, it seemed that CMC was unstable. Many times the simulations would crash or be unable to solve. This is presumably a result of the inherent difficulty of solving forward dynamics problems, on top of the complex kinematics of this experimental motion that made use of IMU-generated joint angle data. As we are interested in a reliable and computationally efficient method that would allow for relative comparison between unsuited and suited conditions (as opposed to the exact precision of individual muscle forces in each of the conditions), Static Optimization was chosen for use in this study.

4.3 ANALYTICAL METHODS The chosen musculoskeletal model includes shoulder, upper arm, forearm, and wrist muscles that are required to make a range of shoulder and elbow movements. For the purposes of this study, we are most interested in shoulder muscles that contribute to the shoulder abduction/adduction motion. In addition to shoulder abductors and adductors, this also includes the rotator cuff muscles, which contribute to the stability of the shoulder joint (Williams and Johnson, 2003). These four muscles—the supraspinatus, infraspinatus, subscapularis, and the teres minor, shown in Figure 4.8—work with other shoulder muscles to maintain shoulder stability and allow it to rotate. The supraspinatus also aids in the first portion of shoulder abduction. Shoulder FIGURE 4.8: ROTATOR CUFF MUSCLES. (IMAGE abductors in the chosen model include the anterior ADAPTED FROM WILLIAMS AND JOHNSON, deltoid, middle deltoid, and supraspinatus, while 2003)

57 shoulder adductors in the model include the latissimus dorsi (thoracic, lumbar, and iliac), pectoralis major (clavicular, sternal, and ribs), and teres major muscles. The locations of these muscles are shown below in Figure 4.9. Muscle forces for each of these muscle groups and individual muscles were determined for both the unsuited and Mark III-suited conditions.

FIGURE 4.9: SHOULDER ABDUCTORS AND ADDUCTORS OF THE CHOSEN MODEL. AT LEFT: CHOSEN ABDUCTOR MUSCLES INCLUDE THE ANTERIOR AND MIDDLE DELTOIDS, AS WELL AS THE SUPRASPINATUS. AT RIGHT: CHOSEN ADDUCTOR MUSCLES INCLUDE THE LATISSIMUS DORSI, TERES MAJOR, AND PECTORALIS MAJOR. (IMAGE ADAPTED FROM WIKIMEDIA COMMONS.)

Statistical tests were used to determine which muscle groups and individual muscles experience significant changes from the unsuited to suited conditions. The values compared were the peak forces in each muscle group or individual muscle across the 8 repetitions, for each of the unsuited and suited conditions. Peak forces were tested for normality using the Kolmogorov-Smirnov test, however most of the samples did not satisfy normality, likely due to the small sample size. As a result, the non-parametric Wilcoxon signed rank test was used to compare the peak forces between the unsuited and suited conditions. For all tests, an initial alpha value of 0.05 was used, and a family-wise error correction was implemented to limit the probability of Type I error.

4.4 RESULTS

4.4.1 MUSCLE GROUPS Total forces for shoulder abductors, adductors, and stabilizer muscles are shown for one representative repetition (Movement Group 3, Repetition 4) in Figure 4.10. The Figure 4.10a depicts the total force being exerted by the anterior deltoid, middle deltoid, and supraspinatus muscles, Figure 4.10b shows the total force being exerted by the latissimus dorsi, pectoralis major, and teres major muscles, and Figure 4.10c shows the total force being exerted by the infraspinatus, subscapularis, and teres minor muscles. Appendix E shows the same graphs for all eight of the repetitions.

A few general observations can be made from these graphs. Looking specifically at Figure 4.10a, it appears that the total force exerted by the shoulder abductors is substantially higher in the suited condition, although the force profiles of the unsuited and suited cases follow a similar shape. The frequent peaks and valleys in the profiles are simply a result of the realistic joint angle data. If we had run this simulation on an idealized shoulder abduction/adduction motion with smooth velocity and acceleration, the results would have followed a more parabolic shape. However, the true motion of the suited subject is of variable velocity and acceleration, resulting in the “bumpier” pattern seen in Figure 4.10a. Now considering the adductors in Figure 4.10b, we see that the total forces reached overall are small in magnitude compared to the abductors and stabilizers, and that there only appears to be adductor activation at the start and end of the motion. Also, there does not appear to be a substantial difference between the unsuited and suited forces. Finally, looking at Figure 4.10c, it seems that the stabilizer muscles experience slightly higher forces in the suited condition, but that the force profiles of the unsuited and suited cases follow a similar shape. For all three graphs, it should be noted that the unsuited and suited motions do not differ until approximately 20% of the movement. This is a result of the fact that the lower bound of suit torque data was 27 degrees abduction, as mentioned previously.

However, these graphs only depict the force profiles for one of the eight repetitions that were analyzed. Thus a Wilcoxon signed rank test was used to determine whether the difference between the unsuited and suited conditions was statistically significant, across all trials. A Bonferroni correction was made to account for the family-wise error rate, which brought the alpha value for significance from 0.05 to 0.0167. These tests suggested that while there was a statistically significant difference for the abductors (p=0.0078), there was no significant difference for the adductors (p=0.625) or stabilizers (p=0.0313). Peak forces across all eight trials in each condition are shown in Table 4.2 (mean + SD), and the corresponding box plots are shown in Figure 4.11.

FIGURE 4.10: TOTAL FORCE BY MUSCLE GROUP. AT TOP LEFT: TOTAL ABDUCTORS MUSCLE FORCE. AT TOP RIGHT: TOTAL ADDUCTORS MUSCLE FORCE. AT BOTTOM: TOTAL STABILIZERS MUSCLE FORCE. GRAPHS SHOWN FOR ONE REPRESENTATIVE MOTION. BODY DIAGRAMS SHOW MUSCLES INCLUDED IN THE GRAPH.

TABLE 4.2: PEAK FORCES BY MUSCLE GROUP, IN NEWTONS (N). LISTED AS MEAN + SD, AND SHOWING P-VALUES. ASTERISK * INDICATES A STATISTICALLY SIGNIFICANT DIFFERENCE BETWEEN THE UNSUITED AND SUITED CONDITIONS.

Muscle Group Unsuited Suited P-Value Abductors 289.2 + 43.0 379.1 + 65.0 0.0078* Adductors 41.4 + 30.1 41.5 + 30.2 0.625 Stabilizers 223.1 + 44.0 275.6 + 67.1 0.0313

FIGURE 4.11: BOXPLOTS COMPARING UNSUITED AND SUITED CASES, BY MUSCLE GROUP. AT TOP LEFT: TOTAL ABDUCTORS MUSCLE FORCE. AT TOP RIGHT: TOTAL ADDUCTORS MUSCLE FORCE. AT BOTTOM: TOTAL STABILIZERS MUSCLE FORCE. PLOTS INCLUDE DATA FROM ALL 8 SIMULATIONS.

4.4.2 INDIVIDUAL MUSCLES Given the results of the muscle group analysis, it became apparent that shoulder abductors were the muscle group of largest concern, but that the stabilizers also seemed to be a potential concern. Thus a similar investigation was performed looking at individual muscle forces within the abductor and stabilizer muscle groups. Individual muscle forces for shoulder abductors and stabilizers are shown for one representative repetition (Movement Group 3, Repetition 3) in Figures 4.12 and 4.13. Figure 4.12 depicts the forces being exerted by the anterior deltoid (DELT1), middle deltoid (DELT2), and supraspinatus muscles (SUPRA), for the unsuited and suited conditions. Figure 4.13 shows the force being exerted by the infraspinatus (INFRA), subscapularis (SUBSC), and teres minor (TMIN) muscles, for the unsuited and suited conditions. Appendix F shows the same graphs for all eight of the repetitions.

FIGURE 4.12: INDIVIDUAL ABDUCTOR MUSCLE FORCES. AT LEFT: INDIVIDUAL MUSCLE FORCES IN THE UNSUITED CONDITION. AT RIGHT: INDIVIDUAL MUSCLE FORCES IN THE SUITED CONDITION. GRAPHS SHOWN FOR ONE REPRESENTATIVE MOTION. BODY DIAGRAMS SHOW MUSCLES INCLUDED IN THE GRAPH.

A few general observations can be made from these graphs. Looking first at Figure 4.12a, which shows the abductor muscles in the unsuited condition, we see that the middle deltoid (DELT2) is the primary force contributor, although the supraspinatus (SUPRA) also plays a role in approximately the first 30% of the motion. In this particular repetition, the anterior deltoid (DELT1) does not appear to be contributing to the motion. Now comparing Figure 4.12a and 4.12b, we see that the suited condition significantly increases the force experienced by the middle deltoid. The supraspinatus also appears to have a slight increase in force. Then looking at Figure 4.13a, we see that both the infraspinatus (INFRA) and subscapularis (SUBSC) contribute to the first 30% of the motion, while the infraspinatus continues to play a role in the duration of the motion. The teres minor (TMIN) does not appear to contribute to this motion. Comparing Figures 4.13a and 4.13b, we see that the suited condition does increase the force in both the infraspinatus and the subscapularis from about 20% of the movement through 70% of the movement. The suit does not appear to make a difference in the force experienced by the teres minor.

FIGURE 4.13: INDIVIDUAL STABILIZER MUSCLE FORCES. AT LEFT: INDIVIDUAL MUSCLE FORCES IN THE UNSUITED CONDITION. AT RIGHT: INDIVIDUAL MUSCLE FORCES IN THE SUITED CONDITION. GRAPHS SHOWN FOR ONE REPRESENTATIVE MOTION. SHOULDER DIAGRAMS SHOW MUSCLES INCLUDED IN THE GRAPH. Again, however, these graphs only depict the force profiles for one of the eight repetitions that were analyzed. Thus a Wilcoxon signed rank test was used to determine whether the difference between the unsuited and suited conditions was statistically significant, and a Bonferroni correction was made to account for the family-wise error rate, bringing the alpha value for significance from 0.05 to 0.0083. These tests suggested that while there was a statistically significant difference for the middle deltoid (p=0.0078), there was no significant difference for the other abductors or for any of the stabilizers. Table 4.3 shows the peak force for each of the individual muscles (mean + SD) across all eight repetitions and their corresponding p-values. The corresponding box plots are shown in Figure 4.14.

TABLE 4.3: PEAK FORCES FOR INDIVIDUAL MUSCLES, IN NEWTONS (N). LISTED AS MEAN + SD, AND SHOWING P-VALUES. ASTERISK * INDICATES A STATISTICALLY SIGNIFICANT DIFFERENCE BETWEEN THE UNSUITED AND SUITED CONDITIONS. Muscle Unsuited Suited P-Value Anterior Deltoid 51.3 + 58.2 51 + 58.1 0.50 Abductors Middle Deltoid 242 + 87.8 337.1 + 115.3 0.0078* Supraspinatus 38.1 + 17.4 43.1 + 15.2 0.0156 Infraspinatus 139.3 + 61.9 168.3 + 85.2 0.0313 Stabilizers Subscapularis 89.9 + 22.3 113.4 + 17.1 0.0156 Teres Minor 7.7 + 5.2 7.9 + 5.9 0.6875

FIGURE 4.14: BOXPLOTS COMPARING UNSUITED AND SUITED CASES, FOR INDIVIDUAL MUSCLES. AT TOP LEFT: ANTERIOR DELTOID. AT TOP RIGHT: MIDDLE DELTOID. AT MIDDLE LEFT: SUPRASPINATUS. AT MIDDLE RIGHT: INFRASPINATUS. AT BOTTOM LEFT: SUBSCAPULARIS. AT BOTTOM RIGHT: TERES MINOR. PLOTS INCLUDE DATA FROM ALL 8 SIMULATIONS.

4.5 DISCUSSION When considering the total forces for each of the muscle groups, we saw in Figure 4.10a that the total force exerted by the shoulder abductors is substantially increased from the unsuited to suited conditions. This increase in force makes sense since the suited pressurized condition adds an additional torque that the primary abduction muscles must overcome to perform the shoulder abduction/adduction motion. An interesting observation was that the shoulder adductors did not contribute substantially to the motion, and that the comparison between unsuited and suited conditions yields only minor changes (Figure 4.10b). One would expect that the adductors would be particularly involved in the second half of the motion, corresponding to the adduction phase. The fact that the adductors appear active at the start and end of the motion suggests that perhaps the “impure” nature of this shoulder abduction/adduction motion is the reason that we do not see expected results. In this context, “impure” points to the fact that the subject did not perform a perfectly planar shoulder abduction/adduction motion. This might suggest that muscles other than the shoulder adductors (such as those that facilitate rotation or lateral translation of the humerus) might be the primary contributors to the second half of the shoulder abduction/adduction motion. Finally, when considering the stabilizers, it was noted that they contribute substantially to the first 30% of the motion, and then play a more minor role for the remaining duration of the motion (Figure 4.10c). Interesting to note is that the stabilizers are active between 30% and 80% of the movement, which is precisely the portion of the motion in which the adductors are not contributing. Since the stabilizer muscles also allow for internal and external rotation of the humerus, it therefore seems likely that, as suggested previously, the rotated position of the arm is the reason that the typical adductors did not contribute substantially to this motion.

From the statistical analysis, it was found that there was a statistically significant difference between the peak pressures that arose in the unsuited and suited conditions for the abductors, but not for the adductors or stabilizers. As the abductors were the primary contributors to this motion, it is no surprise that the presence of the spacesuit has a substantial effect on the muscle forces required to perform the motion. On the other hand, if we consider the adductors, we do not see a significant difference between the forces in the unsuited and suited conditions since the adductors do not contribute substantially to this motion. Statistically, the stabilizer muscles did not see a significant change between the unsuited and suited conditions either. This result is a bit surprising since these muscles appear to contribute substantially to the motion (Figure 4.10c). Looking at Figure 4.10c and the corresponding boxplot in Figure 4.11, we see that there does appear to be a noticeable increase from the unsuited to suited muscle forces. However, this difference was not statistically significant given the correction for family-wise error.

After investigating the muscle groups that contribute to the shoulder abduction/adduction motion, both the shoulder abductors and stabilizers were investigated in further detail. We saw in Figure 4.12a that the main abductor contributing to the motion was the middle deltoid, and that the supraspinatus was a secondary contributor for the first 30% of the motion. These results make sense since the middle deltoid is the primary muscle associated with shoulder abduction and the supraspinatus contributes to the first 20 degrees of abduction (Williams and Johnson, 2003). Since the middle deltoid is the primary contributor, it makes sense that its force increases substantially in the suited condition (Figure 4.12b) as it is the main muscle countering the spacesuit’s joint torque. Then, looking at the individual stabilizer muscles in Figure 4.13a, we saw that both the infraspinatus and subscapularis contribute largely to the first 30% of the motion,

65 and that the infraspinatus continues to be active throughout the rest of the motion. Both of these muscles are used for general stabilization of the joint, while the infraspinatus aids in external rotation of the humerus and the subscapularis is used in internal rotation. Comparing from the unsuited to suited condition, both of these muscles saw an increase in force (Figure 4.13b).

From the statistical analysis, it was found that out of all of the abductors and stabilizers, the middle deltoid was the only one that experienced a statistically significant increase from the unsuited to suited conditions. As the middle deltoid is the primary contributors to the abduction motion, it is no surprise that the presence of the spacesuit has a substantial effect on the muscle forces required to perform the motion. However, if we consider the rotator cuff muscles, we do not see a significant difference between the forces in the unsuited and suited conditions. Looking at Table 4.3 and the corresponding boxplots in Figure 4.14, we see that there does appear to be somewhat of an increase from the unsuited to suited muscle forces for a few of the rotator cuff muscles; however, this difference was not statistically significant. A summary of all results is presented in Table 4.4, shown below.

TABLE 4.4: SUMMARIZED RESULTS – PEAK FORCES FOR MUSCLE GROUPS AND INDIVIDUAL MUSCLES, IN NEWTONS (N). LISTED AS MEAN + SD, AND INDICATING WHETHER THE UNSUITED AND SUITED CONDITIONS WERE FOUND TO BE STATISTICALLY SIGNIFICANT.

Name Unsuited Suited Stat. Significant? Abductors 289.2 + 43.0 379.1 + 65.0 Yes Muscle Groups Adductors 41.4 + 30.1 41.5 + 30.2 No Stabilizers 223.1 + 44.0 275.6 + 67.1 No Anterior Deltoid 51.3 + 58.2 51 + 58.1 No Middle Deltoid 242 + 87.8 337.1 + 115.3 Yes Muscles Supraspinatus 38.1 + 17.4 43.1 + 15.2 No Infraspinatus 139.3 + 61.9 168.3 + 85.2 No Subscapularis 89.9 + 22.3 113.4 + 17.1 No Teres Minor 7.7 + 5.2 7.9 + 5.9 No

Finally, we can relate these results to the potential for muscle injury. One measure of injury susceptibility is the peak muscle force developed during a motion (Salmons, 1997; Zatsiorsky, 2000; Diaz and Newman, 2014). Previous studies have shown that there is a high probability of injury if the peak force is greater than or equal to 150% the muscle’s maximum isometric force (Salmons, 1997). Table 4.5 shows the maximum isometric forces of each of the abductor and stabilizer muscles investigated in this chapter, as extracted from the musculoskeletal model (Holzbaur et al., 2005). While we could directly compare these values to the peak forces reached in our suited simulation, we would not see any peak forces nearing the injury threshold. This is because OpenSim distributes the forces across muscles using an optimization algorithm specifically designed so as to make the motion a feasible one. Nevertheless, if we see that a particular muscle is approaching forces close to the maximum isometric force and that subsequently a muscle with similar function increases in activation, this may be indicative of a muscle approaching forces of concern.

TABLE 4.5: MAXIMUM ISOMETRIC FORCES OF CONSIDERED MUSCLES (N).

4.6 LIMITATIONS While this type of analysis can be very useful in understanding the potential for shoulder muscle injury, there are a number of limitations associated with this particular study that could be improved upon in future work.

The first limitation is the musculoskeletal model itself, which, as mentioned previously, uses a simplified shoulder model that is modeled as a ball-and-socket joint. Even though the magnitudes of muscle forces found using this simplified model were similar to experimental EMG data for a shoulder abduction/adduction motion, the accuracy would likely be improved by using a more realistic shoulder joint model. While we expect the relative increase in muscle forces from unsuited to suited conditions to be similar between the simplified and complex models, the absolute magnitudes of these forces will likely differ between the two models. Thus to obtain more accurate muscle force magnitudes, future work should integrate the more complex model (Saul et al., 2015) to see if the results of this study are affected.

Another limitation that must be taken into account when considering the results of the study is the torque-angle data. As mentioned in the methodology, this data was collected using the modified fish-scale method, a technique which tends to underestimate the torques that would be measured if a human were in the suit (Schmidt, 2001). While this is a limitation of the study, the document used to obtain these torque-angle values is the most complete suit torque test report that is available for use at this time.

The final limitations to be mentioned regard the small number of repetitions used and the variability within the subject’s motions. The small sample size of eight repetitions both makes it difficult to find statistical trends and limits the generalization of results. Using a larger sample size and expanding the analysis to multiple subjects would allow for an interesting investigation into which muscles are most affected by the spacesuit for individual subjects, and whether there are any similar results across subjects. A related point is that even though all of the data used here

67 was from one subject performing eight repetitions of what was deemed a shoulder abduction/adduction motion, the actual shoulder angles for each of these repetitions varied substantially. Variation in how the subject performs the motion has a large effect on the muscle forces that arise, causing a large spread in the data. Analyzing a small number of widely variable motions makes finding trends particularly difficult. While future work can certainly increase the sample size, variation of subjects’ motion is difficult to control. Further investigations could consider improved methods for ensuring subjects’ consistency of motion, or for taking this variability into account when analyzing the results.

4.7 CONCLUSION In summary, the work presented in this chapter determines which muscles associated with a shoulder abduction/adduction motion are most affected by the Mark III spacesuit. We can directly address the research questions posed in Chapter 1, restated below.

From this analysis, we found that of the three investigated muscle groups—shoulder abductors, adductors, and stabilizers—only the abductors experience a statistically significant change in total muscle force between the unsuited and suited conditions. Looking specifically at the individual muscles that constitute the abductors and stabilizers, we found that only the middle deltoid experienced a statistically significant change in force from the unsuited to suited condition. A number of explanations were provided for the observed force profiles and the statistical results. The results were summarized in Table 4.4, shown again below.

TABLE 4.4: SUMMARIZED RESULTS – PEAK FORCES FOR MUSCLE GROUPS AND MUSCLES, IN NEWTONS (N). LISTED AS MEAN + SD, AND INDICATING WHETHER THE UNSUITED AND SUITED CONDITIONS WERE FOUND TO BE STATISTICALLY SIGNIFICANT. Name Unsuited Suited Stat. Significant? Abductors 289.2 + 43.0 379.1 + 65.0 Yes Muscle Groups Adductors 41.4 + 30.1 41.5 + 30.2 No Stabilizers 223.1 + 44.0 275.6 + 67.1 No Anterior Deltoid 51.3 + 58.2 51 + 58.1 No Middle Deltoid 242 + 87.8 337.1 + 115.3 Yes Muscles Supraspinatus 38.1 + 17.4 43.1 + 15.2 No Infraspinatus 139.3 + 61.9 168.3 + 85.2 No Subscapularis 89.9 + 22.3 113.4 + 17.1 No Teres Minor 7.7 + 5.2 7.9 + 5.9 No

The results of this modeling and analysis can be used to assess which muscle groups and individual muscle forces may be most susceptible to injury in the spacesuit given their relative increase from the unsuited to suited condition. Future work should expand this analysis to

68 include a larger sample size and a more complex upper extremity model, as these improvements would increase the accuracy of the absolute muscle forces calculated by the simulation. These muscle forces could then be compared to published thresholds for muscle injury, thus determining not only which muscles experience a statistically significant increase from the unsuited to suited conditions, but also which muscles are above the threshold for injury in the suited condition.

The ultimate goal of this work would be to expand this musculoskeletal framework to other joints and spacesuits, thereby providing useful information to EVA operations teams, flight doctors, and spacesuit designers alike. Performing this analysis for more complex EVA tasks—whether it involves the upper body, lower body, or full body—could provide information on which movements or tasks should be avoided or performed minimally to prevent injury. The resulting muscle forces could also be used to set limits on the joint torques that are engineered in future spacesuits.

CONCLUSION

Extravehicular activity (EVA) is a critical and complex aspect of human spaceflight missions. To prepare for safe and successful execution of required tasks, astronauts undergo extensive training and hours of repetitive motions at various orientations, while wearing a pressurized spacesuit. The current U.S. spacesuit—the Extravehicular Mobility Unit (EMU)—is pressurized to 29.6 kPa (4.3 psi) and requires astronauts to exert a substantial amount of energy in order to move the suit into a desired position. The pressurization of the suit therefore limits human mobility, causes discomfort, and leads to a variety of contact and strain injuries. Shoulder injuries are one of the most severe injuries that astronauts contend with, and they are mainly attributed to the EMU’s hard upper torso (HUT). The HUT restricts shoulder mobility and can cause injuries when astronauts must support their body weight against the HUT. While suit-related injuries have been observed for many years and basic countermeasures have been implemented, there is still a lack of understanding of how humans move within the spacesuit.

Prior to the research of this thesis and related work under the same grant (Anderson et al., 2014; Bertrand et al., 2014; Diaz and Newman, 2014; Anderson, 2015), there was no conventional method to evaluate human movement within the spacesuit, and few attempts to determine the quantitative nature of this interaction had been made. At this point in time, suited evaluations typically consider the human-suit system as a whole, judging performance with measures such as joint torque, range of motion, or work envelope (Morgan et al., 1996; Jaramillo et al., 2008;

Matty and Aitchison, 2009; Norcross et al., 2010; Aitchison, 2012; Valish and Eversley, 2012). Furthermore, while many studies have been published on the incidence and mechanisms of both general astronaut injury (Wear, 1999; Johnson et al., 2004; Viegas et al., 2004; Strauss et al., 2005; Scheuring et al., 2009) and shoulder-specific astronaut injury (Graziosi et al., 2000; McCluskey, 2002; Williams and Johnson, 2003; Johnson et al., 2004; Scheuring et al., 2012), the suggested causes of shoulder injury have not been evaluated quantitatively. The work presented here addresses this research gap by providing quantitative and repeatable methods of evaluating human-spacesuit interaction in the shoulder region.

The five primary contributions of this research are the following: 1. Presenting the first quantitative “look” inside the spacesuit at the shoulder interface 2. Developing quantitative methods for analyzing pressure distributions and pressure profiles that arise during dynamic shoulder motions 3. Quantitatively evaluating the effects of anthropometric region and motion type on pressures experienced at the region 4. Expanding an existing methodology for musculoskeletal modeling of human-spacesuit interaction to a complex shoulder joint motion 5. Improving an existing methodology for musculoskeletal modeling of human-spacesuit interaction through the use of suited body joint angles obtained from inertial measurement units

The primary objective of this work is to develop an increased understanding of the potential for shoulder injury in pressurized spacesuits, a goal that is accomplished through two specific aims:

1) Quantifying and analyzing the pressures that arise at the human-suit shoulder interface

2) Assessing the effects of spacesuits on the muscle forces that arise during shoulder motions, using an adjusted methodology for modeling human-spacesuit interaction

A variety of graphical and statistical analyses were performed to evaluate the pressure distributions and pressure profiles that arose at the interface between the human shoulder and the torso of the spacesuit. The graphical analysis provided an “inside look” of how the spacesuit affects the pressure distributions and pressure profiles experienced at the shoulder. For this particular study, it was revealed that 1) the least experienced subject generated the highest pressures, 2) the region just above the clavicle over the soft musculature at the top of the shoulder was of particular concern, as pressure was concentrated in this location for the majority of movements for all subjects, and it was also one of the regions in which maximum pressure is located most frequently, and 3) the top of the shoulder blade was a secondary region of concern for some subjects. Pressure profile analysis further showed that 1) for most subjects, general profile trends varied in shape across movement groups, 2) repetitions within each movement group were consistent in shape, and for most subjects also in magnitude, 3) the highest pressures were typically found near the top of the shoulder, and 4) the shoulder blade area was of concern for at least one subject. The statistical analysis considered the effects of motion type and anthropometric region on peak pressure magnitudes, showing that results cannot be generalized across subjects as they are likely affected by individual anthropometry, suit fit, and the biomechanics of how each subject performs the motion. However, a number of interesting trends regarding which motions or regions yielded higher pressures were found for each of the individual subjects. For improvement of this work, further studies should integrate pressure data

72 with joint angle information, thus providing more insight as to how different movement strategies affect the pressure profiles and distributions of pressure, and how suited motions compare to unsuited ones. Such an analysis would allow for more accurate determination of when injury is most probable, and therefore help to prevent such injuries.

Additionally, an existing framework for musculoskeletal modeling of human-spacesuit interaction was expanded to evaluate which shoulder muscles are most affected by the spacesuit. Looking specifically at a shoulder abduction/adduction motion, it was found that of the three investigated muscle groups—shoulder abductors, adductors, and stabilizers—only the abductors experience a statistically significant change in total muscle force between the unsuited and suited conditions. Looking specifically at the individual muscles that constitute the abductors and stabilizers, only the middle deltoid was found to experience a statistically significant change in force from the unsuited to suited condition. A number of explanations were provided for the observed force profiles and the statistical results. In order to increase the accuracy of the results, future studies should expand this analysis to a larger sample size and a more complex upper extremity model. Comparing the results to published thresholds for muscle injury could then provide information as to which of these muscles are above the threshold for injury in the suited condition.

Each of the specific aims addressed in this thesis provides a different avenue for approaching the issue of shoulder injury in the spacesuit. While pressure analysis contributes to the understanding of human-spacesuit interaction by identifying the anthropometric regions that might be most susceptible to contact injury, musculoskeletal analysis demonstrates which individual muscles are most susceptible to strain injury. Quantifying the interaction between the human and the spacesuit could aid in the suit fit process, by informing appropriate sizing of the HUT or placement of padding to reduce the risk of injury. Modeling the interaction contributes valuable information to flight doctors, EVA operations teams, and astronauts as to which movements or tasks should be avoided or performed minimally to prevent injury, while the resulting muscle forces could be used by spacesuit designers to set limits on the suit joint torques. All in all, both of these quantitative, evidence-based approaches help to develop an increased understanding of the potential for shoulder injury in the spacesuit—an issue that must be overcome if we plan to continue toward the ultimate goal of human planetary exploration.

REFERENCES

Aitchison, L. (2012). A Comparison of Methods for Assessing Space Suit Joint Ranges of Motion. International Conference on Environmental Systems. San Diego, CA, American Institute of Aeronautics and Astronautics: 12.

Anderson A., Understanding Human-Space Suit Interaction to Prevent Injury During Extravehicular Activity, Doctoral Dissertation. Dept. Aeronautics and Astronautics, Massachusetts Institute of Technology, Cambridge, MA. June 2014.

Anderson, A. P., Newman, D. J., R. E. Welsch, (2015). “Statistical Evaluation of Causal Factors Associated with Astronaut Shoulder Injury in Space Suits.” Aerospace Medicine and Human Performance 86(7): 1-8.

Anderson, A., Diaz, A., Kracik, M., Trotti, G., Hoffman, J., & Newman, D. (2012). Developing a Spacesuit Injury Countermeasure System for Extravehicular Activity: Modeling and Analysis. In 42nd International Conference on Environmental Systems.

Bertrand P. J., Anderson A., Hilbert A., Newman D. J. (2014). “Feasibility of Spacesuit Kinematics and Human-Suit Interactions.” International Conference on Environmental Systems, Tucson, AZ.

Delp, S. L., Anderson, F. C., Arnold, A. S., Loan, P., Habib, A., John, C. T., & Thelen, D. G. (2007). OpenSim: open-source software to create and analyze dynamic simulations of movement. Biomedical Engineering, IEEE Transactions on, 54(11), 1940-1950.

"Deltoideus". Licensed under CC BY-SA 3.0 via Wikimedia Commons - .

Diaz A., Newman, D.J., “Musculoskeletal Human-Spacesuit Interaction Model”, IEEE Aerospace Conference, Big Sky, MT, March, 2014.

Escamilla, R. F., Yamashiro, K., Paulos, L., & Andrews, J. R. (2009). Shoulder muscle activity and function in common shoulder rehabilitation exercises.Sports medicine, 39(8), 663-685.

Gernhardt, M., et al. (2009). Risk of Compromised EVA Performance and Crew Health Due to Inadequate EVA Suit Systems. Houston, TX, Human Research Program.

Gonzalez L. J., Maida J. C., Miles E. H., et al. (2002). Work and fatigue characteristics of unsuited and suited humans during isolated isokinetic joint motions. Ergonomics 2002; 45: 484-500.

Graziosi, D., Stein, J., & Kearney, L. (2000, July). Space Shuttle small Extravehicular Mobility Unit (EMU) development. In 30 th SAE International Conference on Environmental Systems. Toulouse, France.

Holzbaur, K. R., Murray, W. M., & Delp, S. L. (2005). A model of the upper extremity for simulating musculoskeletal surgery and analyzing neuromuscular control. Annals of biomedical engineering, 33(6), 829-840.

Jaramillo, M. A., et al. (2008). Refinement of Optimal Work Envelope for Extravehicular Activity Suit Operations. Houston, TX, Johnson Space Center.

Johnson, B. J., Cupples, J. S., & Williams, D. (2004, January). Results from an Investigation into Extra-Vehicular Activity (EVA) Training-Related Shoulder Injuries. In AIAA Space 2004 Conference and Exposition. United States: American Institute of Aeronautics and Astronautics Inc (pp. 610-624).

Jordan, N. C., Saleh, J. H., & Newman, D. J. (2005, August). The Extravehicular Mobility Unit: case study in requirements evolution. In Requirements Engineering, 2005. Proceedings. 13th IEEE International Conference (pp. 434-438). IEEE.

Kuo, A. D. (1998). A least-squares estimation approach to improving the precision of inverse dynamics computations. Journal of biomechanical engineering, 120(1), 148-159.

Matty, J. and L. Aitchison (2009). A Method for and Issues Associated with the Determination of Space Suit Joint Requirements. International Conference on Environmental Systems. Berlin, Germany, SAE International.

McCluskey R. NASA Memo t oDr. Craig Fischer, Space and Life Sciences Directorate, Johnson Space Center 2002.

McConville, J. T., Churchill, T. D., Kaleps, I, Clauser, C.E., Cuzzi J. 1980. Anthropometric relationships of body and body segment moments of inertial. Yellow Springs (OH): Wright-Patterson Air Force Base.

Morgan, D. A., et al. (1996). Comparison of Extravehicular Mobility Unit (EMU) Suited and Unsuited Isolated Joint Strength Measurements. Houston, TX, Johnson Space Center.

NASA (2002). NASA Extravehicular Mobility Unit (EMU) LSS/SSA Data Book, Hamilton Sundstrand. Rev H.

NASA Johnson Space Center (2010). Constellation Space Suit Element Test Report: CTSD-CX-5304. Houston, TX, NASA.

NASA Johnson Space Center / Hamilton Standard. NASA Extravehicular Mobility Unit, LSS / SSA data book. Windsor Locks, CT: Hamilton Standard; Oct 1998. JG8353001cv, Rev. E.

NASA Johnson Space Center. Hazard analysis for class 3 SSA and ancillary equipment. Rev B. Houston, TX: NASA; April 2002: 13-32.

NASA Johnson Space Center. Neutral Buoyancy Laboratory standard operating procedures. Houston, TX: NASA; 2002. DX12-0002.

Newman D. J., Schmidt P. B., Rahn D. B., et al. (2000). Modeling the extravehicular mobility unit (EMU) space suit – physiological implications for EVA. 30th SAE International Conference on Environmental Systems. July 10-13 2000; Toulouse, France. Warrendale, PA: SAE International. No 2000-01-2257.

Newman D.J., Anderson A., Bertrand P., Diaz A., Hilbert A., Kracik M., Hoffman J., Trotti G. (2015). “Spacesuit Trauma Countermeasures Research: Injury Prevention and Comfort Protection.” Human Research Program Investigator’s Workshop. Talk, Galveston, TX.

Newman D.J., Anderson A., Diaz A., Kracik M., Hilbert A., Bertrand P., Hoffman J., Trotti G. (2014). “Spacesuit Trauma Countermeasures Research: Injury Prevention and Comfort Protection Design.” Human Research Program Investigator’s Workshop. Talk, Galveston, TX.

Newman, D. J., Hoffman, J., et al. (2014). Spacesuit Trauma Countermeasure System for Intravehicular and Extravehicular Activities: Final NASA HRP Grant Report. Cambridge, MA, Massachusetts Institute of Technology.

Nikooyan, A. A., Veeger, H. E. J., Chadwick, E. K. J., Praagman, M., & van der Helm, F. C. (2011). Development of a comprehensive musculoskeletal model of the shoulder and elbow. Medical & biological engineering & computing, 49(12), 1425-1435.

Norcross, J., et al. (2010). Metabolic Costs and Biomechanics of Level Ambulation on a Planetary Suit. Houston, TX, Johnson Space Center: 74.

Opperman, R. (2010). Astronaut Extravehicular Activity – Safety, Injury, and Countermeasures & Orbital Collisions & Space Debris – Incidence, Impact, and International Policy. Aeronautics and Astronautics, Technology Policy Program, Cambridge, MA, Massachusetts Institute of Technology. M.S.: 183.

Opperman, R. A., et al., (2010). “Probability of spacesuit-induced fingernail trauma is associated with hand circumference.” Aviat Space Environ Med 81(10): 907-913.

Pliance Sensing System (2013). Novel: Quality in Measurement. Technical Data and Use Manual.

Prinold, J. A., Masjedi, M., Johnson, G. R., & Bull, A. M. (2013). Musculoskeletal shoulder models: A technical review and proposals for research foci. Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine, 0954411913492303.

Reich, J., & Daunicht, W. J. (2000). A rigid body model of the forearm. Journal of biomechanics, 33(9), 1159-1168.Salmons, S (1997). Muscle Damage. Oxford University Press.

Saul, K. R., Hu, X., Goehler, C. M., Vidt, M. E., Daly, M., Velisar, A., & Murray, W. M. (2015). Benchmarking of dynamic simulation predictions in two software platforms using an upper limb musculoskeletal model. Computer methods in biomechanics and biomedical engineering, 18(13), 1445-1458.

Scheuring, R. A., Mathers, C. H., Jones, J. A., & Wear, M. L. (2009). Musculoskeletal injuries and minor trauma in space: incidence and injury mechanisms in US astronauts. Aviation, space, and environmental medicine,80(2), 117-124

Scheuring, R., et al., (2008). “The Apollo Medical Operations Project: Recommendations to Improve Crew Health and Performance for Future Exploration Missions and Lunar Surface Operations.” Acta Astronautica 2008(63): 8.

Scheuring, R., et al., (2012). Shoulder Injuries in US Astronauts Related to EVA Suit Design. Aerospace Medical Association. Atlanta, GA.

Schmidt, P. (2001). An Investigation of Space Suit Mobility with Applications to EVA Operations. Aeronautics and Astronautics. Cambridge, MA, Massachusetts Institute of Technology. PhD: 254.

Shields N Jr, King LC. Testing and evaluation for astronaut extravehicular activity (EVA) operability. Human performance in extreme environments. Hum Perf Extrem Environ 1998; 3:145-9.

Stankiewicz T, Dionne S, Prouty B, Case M. Redesign of the shuttle extravehicular mobility unit (EMU) hard upper torso (HUT) to improve overall system safety and reduce system cost. 23rd International Conference on Environmental Systems; Jul 12-15, 1993; Colorado Springs, CO. Warrendale, PA: SAE International; 1993. No. 932100.

Strauss, S. (2004). Extravehicular Mobility Unit Training Suit Symptom Study Report. Houston, TX, Johnson Space Center.

Strauss, S., Krog, R. L., & Feiveson, A. H. (2005). Extravehicular mobility unit training and astronaut injuries. Aviation, space, and environmental medicine,76(5), 469-474.

Thelen, D. G., & Anderson, F. C. (2006). Using computed muscle control to generate forward dynamic simulations of human walking from experimental data. Journal of biomechanics, 39(6), 1107-1115.

Valish, D. and K. Eversley (2012). Space Suit Joint torque Measurement Method Validation. International Conference on Environmental Systems. San Diego, CA, American Institute of Aeronautics and Astronautics: 14.

Viegas, S., et al., (2004). “Physical Demands and Injuries to the Upper Extremity Associated with the Space Program. Journal of Hand and Surgery. 29(3): 7.

Wear M. Injury rate of shuttle astronauts. The Longitudinal study of Astronaut Health Newsletter, December 1999; 8(2):1, 4.

Williams, D.R. and B.J. Johnson (2003). EMU Shoulder Injury Tiger Team Report. Houston, TX: 104.

Witt, J. and J. Jones (2007). Evaluation of the Hard Upper Torso Shoulder Harness. Houston, TX, NASA.

Zatsiorsky, V. M. (2000). “Biomechanics in Sport. Performance Enhancement and Injury Prevention.” The Encyclopedia of Sport Medicine, Volume IX, International Olympic Committee.

APPENDIX A: EXPERIMENT TEST PLAN

1.0 BACKGROUND

1.1 Test Objective

The objective of this research is to develop an understanding of how the person interacts with the space suit, and use that information to assess and mitigate injury. Our approach is to quantify and evaluate human-space suit interaction with a pressure sensing tool, focusing on the arm and shoulders under different loading regimes. Additionally, inertial measurement units (IMUs) will be placed both internal and external to the space suit arm to assess biomechanics. Informal objectives include evaluating the feasibility of pressure sensing as a platform used inside the space suit. Additionally, this process will foster a collaborative relationship between MIT and EC5 as both organizations independently pursue tools that aid in characterizing the human/suit interface. EC5 has indicated that this capability is of high priority for future development and characterization of spacesuit prototypes. Similarly, MIT has recognized both the need and the importance of this capability and has developed a prototype platform, consisting of custom and off-the-shelf sensors and an integrated data acquisition system, all incorporated into a modified athletic garment, capable of sensing pressure at various locations along the arm and shoulder. Establishing a precedent and proof of concept for this methodology will open the doorway for future collaboration and technology development.

1.2 Test Approach Summary

Multiple subjects will be asked to perform the test protocol in the Mark III (MkIII) space suit. Subjects will be selected based on availability from a group of NASA personnel who meet the medical requirements for in-suit testing. These individuals have a great deal of experience working inside the space suit so will not have to develop new, potentially confounding movement strategies. A minimum of 3 subjects is needed for the experiment. The subjects will be wearing the pressure sensing and IMU systems while performing the tests, and pressure profiles and angle histories will be recorded. The test protocol will consist of 12 repetitions of 5 motions inside the space suit. The selected movements use the upper body where the sensors are placed. The 5 motions are 3 isolated joint movements (Elbow flexion/extension, Shoulder flexion/extension, and Shoulder abduction/adduction) and 2 functional tasks (Hammering Overhand, Cross Body Reach). Prior to the test, subjects will be trained on each movement and allowed to repeat it as many times as they desire. For each movement, the 12 repetitions will be further subdivided into 3 groups of 4 repetitions each. This is done to evaluate subject fatigue or potential change of biomechanical strategies over the course of the test period. After each group of movements, qualitative information on subject comfort and hot spots will be collected. The information will also be collected after training. Each of these test conditions will be counterbalanced and randomized for each subject. Unsuited data will also be collected after the suited test to form the baseline pressure profile used to mitigate the effects of erroneous readings caused by movement without contact with the suit.

1.3 Experiment Pilot Study

The outlined procedures will be performed in alternative pressure suit environments. Two environments will be used. The first is in a vacuum chamber in the Man Vehicle Lab at MIT with space suit components. The second is at the David Clark Company inside one of their experimental space suits. In the David Clark series of tests, this protocol will be replicated and evaluated for shortcomings. Any hardware or data collection issues will be noted and addressed. It will also serve to allow the experimenters additional data with which to test their analysis techniques and ensure the JSC tests will be performed more smoothly.

2.0 TEST HARDWARE

 Mark III Suit  Liquid cooling garment, thermal comfort undergarment, wristlets, comfort gloves, socks  Space Suit Audio Communication Interface System II (SPACIS II) Integrated communication system with in-suit microphones & speakers  Standard Mark III Suit Donning Stand  PGA Test Stand  ASL Chiller Cart  Novel Pressure sensing system o Single S2073 mat with 128 sensor locations o 12V battery pack and Pliance data acquisition hardware  Polipo Pressure sensing system o 12 hyperelastic eutectic gallium indium (eGaIn) pressure sensors o Arduino Micro controller and custom data shield o Athletic shirt with elastic loop material for sensor mounting  APDM Opal IMU sensing system o 4 Low profile 3-axis IMUs, 3 mounted internal to the suit, 3 mounted externally o 3.7V Lithium Ion battery  Two GoPro cameras and mounts

3.0 TEST PERSONNEL

 Test Conductor  Suit Technicians  Suit Test Engineer  Sensory systems specialists (1 or 2)  Suited test subject

4.0 SUIT AND TEST CONFIGURATIONS

The suit hardware and ancillary support equipment provide the necessary functions and interfaces to conduct manned pressurized suit operations when combined with (a) a suitable gas supply system, (b) cooling water supply and (c) suitable communication system.

 Mark III Suit o 0-8.3psid (4.3 psid for this test) o No PLSS mockup will be used for this test o Receiving certified breathing air at 6 ACFM (actual cubic feet per minute) from bottle trailers outside B7 through PGA Test Stand o Receiving cooling via cooling line pass-thru located at the connector on the rear entry door, which allows the modified shuttle liquid cooling garment (LCG) cooling lines to be routed through the fitting o Communications provided by Space Suit Audio Communication Interface System II (SPACIS II), which is the primary external comm. interface for space suit testing o The donning stand secures the suit at the waist ring to allow for rear entry o Ancillary equipment including modified Shuttle LCVG (Liquid Cooling Ventilation Garment); TCU (Thermal Comfort Undergarment)

Figure 1: Mark III Suit and Support Equipment

 Novel Pressure Sensing System o One commercially available Novel pressure mat, S2073 with 128 sensors o Each sensor is 1.4cm in each dimension and pressure range between 20-600kPa o Mat slips into pocket over shoulder o On-board data collection with electronics mounted at the base of the back o Similar system used inside the extravehicular mobility unit by the Anthropometry and Biomechanics Facility without modification for a shoulder load study o Sensor runs at 330mA current o Battery is 10 1.2V nickel metal hydride and encased in housing making it moisture resistant under normal EVA conditions o The system is certified to the European safety standard 93/42/EEC (Annex 1X). o The pedar NiMH 2000mAh battery pack is internally secured with an overheating and an overcurrent protection (Polyswitch). o Puncture or electrical shock is unlikely, resulting in a maximum current of 8 - 20 mA(pp).

Figure 2: Novel Pressure sensing system. A) Sensor mat. B) Data acquisition and battery

 Polipo Pressure Sensing System o 12 developmental sensors distributed over the arm o Detachable system transferrable between subjects with velcro o On-board data collection with electronics mounted at the base of the back

o Each sensor powered with constant current of .5mA . Microcontroller shown; new board will be fabricated (not shown) o The entire board in nominal operation with 12 sensors is estimated to be around 100mA o Battery selection is TBD but at the moment may be an off the shelf 9V battery. A typical 9V has about 500mAh, so we are estimated to have 4 hours of use o Shorting the sensor wires of one circuit will result in ~.5mA through the short o Electrical shock is unlikely from a worst case scenario short, giving ~6mA current draw

Figure 2: Polipo pressure sensing system. A) Sensors mounted on sleeve. B) Microcontroller. Arduino shield not shown

 APDM IMU Sensing System o Commercially available APDM Opal inertial measurement unit (IMU) o 3 internally mounted sensors on the upper and lower arm and chest o 3 externally mounted sensors: upper and lower spacesuit arm and suit torso o Each is 4.8x3.6x1.3 cm (lxwxh) and weighs less than 22g o Lithium Ion battery at 3.7V nominal o Maximum current seen through the sensor is approximately 56 mA, and battery failure is highly unlikely.

Figure 4: Opal IMU sensor from APDM

5.0 DETAILED TEST PLAN

5.1 Pressure Sensing Measurement

The human-suit interface is currently an unknown in space suit characterization. Pressure measurements would allow greater insight into how these interactions occur and help characterize suit performance. Additionally it would allow us to prevent injury incurred by motion inside the suit. There is currently no

82 method by which to characterize this pressure. The two systems selected to measure pressure are targeted at different pressure sensing regimes.

The pressure sensing system is integrated into one conformal athletic garment. This shirt is worn by the subject and has targets over which the low pressure sensors are mounted. The Polipo, or octopus in Italian, is the system of 12 sensors which were developed as part of this research effort for low-pressure sensing under the soft goods. These sensors are placed over the arm in a way that targets anticipated hot spots, and secondarily for uniform coverage. The sensors are detachable from the athletic garment, allowing independent pressure sensing system. It also allows for shifting the sensors to concentrate them over a certain region of the body. The garment has a pocket interface over the shoulder to house the Novel pressure sensor, which is used for the high-pressure sensing regime. The high pressure regime is at the interface between the person’s body and the hard upper torso of the suit. A Novel pressure sensing mat has been used previously in a study by the Anthropometry and Biomechanics Facility (ABF) on an Extravehicular Mobility Unit hard upper torso. Finally, a cover shirt slides easily over the sensors to prevent catching and ensure proper sensor placement.

5.2 IMU Sensing Measurement

Additional information about the human-suit interface may be gathered using IMU data collected inside the space suit. There is a joint angle difference between the person’s movement and that of the space suit. This is due to the resistance of a gas pressurized suit to movement, as well as (in some instances) anatomically inaccurate rotation due to bearing movement. Calculating the joint angles measured internal and external to the suit would help elucidate these differences. Previous studies performed by the ABF and researchers at the University of Maryland have evaluated the use of IMUs inside a gas pressurized space suit. This experiment uses similar methods, mirroring a previous study performed by our research group inside a gas pressurized suit at David Clark Company. This data will be used not only to determine biomechanical differences but also to help find points of maximum and minimum movement to analyze the pressure profiles. It will be matched with video data to improve the results. Sensors will be placed on the lower and upper arm of the subject, not in contact with the pressure sensors. An additional chest mounted IMU will serve as a reference for shoulder rotation data. Three sensors will be placed external to the suit, two on the upper and lower arm and one on the suit upper torso.

5.3 Functional and Isolated tasks and supporting equipment

The following tasks will be performed by all subjects, both suited and unsuited as described in the configurations above. The first time the task is performed suited, each task will be repeated for two minutes, or through 4 repetitions, whichever comes first. After each of the 5 movements is performed, the subject will rest for 5 minutes and subjective data will be collected. The subject will then repeat the movement sequence and rest period two additional times.

Below are the tasks the subject will be performing in this test campaign.

Elbow Flexion/Extension

The subject stands away from the donning stand supported by their own effort. Beginning with both arms relaxed at their side, palms facing anterior, the subject bends the arms at the elbow through their maximum range of motion. The subject then releases to the relaxed position.

Shoulder Flexion/Extension

The subject stands away from the donning stand supported by their own effort. Beginning with both arms relaxed at their side, the subject bends the arms at the shoulder through the sagittal plane. The subjects move through their maximum range of motion. The subject then releases to the relaxed position.

Shoulder Abduction/Adduction

The subject stands away from the donning stand supported by their own effort. Beginning with both arms relaxed at their side, the subject bends the arms at the shoulder through the coronal plane. The subject moves through his or her maximum range of motion. The subject then releases to the relaxed position.

Cross Body Reach

The subject stands away from the donning stand supported by their own effort. Beginning with both arms relaxed at their side, the subject will reach across their body in an attempt to touch their hip on the opposite side. The subject will then move their arm up to chest level and sweep their arm in front of their body in the horizontal plane. When the arm is extended straight in front of the shoulder, the subject will then attempt to touch the helmet at the position of their ear on the same side. The movement is then repeated with the opposite arm.

Overhead Hammering

The subject stands away from the donning stand supported by their own effort. Subjects will be given a rubber mallet to be grasped with both hands. The subject will be instructed to hammer flat 7” rubberized square pads, which are fixed to a repurposed ergometer stand, using the mallet. The pads will be fixed in a horizontal orientation at 39” height and the subject will hammer with both hands beginning the movement overhead and ending at the height of the stand (approximately waist level).

Table 1 shows a summary of all functional tasks each run will consist of, and a cumulative time for the run as currently scheduled. Each subject will perform these tasks in the order specified.

Minute Task Type Description Min Count

Elbow Isolated Stand free of donning stand and bend elbow 2 2 Flexion/Extension

Shoulder Isolated Stand free of donning stand and bend shoulder 2 4 Flexion/Extension

Shoulder Isolated Stand free of donning stand and bend shoulder 2 6 Abduction/Adduction

Stand at task table and hammer rubber plate with both Overhead Hammer Functional 2 8 hands

Stand free of donning stand and reach from overhead Cross Body Reach Functional 2 10 across the body, alternating arms

Rest mounted in donning stand. Qualitative Rest 5 15 information collected Table 1: Test variables matrix

Each of the three subjects will perform this series of tasks identified in Table 1 two different ways:

 Suited in Mark III Suit at a natural pace  Unsuited at a pace matching that of Mark III run The order of these tasks will vary between subjects, but the matched-pace unsuited run will always be completed last. In addition, some familiarization time is built into the test plan for each suit to allow the

84 subject to become comfortable performing each task in the suit he or she has just donned. Not only will this make the subject more comfortable and safe while performing the tasks, but it will also reduce the possibility of familiarization of a task negatively affecting the outcome of the test. Test condition order is shown in Appendix A.

5.5 Subject Selection

Three test subjects will be selected from the pool of available candidates who have been fitchecked in the Mark III Suit. The criteria for suited subject selection for this test are as follows:

 Current fitcheck in Mark III Suit  Current test subject services approval (physical)  Extensive experience working in the pressurized suit to aid in comfort and stability while performing identified functional tasks

5.6 Test and Trial Termination Criteria

Test Termination Criteria:

 At any time for any reason by test subject or test personnel  Unrecoverable suit or hardware failure

This test may be terminated by the subject at any time for any reason, or by the test conductor, suit technicians or suit engineer due to any safety or hardware concerns or concern for the suited subject. Between movement groups, subjective data will be taken from the subject. This will be used as an indicator of subject fatigue and desire to terminate the test. An outline of the questions to be asked is shown in Appendix B.

The test will also be terminated in the event of unrecoverable suit system failure. Standard Advanced Suit Lab procedures will be followed regarding the failure of any suit system part, or any suit emergency.

6.0 DETAILED TEST PROCEDURES

6.1 Test-Specific Pre-Test Safety Briefing

1. Anyone can stop this test at any time for any reason

2. Test personnel: Manage video camera, extension cords and functional task props at all times.

3. Suited Subject: We will ask you how you’re feeling between each task, absent any other reports from you. After each series of 5 tasks, which will last approximately 2 minutes each for a total of 10 minutes, you will rest for at least 5 minutes.

6.2 Detailed Test Procedure (Suited Run)

1. Review summary of test with subject

2. Conduct test-specific pre-test safety briefing

3. ______Explain IMU calibration movements to the subject

4. ______Explain functional task movements to the subject and allow them to practice

5. ______Turn on video cameras

6. Test personnel places IMUs on the subject’s arm and body and notes location on the body. IMU 1: Chest ______IMU 2: Upper Arm_____ IMU 3: Lower Arm _____

7. Subject dons pressure sensing systems

8. Pressure sensing systems are turned on and data collection is initiated

9. ______Subject dons cover shirt

10. ______Subject dons LCG

11. Subject performs IMU calibration movements (4 repetitions)

a. ____ Wrist pronation/supination to 180 degrees

b. ____ Elbow flexion/extension to 90 degrees

c. ____ Shoulder flexion/extension to 90 degrees

d. ____ Shoulder abduction/adduction to 90 degrees

12. ______Data is collected for 30 seconds to obtain baseline resting values for sensors

13. Subject dons suit per CTSD-ADV-197 (Mark III)

14. ASL technicians pressurize suit to 4.3 psi at 6 ACFM (gauge indicated: 10 SCFM) flow rate per CTSD-ADV-197. Instruct suited subject to rest during this time as much as possible

15. Test personnel places IMUs on the suit’s arm and body and notes location on the body.

16. Subject performs IMU calibration movements (5 repetitions)

a. ____ Wrist pronation/supination to 180 degrees

b. ____ Elbow flexion/extension to 90 degrees

c. ____ Shoulder flexion/extension to 90 degrees

d. ____ Shoulder abduction/adduction to 90 degrees

Resting Data Collection

1. ASL technicians assist subject in moving from donning stand to functional task area

2. ______Call out global time

3. ______Data is collected for 30 seconds to obtain baseline resting values for sensors

NOTE: Instruct subject to complete these tasks at what they consider to be a natural pace

NOTE: Request a report of any symptoms from suited subject after each task and ask qualitative questions from Appendix B during the 5-minute rest periods.

Suited Familiarization Session

1. ______Call out global time

2. Subject practices elbow flexion/extension

3. Subject practices shoulder flexion/extension

4. Subject practices shoulder abduction/adduction

5. Subject practices hammering task

6. ______Subject practices cross body reach task

7. ______Call out global time

NOTE: Subject may return to donning stand for rest if necessary at any point

8. Subject returns to donning stand for rest (at least two minutes)

9. ______Subjective comfort and fatigue data collected from Appendix B.

Suited Data Collection Run

NOTE: The subject completes 4 repetitions of the task. Time is not limited, but the task may be terminated if the subject is unable to complete 4 repetitions.

NOTE: Five minute break (minimum) is enforced between each group of movement tasks. Allow suited subject to take additional rest time as needed.

NOTE: Subject task order is counterbalanced for each subject and each movement run. The task order is provided in Appendix A.

10. ASL technicians assist subject in moving from donning stand to task area

11. ______Call out global time

12. Subject performs elbow flexion/extension task

a. Time to completion: 1) ______2) ______3) ______

b. ______Prompt for subjective feedback

13. ______Call out global time

14. Subject performs shoulder flexion/extension task

a. Time to completion: 1) ______2) ______3) ______

b. ______Prompt for subjective feedback

15. ______Call out global time

16. Subject performs shoulder abduction/adduction task

a. Time to completion: 1) ______2) ______3) ______

b. ______Prompt for subjective feedback

17. ______Call out global time

18. Subject performs cross body reach task

a. Time to completion: 1) ______2) ______3) ______

b. ______Prompt for subjective feedback

19. ______Call out global time

20. Subject performs overhead hammering task

a. Time to completion: 1) ______2) ______3) ______

b. ______Prompt for subjective feedback

21. ASL technicians assist subject in returning to the donning stand

22. ______Subjective comfort and fatigue data collected from Appendix B

23. ______Data collection is repeated twice

24. ______Return to CTSD-ADV-197 (Mark III) for suit depressurization and doffing

25. ______External IMUs removed from suit

26. ______Subject remains in LCVG with pressure sensors and IMUs in place

6.3 Detailed Test Procedure (Unsuited Run)

1. Review summary of test with subject

2. Conduct test-specific pre-test safety briefing

3. Test personnel confirms by feel the IMU placement on the subject’s arm and body and notes any changes

a. IMU 1: Chest ______

b. IMU 2: Upper Arm______

c. IMU 3: Lower Arm______

4. Subject performs IMU calibration movements (4 repetitions)

a. ____ Wrist pronation/supination to 180 degrees

b. ____ Elbow flexion/extension to 90 degrees

c. ____ Shoulder flexion/extension to 90 degrees

d. ____ Shoulder abduction/adduction to 90 degrees

5. ______Call out global time

6. ______Data is collected for 30 seconds to obtain baseline resting values for sensors

NOTE: Request a report of any symptoms from suited subject during the 2-minute rest periods if subject does not provide comments

Unsuited Familiarization Session

NOTE: Instruct the subject to try and recreate the movements they performed inside the pressurized space suit. Provide subject with aids such as task repetition interval (“You completed this task in approximately X seconds per repetition”) to help them best match suited Mark III run cadence.

1. ______Call out global time

2. Subject practices elbow flexion/extension

3. Subject practices shoulder flexion/extension

4. Subject practices shoulder abduction/adduction

5. Subject practices hammering task

6. ______Subject practices cross body reach task

7. Subject returns to resting area (at least two minutes)

8. ______Subjective comfort and fatigue data collected from Appendix B

Unsuited Data Collection Run

7. ______Subject is moved in line with both GoPro cameras

8. ______Call out global time

9. Subject performs elbow flexion/extension task

a. Time to completion: 1) ______

b. ______Prompt for subjective feedback

10. ______Call out global time

11. Subject performs shoulder flexion/extension task

a. Time to completion: 1) ______

b. ______Prompt for subjective feedback

12. ______Call out global time

13. Subject performs shoulder abduction/adduction task

a. Time to completion: 1) ______

b. ______Prompt for subjective feedback

14. ______Call out global time

15. Subject performs cross body reach task

a. Time to completion: 1) ______

b. ______Prompt for subjective feedback

16. ______Call out global time

17. Subject performs overhead hammering task

a. Time to completion: 1) ______

b. ______Prompt for subjective feedback

18. ______Subjective comfort and fatigue data collected from Appendix B

19. _____ Subject returns to non-testing area for sensing system doffing and debrief

APPENDIX B: ON-BOARD STORAGE FOR NOVEL

On-board storage (or “offline measurement”/”monitor mode” as referred to in the Novel User’s Guide) is an important capability of the Novel system that allows for continuous data collection in the case that neither a hardwired or Bluetooth connection to a computer is feasible. In this mode, the data transmission is controlled by the start/stop trigger and data is stored onto an SD card (maximum size of 2 GB).

Editing the Total Duration of Data Collection

First, you should edit your mat configuration file (the file that you get after calibrating your sensor) to be sure that it will record data for the entire duration that you intend to be testing. The default duration is 1 hour, so if you intend to collect data for longer than this, follow these steps:

1. In the main screen of the Novel software, select your mat configuration file from the “mat configuration” dropdown menu (next to the “confirm” button). This should bring up an image of your sensor in the left half of the screen. 2. Under the “Expert Settings” menu, click “Edit selected mat configuration”. 3. In the window that opens, look for the “Timing of Measurement” section and change the value of the “Buffer size in seconds” field to your desired total duration, in seconds. For example, for data collection of up to 4 hours, enter the value 14400. 4. At the bottom of the same window, be sure that “monitor mode” is selected. Click OK. Back on the main screen, you should now see (in small white text at the top of the sensor image) that the duration has changed to what you entered. Configuring the SD card

Now you must configure the SD card, using the following steps:

1. Insert SD card to computer and open Novel software 2. Under the “Data Acquisition” menu, click “Configure SD card”. In the window that opens, the table at the bottom of the window shows files that are currently on the card. Note that upon formatting the SD card, all previous contents will be lost, so be sure you have saved any necessary files elsewhere. 3. To delete any existing files on the SD card, click “Delete only data files”. 4. Under “Measurement Configuration” select the mfg file that you want to use. 5. Click the “Format” button. 6. In the pop-up window, choose “FAT32” for the format type, select “Quick Format”, and click “Start”. 7. Click the “Configure” button. This will populate the table at the bottom of the window with the files that have been transferred to the SD card. 8. Click “Close”. Your SD card is now configured and ready for data collection. Collecting Data

To collect data with your configured SD card, follow these steps:

1. Insert the SD card into the slot on the left side of the Pliance box.

2. Once the sensor, battery, and start/stop trigger are all connected to the Pliance box, turn on the Pliance box using the switch at the top. Give the system about 10 seconds to warm up. 3. To start data collection, push the trigger button. You will know that data collection has begun if the trigger button lights up green and the SD card light on the Pliance box is rapidly flickering on and off. Note: If you are using a single sensor mat (i.e. not the foot insoles), it should take 3 pushes of the trigger button to initiate data collection. The reason for this is described in the User’s Guide, and has to do with the case of using the foot insoles. 4. Disconnect the trigger from the Pliance box. This will not stop or affect data collection. While not necessary, this is recommended so that the subject doesn’t somehow accidentally push the button and stop data collection during the test. 5. To stop data collection, re-connect the trigger (it will light up upon doing this) and press its button. You will know that data collection has stopped if both the trigger button and the SD card light on the Pliance box turn off. 6. If done testing, turn off the Pliance box. Post-Test: Looking at the Data

After a test, you will want to verify that you have collected data.

1. Insert SD card to computer and open Novel software. 2. Under the “File” menu, click “Open SD card”. 3. In the box below the sensor display, you will see several files listed. From the dates and time stamps (dates listed in DD.MM.YYYY format) and the duration of the file, you should be able to tell which one corresponds to your data collection period. Select the desired file and click the “Show” button. This will bring up your data in the main display. 4. Scroll through your data to ensure that it seems to have worked appropriately. Hopefully you will be able to see changes in pressure, force, and area across the duration of your test. After verifying that you have collected data, you should save that data to a secondary location other than the SD card, just for back-up. To save the raw data from the SD card, you will simply want to copy the data files (they will be named: #+4 digits without any file extension, e.g. #0001, #0002, etc.) and the mat configuration file (.mfg) into a directory on another drive.

If you want to save your data in a tab delimited format that can be opened in Excel, Notepad, or other program, follow these steps:

1. Under the “File” menu, click “Save as”. 2. In the window that appears, select your file format. The file format options are described under Section 9.3 “Pliance Exported Files” of the Novel User’s guide. Read through this to determine which option works best for the type of data analysis you have in mind. For my purposes I primarily used the ASC format, although MVA would have been useful as well.

APPENDIX C: SYNCHRONIZATION WITH NOVEL

A preliminary synchronization capability was developed to enable data synchronization between the Novel sensor and the APDM Opal inertial measurement units (IMUs). For the purposes of the human-suit interaction experiment, it was not an option to keep the two systems connected while the test was being performed, since that would have meant that wires would have had to be fed out of the suit and attached to external devices. As a result, the current synchronization system was set up such that it only allows for initial synchronization – in other words, the initiation of data collection in both systems is synchronized, but there was not continuous synchronization throughout the experiment, nor was the end of data collection synchronized. Data from the first full experiment using the synchronization device (David Clark test, April 2015) has not yet been analyzed to see if the two data streams appear to be appropriately aligned. Either way, this capability should be improved upon to enable a more reliable and useful synchronization capability.

This tutorial describes how to send the synchronization signal from the Novel to the IMUs, as opposed to vice-versa. Refer to both systems’ User’s Guides if you need the IMUs to trigger the Novel.

Novel Synchronization Capability

The Novel system owned by the Man-Vehicle Lab comes with the necessary elements to perform wired synchronization (not wireless). A description of the wired synchronization components provided by Novel are presented in Subsection 5.1.1 of the “Pliance-X Sync Boxes and Software” section of the User’s Guide. An explanation of how to connect the sync box to the Pliance box is shown in Subsection 5.1.5. To summarize, the components should be connected as shown in the diagram at right (taken from the Novel User’s Guide). In the image, you can see (as indicated by the red circle) that the sync cable simply comes with wires and pins. While some measurement systems such as the APDM Opal IMUs come with standard synchronization connectors (e.g. BNC or RCA cables), the Novel sync cable simply comes with open wire ends that must be soldered to custom-made connectors made to match with whichever system it is being synchronized with. Thus although we created connectors in order to connect to the IMUs, it would require different connectors if you wanted to sync the Novel with a motion capture system or other device.

If you intend to synchronize with the IMUs, you should be able to use the connector we created, shown at right. To create our connector, BNC cables were soldered to each of the “sync in” and “sync out” wires of the Novel sync cable (the green and yellow wires, respectively), and BNC gender changers were used to connect from this newly created cable to the IMU

93 synchronization cables. A photo showing all connections between the Novel and IMU system is shown below under “Initiating Synchronized Data Collection of the Novel and APDM IMUs”.

If you are trying to connect with a system other than the IMUs, you may have to create a new connector or build off of ours. If you are building a new connector, refer to the diagram at right for a description of each of the sync cable wires (taken from Section 5.1.3 of the User’s Guide). You should read all of Section 5 “Pliance-X Sync Boxes and Software” in the User’s Guide to obtain a full understanding of the necessary connections.

Configuring the Novel for Synchronization

To prepare the Novel for synchronization using the current capability, follow these steps:

5. In the main screen of the Novel software, select your mat configuration file from the “mat configuration” dropdown menu (next to the “confirm” button). This should bring up an image of your sensor in the left half of the screen. 6. Under the “Expert Settings” menu, click “Edit selected mat configuration”. 7. In the window that opens, go to the “Sync Setup” tab. In the “Trigger output” section, select “first picture”, a pulse width of “10 ms” and a pulse polarity of “negative”. In the “Trigger input” section, select “none”. Click OK. You have now edited the mat configuration file to allow for a sync out capability. Selecting “first picture” option simply means that a trigger output signal is sent with the first frame of data collected by the Pliance. The other two settings are selected to match the synchronization set-up of the IMUs. 8. If you will be using on-board storage and have already configured your SD cards for that, you will need to re-configure the cards with this updated mat configuration file. Initiating Synchronized Data Collection of the Novel and APDM IMUs

To initiate synchronized data collection of the two systems, follow these steps:

1. Set-up each of the data collection systems. For example, if you are doing on-board storage, insert the SD card into the Pliance box and have all components connected appropriately. For the IMU system, have all connections made as well. 2. After ensuring that Motion Studio (the APDM IMU software) is closed, connect the IMUs and the Novel system through the sync box. The appropriate connections are shown in the image at right. Note that the connections shown are in addition to the ones shown in the first image of this Appendix, as well as the required IMU system connections. 3. Once everything is connected as shown above, turn on the Pliance box and the sync box.

4. Open Motion Studio. Under the “Tools” menu, click “Setup External Synchronization”. 5. In the window that appears, select the following “Input Trigger” options: Shape – “Edge”, Level – “High”, Trigger - “Start”. Be sure that the “Output Trigger” is disabled. Click OK. You have now configured the IMUs to receive an input trigger from the Novel.

6. On the main screen of Motion Studio, click “Stream”. You should now see a new window although data will not be recording yet. 7. Push the Novel start/stop trigger to begin simultaneous recording of both systems. You will know that this has worked if the SD card light on the Pliance box begins to flash rapidly and if Motion Studio also shows that it has started recording. 8. You can disconnect both systems from the synchronization cables without interrupting data collection. To do this for the Novel, simply unplug the fiber optic cables from the Pliance box.

APPENDIX D: DATA PROCESSING IN MATLAB

For the purposes of the human-suit interaction experiment, use of on-board storage meant that data files included nearly 3 hours of pressure data. It was therefore necessary to create code that would process the data and extract the time periods of interest. Data processing was performed in MATLAB.

The methods and MATLAB code referred to here are simply the way I chose to do this. Improvements could certainly be made to the efficiency of this code, particularly if it were to be automated. It is often difficult to use code written by someone else, so future users of this experiment should perform data processing in whichever way suits them best.

This appendix provides brief descriptions of the code included in my repository. All code referred to here can be found on the Man-Vehicle Lab’s EVA server, under “Past Students” -> “Alexandra Hilbert”.

 “Timer.m” – This is a timing tool used during the experiment. It allows the user to keep track of when the subject is performing the desired tasks. After launching the code, prompts will appear in the Command Window. By pressing “Enter”, you store the time of the prompt. By pressing “0”, you skip the prompt. Be sure to save the workspace at the end of an experiment, so you can come back and use the timing information later.

 “filter_subj1.m” – This code takes the large 3-hour data file and separates it into the time periods of interest. To do so, the user must enter the time stamps of the start and end of each movement group and each set of 4 repetitions of a particular motion. These values will come from your timing tool. (Note: This code would benefit from automation.)

 “max_pressures.m” – After running “filter_subj1.m”, you can use this code to create the maximum pressure curves for all of the motions of interest. There is also a section that allows for removal of any pressure discontinuities that may have been found.

 “max_pressure_ind.m” – This code is a precursor to determining the locations of maximum pressure that were shown in Section 3.3.2.2 of this thesis. For a particular set of repetitions, it identifies all sensiles on the mat that experience maximum pressure at some point during this motion. An example of the output is shown below.

 “normalizing_S1.m” – This code takes the maximum pressure curves from “max_pressures.m” and normalizes each movement group with respect to time.

 “peak_locations_freq_S1.m” – This code creates the diagrams of locations of maximum pressure that were shown in Section 3.3.2.2 of this thesis. Values can be adjusted to make the green, yellow, and red signs signify different frequency thresholds.

 “peak_maps_S1.m” – This creates the pressure distribution diagrams used in Section 3.3.2.1 of this thesis.

 “profiles_ind.m” – This creates the pressure profile graphs showing the general trends of sensor response and corresponding sensor mat diagrams shown in Section 3.3.2.3 of this thesis. An example is shown below.

 “loaded_sensors.m” – This creates a diagram that indicates which sensors were loaded over the course of the experiment, as shown at right. This can be useful in determining if placement of the sensor mat was appropriate.

 “max_pressures_regional_anthro_withShBlade.m” – This allows the user to extract the peak pressures that arise within various regions of the sensor. It is currently set up for subject-specific anthropometric regions, and must therefore be adjusted for a new data set data set.

APPENDIX E: TOTAL FORCE BY MUSCLE GROUP

Total Forces for Shoulder Abductors – All 8 repetitions shown

Total Forces for Shoulder Adductors – All 8 repetitions shown

100

Total Forces for Shoulder Stabilizers – All 8 repetitions shown

101

102

APPENDIX F: INDIVIDUAL MUSCLE FORCES

Individual Forces for Shoulder Abductors – All 8 repetitions shown

103

104

105

Individual Forces for Shoulder Stabilizers – All 8 repetitions shown

106

107

108