Special Topics in Structural Dynamics, Volume 6

River Rapids Conference Proceedings of the Society for Experimental Mechanics Series Special Topics in Structural Dynamics, Volume 6 Randall Allemang James De Clerck Christopher Niezrecki Alfred Wicks Proceedings of the 31st IMAC, A Conference on Structural Dynamics, 2013 River Publishers

Conference Proceedings of the Society for Experimental Mechanics Series Series Editor TomProulx Society for Experimental Mechanics, Inc., Bethel, CT, USA

River Publishers Randall Allemang • James De Clerck • Christopher Niezrecki • Alfred Wicks Editors Special Topics in Structural Dynamics, Volume 6 Proceedings of the 31st IMAC, A Conference on Structural Dynamics, 2013

Published, sold and distributed by: River Publishers Broagervej 10 9260 Gistrup Denmark www.riverpublishers.com ISBN 978-87-7004-878-1 (eBook) Conference Proceedings of the Society for Experimental Mechanics An imprint of River Publishers © The Society for Experimental Mechanics, Inc. 2013 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, or reproduction in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Preface Special Topics in Structural Dynamics Volume 6: Proceedings of the 31st IMAC, A Conference on Structural Dynamics, 2013 represents one of seven volumes of technical papers presented at the 31st IMAC, A Conference and Exposition on Structural Dynamics, 2013 organized by the Society for Experimental Mechanics, and held in Garden Grove, California February 11–14, 2013. The full proceedings also include volumes on Nonlinear Dynamics; Experimental Dynamics Substructuring; Dynamics of Bridges; Dynamics of Civil Structures; Model Validation and Uncertainty Quantification; and, Modal Analysis. Each collection presents early findings from experimental and computational investigations on an important area within Structural Dynamics. Special Topics in Structural Dynamics represents papers on enabling technologies for Modal Analysis measurements such as Sensors & Instrumentation, as well as applications of Modal Analysis in specific application areas. Topics in this volume include: Teaching Experimental & Analytical Structural Dynamics Sensors & Instrumentation Aircraft/Aerospace Bio-Dynamics Sports Equipment Dynamics Advanced ODS & Stress Estimation Shock & Vibration Full-Field Optical Measurements & Image Analysis Structural Health Monitoring Operational Modal Analysis Wind Turbine Dynamics Rotating Machinery Finite Element Methods The organizers would like to thank the authors, presenters, session organizers, and session chairs for their participation in this track. Cincinnati, OH, USA Randall Allemang Lowell, MA, USA Christopher Niezrecki Houghton, MI, USA James De Clerck Blacksburg, VA, USA Alfred Wicks v

Contents 1 Safety Improvement of Child Restraint System by Using Adoptive Control ..................................... 1 Takayuki Koizumi, Nobutaka Tsujiuchi, and Shin Ito 2 Dynamic Response and Damage Estimation of Infant Brain for Vibration....................................... 11 Takayuki Koizumi, Nobutaka Tsujiuchi, Keisuke Hara, and Yusuke Miyazaki 3 Mechanical Strength of Bone Cement with and Without Adjuvant Screw Fixation............................. 19 Ryan Keyser, Robert Migliori, Tessa Morgan, Steven R. Anton, Kevin M. Fariholt, and R. Michael Meneghini 4 Development of a Bench for Testing Leg Prosthetics ................................................................ 35 H. Giberti, F. Resta, E. Sabbioni, L. Vergani, C. Colombo, G. Verni, and E. Boccafogli 5 Application of Modal Testing and Analysis Techniques on a sUAV................................................ 47 Kaci J. Lemler and William H. Semke 6 Progress in Operational Analysis of Launch Vehicles in Nonstationary Flight................................... 59 George James, Mo Kaouk, and Tim Cao 7 Influence of Test Conditions on Comfort Ranking of Road Bicycle Wheels ...................................... 77 Julien Le´pine, Yvan Champoux, and Jean-Marc Drouet 8 Direct Measurement of Power on a Gravity Independent Flywheel-based Ergometer.......................... 83 F. Braghin, M. Bassetti, P. Crosio, and D. Locati 9 Instrumented Treadmill for Cross-Country Skiing Enhanced Training .......................................... 87 M. Bassetti, F. Braghin, and S. Maldifassi 10 Instrumenting a Rowing Ergometer for Improved Training ....................................................... 93 G. Cazzulani, M. Bassetti, G. Picardi, L. Mariella, J. Verdonkschot, A. Benecchi, and Dario Dalla Vedova 11 A Laboratory Technique to Compare Road Bike Dynamic Comfort .............................................. 99 Yvan Champoux, Julien Le´pine, Philippe-Aubert Gauthier, and Jean-Marc Drouet 12 Exploring Experimental Structural Dynamics in EMA/ME 540 at UW-Madison............................... 107 Matthew S. Allen 13 The ABRAVIBE Toolbox for Teaching Vibration Analysis and Structural Dynamics .......................... 131 Anders Brandt 14 Structural Dynamics Teaching Example: A Linear Test Analysis Case Using Open Software................. 143 Per-Olof Sturesson, Anders Brandt, and Matti Ristinmaa 15 Testing Anti-Ram Barrier Protection Systems ....................................................................... 155 Kurt Veggeberg vii

viii Contents 16 Fiber Optic Accelerometers and Sensors for Dynamic Measurements............................................ 161 Kurt Veggeberg 17 Nonlinear Model Tracking for Varying System Geometries........................................................ 167 Timothy A. Doughty, Matthew R. Dally, Mikah R. Bacon, and Nick G. Etzel 18 Fuzzy Arithmetical Assessment of Wave Propagation Models for Multi-Wire Cables .......................... 177 Christoph Schaal and Michael Hanss 19 A Vibro-Haptic Human-Machine Interface for Structural Health Monitoring Applications................... 187 Christina Brown, Martin Cowell, C. Ariana Plont, Heidi Hahn, and David Mascaren˜as 20 Technologies for Seismic Safety Management of Existing Health Facilities ...................................... 199 C. Rainieri and G. Fabbrocino 21 Wave-Induced Vibration Monitoring for Stability Assessment of Harbor Caisson.............................. 207 So-Young Lee, Thanh Canh Huynh, Han-Sam Yoon, Jeong-Tae Kim, and Sang-Hun Han 22 Damage Assessment of a Beam Using Artificial Neural Networks and Antiresonant Frequencies............. 217 V. Meruane and J. Mahu 23 Case Studies of Tools Used in Teaching Structural Dynamics...................................................... 225 Kurt Veggeberg 24 “Structural System Testing and Model Correlation”: An Industry-University Collaborative Course in Structural Dynamics ........................................................................................ 233 Michael Todd, Dustin Harvey, David Gregg, Bill Fladung, Paul Blelloch, and Kevin Napolitano 25 Visualizing Structural Vibrations Using Stroboscopic Light in a Novel Setup ................................... 241 Markus J. Hochrainer 26 Analytical and Experimental Learning in a Vibrations Course at the University of Massachusetts Lowell .. 249 Pawan Pingle and Peter Avitabile 27 Around the World in 80 Courses....................................................................................... 265 David Ewins 28 Review of a Pilot Internet System Dynamics Course ................................................................ 271 C.C. Claeys, S. Leuridan, D. Brown, and J. Connor 29 Using Random Response Input in Ibrahim Time Domain.......................................................... 281 Peter Olsen and Rune Brincker 30 Modal Parameter Identification of New Design of Vertical Axis Wind Turbine.................................. 289 Prasad D. Chougule and Søren R.K. Nielsen 31 Predicting Dynamic Strain on Wind Turbine Blade Using Digital Image Correlation Techniques in Conjunction with Analytical Expansion Methodologies ......................................................... 295 Jennifer Carr, Javad Baqersad, Christopher Niezrecki, Peter Avitabile, and Micheal Slattery 32 Dynamic Characterization of a Free-Free Wind Turbine Blade Assembly ....................................... 303 Javad Baqersad, Christopher Niezrecki, Peter Avitabile, and Micheal Slattery 33 Harmonic Analysis on a Le´vy Plate and Its Application to Fatigue Analysis..................................... 313 Nam-Gyu Park, Jung-Min Suh, and Kyeong-Lak Jeon 34 Vibration Level Assessment of Nuclear Power Plant Powerhouse Hall ........................................... 321 G.G. Boldyrev and A.A. Zhivaev 35 Study on the Band Structure of Trigonal Chiral Structures........................................................ 329 Shiyin Xu, Xiuchang Huang, and Hongxing Hua 36 FEM Sensitivity Vector Basis for Measured Mode Expansion ..................................................... 339 Robert N. Coppolino

Contents ix 37 Estimation of Unmeasured DOF’s on a Scaled Model of a 4-Storey Building.................................... 347 Anders Skafte and Rune Brincker 38 Estimation of Rotational Degrees of Freedom by EMA and FEM Mode Shapes ................................ 355 A. Sestieri, W. D’Ambrogio, R. Brincker, A. Skafte, and A. Culla 39 Real-Time Dynamic Stress Response Estimation at Critical Locations of Instrumented Structures Embedded in Random Fields ............................................................................. 367 Eric M. Hernandez and Kalil Erazo 40 Strain Estimation in a Glass Beam Using Operational Modal Analysis........................................... 375 Manuel L. Aenlle, Anders Skafte, Pelayo Ferna´ndez, and Rune Brincker 41 Pressure Measurement Sensor for Jointed Structures .............................................................. 383 G. Chevallier, H. Festjens, F. Renaud, and J.-L. Dion 42 Modal Analysis of Machine Tools Using a Single Laser Beam Device............................................. 389 Christian Brecher, Stephan Ba¨umler, and Alexander Guralnik 43 Valvetrain Motion Measurements in Firing Conditions by Laser Doppler Vibrometer......................... 395 P. Castellini, P. Chiariotti, M. Martarelli, and E.P. Tomasini 44 Using High-Speed Stereophotogrammetry to Collect Operating Data on a Robinson R44 Helicopter........ 401 Troy Lundstrom, Javad Baqersad, and Christopher Niezrecki 45 Principles of Image Processing and Feature Recognition Applied to Full-Field Measurements................ 411 John E. Mottershead and Weizhuo Wang 46 Model Updating Using Shape Descriptors from Full-Field Images ................................................ 425 Weizhuo Wang, John E. Mottershead, Eann Patterson, Thorsten Siebert, and Alexander Ihle 47 Shape-Descriptor Frequency Response Functions and Modal Analysis........................................... 437 John E. Mottershead, Weizhuo Wang, Thorsten Siebert, and Andrea Pipino 48 Dynamic Simulation of the Lunar Landing Using Flexible Multibody Dynamics Model ....................... 447 Huinam Rhee, Sang Jin Park, Tae Sung Kim, Yong Ha Kim, Chang Ho Kim, Jae Hyuk Im, and Do-Soon Hwang 49 A New Approach for a Train Axle Telemetry System............................................................... 453 M. Bassetti, F. Braghin, F. Castelli-Dezza, and M.M. Maglio 50 Triaxial Multi-range MEMS Accelerometer Nodes for Railways Applications .................................. 463 M. Bassetti, F. Braghin, G. Cazzulani, and F. Castelli-Dezza 51 Acoustical Excitation for Damping Estimation in Rotating Machinery........................................... 473 Bram Vervisch, Michael Monte, Kurt Stockman, and Mia Loccufier 52 Numerical Simulations on the Performance of Passive Mitigation Under Blast Wave Loading................ 481 Oruba Rabie, Yahia M. Al-Smadi, and Eric Wolff 53 Finite Element Model Updating Using the Shadow Hybrid Monte Carlo Technique........................................................................................ 489 I. Boulkaibet, L. Mthembu, T. Marwala, M.I. Friswell, and S. Adhikari 54 Pseudo Velocity Shock Data Analysis Calculations Using Octave.................................................. 499 Howard A. Gaberson 55 Analysis and Dynamic Characterization of a Resonant Plate for Shock Testing................................. 515 Richard Hsieh, R. Max Moore, Sydney Sroka, James Lake, Christopher Stull, and Peter Avitabile 56 Resonances of Compact Tapered Inhomogeneous Axially Loaded Shafts ........................................ 535 Arnaldo J. Mazzei and Richard A. Scott 57 Modelling Friction in a Nonlinear Dynamic System via Bayesian Inference ..................................... 543 P.L. Green and K. Worden

x Contents 58 Optimum Load for Energy Harvesting with Non-linear Oscillators............................................... 555 A. Cammarano, A. Gonzalez-Buelga, S.A. Neild, D.J. Wagg, S.G. Burrow, and D.J. Inman 59 Harvesting of Ambient Floor Vibration Energy Utilizing Micro-Electrical Mechanical Devices .............. 561 Joshua A. Schultz and Christopher H. Raebel 60 Robust Optimization of Magneto-Mechanical Energy Harvesters for Shoes ..................................... 571 Stefano Tornincasa, Maurizio Repetto, Elvio Bonisoli, and Francesco Di Monaco 61 Optimization of an Energy Harvester Coupled to a Vibrating Membrane............................................................................................... 577 Levent Beker, H. Nevzat O¨ zgu¨ven, and Haluk Ku¨lah 62 Experimental Localization of Small Damages Using Modal Filters ............................................... 585 G. Tondreau and A. Deraemaeker 63 Output Only Structural Identification with Minimal Instrumentation ........................................... 593 Suparno Mukhopadhyay, Raimondo Betti, and Hilmi Lus 64 Simulation of Guided Wave Interaction with Defects in Rope Structures ........................................ 603 Stefan Bischoff and Lothar Gaul 65 Estimation of Modal Parameters Confidence Intervals: A Simple Numerical Example ........................ 611 Elisa Bosco, Ankit Chiplunkar, and Joseph Morlier 66 A Bayesian Framework of Transmissibility Model Selection and Updating...................................... 621 Zhu Mao and Michael Todd 67 Monitoring of Torsion of Guyed Mast Shafts ........................................................................ 627 Shota Urushadze and Mirosˇ Pirner

Chapter1 Safety Improvement of Child Restraint System by Using Adoptive Control Takayuki Koizumi, Nobutaka Tsujiuchi, and Shin Ito Abstract Wearing a child restraint system (CRS) greatly improves the crash-safety of children. However, the number of children’s injuries in traffic accidents hasn’t decreased. Therefore much further improvement of CRS is requested. Recently, active control of restraint systems for occupants has been studied to improve crash-safety and their effectiveness has been shown. We proposed active harness control for CRS and showed its effectiveness. We constructed a simulation model without harness control. It consists of a child dummy model and a CRS model. We attached a mechanism that changed the harness length in the CRS model. Additionally, we constructed a control system. In this system, head acceleration is fed back as a state quantity and the controller changes the belt length. However, measuring head acceleration in actual car crashes is unrealistic. Therefore we constructed a child linear model and estimate child’s head acceleration in crashes. Also, we designed a controller that determines the harness length from estimated head acceleration. Simulations were executed and the injury risks were decreased compared to the model without control of the harness. Thus we clarified the effectiveness of active control of CRS in numerical simulations. Keywords Child restraint system • Adoptive control • Crash-safety • Property estimation • Optimization 1.1 Introduction According to a National Police Agency survey in 2011, the quotient of death or severe injury in car crashes for child restraint system (CRS) users is 0.72%, while that for occupants not using a CRS is 2.71 [1]. With these statistics, the utility of the CRS is confirmed. However, since children continue to be injured even though they are wearing CRSs, further crash-safety improvement is required [2] and the research had been done to improve CRSs by attaching safety devices [3]. To improve crash-safety, continuous restraint systems have been researched for occupants, and their effectiveness has been shown [4, 5]. Therefore, we proposed a controlling harness that continuously restrains a child to a CRS and showed its effectiveness. Currently, many crash experiments and numerical simulations have investigated child crash-safety and designed safety devices. Numerical simulations are more effective than experiments, especially for repeated trials. In this study, we created a retractor model in a CRS model and attached the harness to the retractor model to create a CRS model control harness to show the effectiveness of controlling harnesses. Numerical simulation employed the CRS and dummy models using the acceleration pulses acquired from sled tests that simulated frontal crashes. The supplied acceleration is compliant with ECE R44. The control harness, we attached a mechanism that changed the harness length in the CRS model. Additionally, we constructed a control system. In this system, head acceleration is fed back as a state quantity and the controller changes the belt length. However, measuring head acceleration in actual car crashes is unrealistic. Therefore we constructed a child linear model and estimated a child’s head acceleration in crashes. Next, injury criteria were set to the objective functions, and the controller gains were set to the optimizing parameters to minimize the objective function. Using an optimized controller, we simulated the effectiveness of the controlling harness and clarified our improved CRS for greater crash-safety. T. Koizumi • N. Tsujiuchi • S. Ito ( ) Department of Mechanical Engineering, Doshisha University, 1-3, Tataramiyakodani, Kyotanabe-city, Kyoto 610-0321, Japan e-mail: tkoizumi@mail.doshisha.ac.jp; ntsujiuc@mail.doshisha.ac.jp; dtl0325@mail4.doshisha.ac.jp R. Allemang et al. (eds.), Special Topics in Structural Dynamics, Volume 6: Proceedings of the 31st IMAC, A Conference on Structural Dynamics, 2013, Conference Proceedings of the Society for Experimental Mechanics Series 43, DOI 10.1007/978-1-4614-6546-1 1, © The Society for Experimental Mechanics, Inc. 2013 1

2 T. Koizumi et al. 1.2 Accident Data In this section, we analyze the accident data to identify the current CRS problems and clarify what should be improved. Table 1.1 shows the accident data by crash type of children injured while wearing CRSs [2]. Frontal crashes comprise the largest number of fatalities and severe injuries. Figure 1.1 shows the accident data of children less than 6-year-olds who were injured while wearing CRSs [6]. The bulk of the data consists of head, neck, and thorax injuries. Therefore, in this study, we assessed the crash-safety of children by head, neck, and thorax injury criteria in frontal crashes. 1.3 Simulation Model In this section, we constructed a non-control model without a harness control. It consists of a child dummy model and a current CRS model. The details of each are described below. Next, we attached a mechanism that changed the harness length in the current CRS model. Additionally, we constructed a control system for the harness and a simulation model that can control the harness in collisions. 1.3.1 Dummy Model In this study, we adopted the Hybrid III 3-year-olds dummy model, whose overview is shown in Fig. 1.2. We modeled the Hybrid III 3-year-olds dummy model, which is generally used for the CRS assessment of front collisions [7]. Since the model exists in the MADYMO database, it has enough validity as a Hybrid III 3-year-olds dummy [8]. This model consists of 28 rigid bodies and 18 joints, and its calculation time is far shorter than finite element models. 1.3.2 CRS Model The CRS model consists of a main part, a 3-point harness, and an ECE test seat (Fig. 1.3). The main part of the CRS model consists of ten rigid bodies interconnected by joints. The 3-point harness was modeled by the finite element and represents the physical harness. The ECE test seat was modeled by two rigid planes, and the CRS model was connected to the reference place by belt elements. To evaluate this model’s validity, we conducted a sled test using the Hybrid III 3-year-olds dummy based on ECE R44 [9]. Displacement of the dummy head, the CRS, and the resultant acceleration of the head were compared in experiments and simulations. Figure 1.3 shows the results of frontal crashes and the CRS model’s validity. Next, we modeled the retractor in the cushion of the CRS model constructed above and attached it to the harness. Therefore, the CRS model has a mechanism that changed the harness length (Fig 1.4). Table 1.1 Accident data of crash type of injured children wearing CRSs Number of cases Number of fatality and severe injured Fatality and severe injured rate [%] Frontal crash 3,005 50 1.66 Side crash 1,438 22 1.53 Rear-end crash 3,245 9 0.28 0% 20% 40% 60% N=4, 321 80% 100% Head Face Neck Thorax Abdomen Leg Other Fig. 1.1 Accident data of injured areas of children under 6-year-olds wearing CRSs

1 Safety Improvement of Child Restraint System by Using Adoptive Control 3 Fig. 1.2 Hybrid III 3-year-olds dummy model Fig. 1.3 CRSmodel 1.4 Linear Model In a paper related to the study of active control, the occupant’s head acceleration is chosen as the property to control. However it is difficult to measure head acceleration in a crash. Therefore we construct a linear model to estimate the infant’s head acceleration. 1.4.1 Infant Model About the linear model in Fig. 1.5, the restraint equation of motion below holds.

4 T. Koizumi et al. a b c 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 −0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 X-Displacement [m] Z-Displacement [m] EXP_HEAD EXP_CRS_FORE EXP_CRS_REAR SML_HEAD SML_CRS_FORE SML_CRS_REAR 0 100 200 300 400 500 600 700 800 0 20 40 60 80 100 120 Time [ms] Acceleration [m/s2] Experiment Simulation 0 100 200 300 400 500 600 700 0 20 40 60 80 100 120 Time [ms] Acceleration [m/s2] Experiment Simulation Fig. 1.4 Results of CRS calibration tests in frontal crashes calibration tests in frontal crash (a) Trajectories of head and CRS, (b) Head resultant acceleration, (c) Thorax resultant acceleration x y x1 y1 x2 y2 L1,1 L1,2 L2,1 L2,2 Fig. 1.5 Example of link model ˆ.q; t/ D 2 66 66 64 x1 L1;1 cos 1 y1 L1;1 sin 1 x2 L2;1 cos 2 x1 L1;2 cos 2 y2 L2;1 sin 2 y1 L1;2 sin 2 3 77 77 75 (1.1) The generalized coordinate is q DŒx1;y1; 1;x2;y2; 2 T. When external force applied to each link is Fxi, Fyi and torque applied to the linear model is T, the generalized force is QDŒFx1;Fy1;T;Fx2;Fy2; T T. A differential algebraic equation is represented below. "M ˆ T q ˆq 0 #" Rq œ # D "QA # ˆq D @ˆi @qi DˆqRq (1.2) Mis mass matrix. is Lagrange constant number. Using the above method, we solved the differential algebraic equation of the linear model with eight rigid bodies and seven joints in Fig. 1.6. Each rigid link represents head, neck, thoracic cavity, abdomen, crural area, leg area, brachial region and antebrachial region. As its length, mass, and inertial moment, the values of the HybridIII 3-year-olds dummy model were adopted. Its joints represent an infant’s arthrogenous characteristics. We illustrate the joint characteristics in the next section. We made the initial position and, initial angle of each link correspond to the dummy model’s initial posture. The CRS was modeled as an object having only mass. We defined the five-contact model, which represents the restraint between the car body and the CRS, and the CRS and the infant. We adopted the Voight model as the contact model. Moreover, we enabled the contact model to move between the thoracic cavity and the CRS to represent the variance of harness length by changing the contact point. We define inter-joint

1 Safety Improvement of Child Restraint System by Using Adoptive Control 5 Contact model Impact acc. Fig. 1.6 Linkmodel force and the restraint force calculated by the contact model as the generalized acting force. The generalized acting force was calculated by inter joint force and the contact model, and the differential algebraic equation of the infant linear model was analyzed by the four-dimensional Runge–Kutta method. 1.4.2 Joint Characteristic The infant model’s joints are modeled by rigid links and enabled to rotate. These joints’ characteristics should be those of the HybridIII 3-year-olds dummy model, but data related to the dummy model does not exist. Therefore, we calculate the 3-year-olds infant’s joint characteristics by scaling of the adult characteristics that was proposed by Y.K. Yang and Robert and substitute the calculated data. We used the geometric scaling method to calculate the joint characteristics. This method is the way to calculate various physical quantities by consulting human body measurement. Regarding the representative length of each item, we defined the adult’s data as X, Y andZ, and the infant’s data as x, y andz. Then, the scaling coefficient of the lengths x, y and z is represented as below. X Dx=X Y Dy=Y Z Dz=Z (1.3) When using the scaling coefficient of the Young’s modulus œE, the scaling coefficient of the load is represented as below. fX=FX D E Y Z fY=FY D E Z X fZ=FZ D E X Y (1.4) The scaling coefficient of the load multiplied by the scaling coefficient of the moment arm equals the scaling coefficient of the torque. The torque is as below. tX=TX D E X Y 2 tY=TY D E Y X 2 tZ=TZ D E Z Y 2 (1.5) As the scaling coefficient of the Young’s modulus, we used scaling of the Young’s modulus of the ligament. We assumed the infant’s range of movement equals the adult’s. As the viscosity characteristic related to each joint’s rotation, we used that of the adult’s estimated by Aoki.

6 T. Koizumi et al. Fig. 1.7 Comparison of behavior between the simulation and link models (a) Simulation model, (b) Linkmodel Fig. 1.8 Comparison of simulation and link model (a) Head resultant acceleration, (b) Trajectories of head and CRS 1.4.3 Validity Verification To verify the validity of the constructed linear model, we compared it to the simulation model constructed in the previous chapter. In a way similar to the previous chapter, simulations were executed without the control of the harness using the acceleration gained by the sled test. Figure 1.7 shows the comparison of the behavior between the linear model and the simulation model in car crashes. From this figure, there is a little difference of the behaviors of the leg between using the linear model and not using. The cause may be that the contact model is not defined between the leg and the CRS. However Fig. 1.8 shows that the head acceleration, head displacement and CRS displacement of the linear model correspond to those of the simulation model. Thus, the validity of the linear model constructed in this paper is confirmed. 1.5 Control System 1.5.1 Block Diagram We designed a PI controller to control the harness length of the CRS model in collisions using MATLAB/Simulink software. Next we constructed a control system to decrease the risk of injuries to the head, neck, and thorax.

1 Safety Improvement of Child Restraint System by Using Adoptive Control 7 Simulation model Linear model Controller Head acc. Estimated head acc. Impact acc. Fig. 1.9 Block diagram of experimental control system Figure 1.5 shows our constructed block diagram. The dummy model sits on the CRS model, and the collision acceleration is input. The PI controller controls the harness length by feeding back the output head accelerations estimated by the constructed linear model to the controller (Fig 1.9). 1.5.2 Optimization In this section, head and thorax injury criteria were set to the objective functions, and the controller gains were set to the optimizing parameters to minimize the objective function. The optimization of the controller gains was performed with optimization software called modeFRONTIER. Numerical simulations were executed repeatedly and parameters were optimized. These parameters were applied to the controller and reconstructed a controller system. A simulation model with this optimized control system is called a control model. 1.5.3 Objective Functions In this study, the crash-safety of children was assessed by head, neck, and thorax injury criteria in frontal crashes. HIC36, Nij, and Thorax Acceleration Tolerance were chosen as the injury criteria and set to the objective functions that should be decreased. The head injury criterion (HIC) is formulated below. HIC Dmax (.t2 t1/ 1 t2 t1 Z t2 t1 adt 2:5) (1.6) a: Head resultant acceleration HIC36is set as t2–t1 D36 [ms] and is widely used in the collision safety field [10]. A threshold value of 1,000 was applied to the Hybrid III 3-year-olds dummy, and we assumed a risk of injury if the criterion exceeded 1,000. The neck injury criterion (Nij), which is formulated below, had four kinds of values determined by the combination of axial force and bending moment. The combination of tension and extension is NTE, the combination of tension and flexion is NTF, the combination of compression and extension is NCE, and the combination of compression and flexion is NCF. Nij, which is their maximum value, is set to an objective function. Nij D Fx Fxc C Mocy Myc (1.7) Fx: Neck axial force [N] Fxc: Critical value of axial force [N] My : Neck bending moment[Nm] Myc: Critical value of bending moment[Nm] A threshold value is 1.0 because Nij is regularized by critical values and might cause injury if the criterion exceeds this value.

8 T. Koizumi et al. As a thorax injury criterion, we chose thorax acceleration tolerance, because it is directly calculated from spinal acceleration. This criterion is peak spinal acceleration sustained for 3 [ms]. The threshold value of thorax acceleration tolerance is 60 G and might cause injury if the criterion exceeds this value. 1.5.4 Optimizing Algorithm We adopted the genetic algorithm as an optimizing algorithm in this study. The genetic algorithm imitates the inheritance and the evolution of life, based on the idea of evolution to a better design by assuming that one design is one gene and repeating such operations as the crossover of two genes, mutation, and natural selection. Sixteen designs were generated as the first generation by semi-random sampling and evolved until the 40th generation. The mutation rate was assumed to be 5%. 1.5.5 Control Model Optimization that minimizes the objective functions was done using the algorithm, and many Pareto optimum solutions were obtained. A Pareto optimum solution means that a better solution doesn’t exist among the solutions, although it is not necessarily more dominant than all other solutions. Therefore, evaluation function J was formulated in this study, and the Pareto optimum solutions were judged by the evaluation function, which regularizes by dividing the injury values of the head, the neck, and the thorax by each threshold value, as described below. J D HIC 1000 2 C Nij 1:0 2 C Thoraxtolerance 60G 2 (1.8) The evaluation function J selected a Pareto optimum solution as a preferred solution. This preferred solution was applied to the controller gains to reconstruct a controller system. In this study, a simulation model with this optimized control system is called a control model. 1.6 Simulation Results To clarify the optimization effect of the harness controller, simulations were executed using the non-control and control models and we compared these models. The control model is the simulation model that consists of the dummy model and the current CRS model without a harness control system. Table 1.2 compares the injury values of the non-control and control models. Additionally, Figs. 1.10 and 1.11 compare responses and behaviors. First, we compared and discussed the responses of both models. Table 1.2 shows that all injury risks of a control model were substantially decreased compared to a non-control model. This results shows that CRS improved crash-safety. Additionally, in Fig. 1.10d, each value temporarily decreased at about 90 [ms], perhaps because the controller changed the length of harness rapidly and harness was loosened temporarily. Figure 1.10a shows the head resultant accelerations. The peak value in the control model is smaller than in the non-control model. Therefore, the head injury value decreased from 596 to 305, and head injury risks decreased. Figure 1.10b shows the thorax resultant accelerations. The peak value in the control model is smaller than in the non-control model and the acceleration increased earlier. Perhaps because the belt was immediately fastened after the impact and the peak value decreased. Therefore, the thorax injury value decreased from Table 1.2 Comparison of the injury values between the non-control and the control models Injury value (Threshold value) HIC(1,000) Nij (1.0) Thorax acceleration tolerance (60G) Non-control 593.9 (59[%]) 2.29 (229[%]) 60.0 G (100%[%]) Control 304.8 (30[%]) 1.25 (125[%]) 56.3 G (94[%]) Ratio of decrease to threshold value 29[%] 104[%] 6[%]

1 Safety Improvement of Child Restraint System by Using Adoptive Control 9 Fig. 1.10 Response comparison between non-control and control models (a) Head resultant acceleration, (b) Thorax resultant acceleration, (c) Neck axial force, (d) Neck bending moment, (e) Trajectories of head and CRS 60 G to 56 G, and the thorax injury risk decreased as a result. Figure 1.10d show the neck axial force and the neck bending moment. Even though major differences of the peak values in the control and non-control models are not seen in the neck axial force, the peak value decreased in the neck bending moment. As a result, the neck injury value decreased from 2.91 to 1.25, and neck injury risks decreased. From Fig. 1.10e, the maximum head displacement became 0.65 [m] in the non-control model and 0.80 [m] in the control model, and the displacement difference was 0.15 [m]. In the control model, the head displacement increased because the controller lengthened the harness to prevent head, neck, and thorax injuries. The crash-safety of CRS was improved by optimized harness controller. 1.7 Conclusion The following conclusions were drawn from this study. 1. Attaching the concept of the adoptive control for the CRSs, the possibility of improving the crash safety was clarified. 2. We constructed a child linear model and estimated a child’s head acceleration to control the harness length. 3. The numerical simulations showed that the head, the neck, and the thorax injury risk decreased with PI controller designed for controlling of the harness.

10 T. Koizumi et al. Fig. 1.11 Comparison of behavior in non-control and control (a) Non control model, (b) Control model Acknowledgements This work was partially supported by Grant-in-Aid for Scientific Research (C) (23560272), Japan Society for the Promotion of Science. References 1. Transportation Authority of National Police Agency (2011) Incidents of transportation accident in 2011 (in Japan) 2. Yoshida R, Goto S, Mori K (2002) The study of child restraint system safety based on ITARDA field survey. In: Proceedings of the 2002 JSAE congress, Kyoto, Japan 3. Koizumi T, Tsujiuchi N, Kurumisawa J (2008) Optimization of child restraint system with load limiter and airbag using child FE human model in frontal crash. In: International conference on noise and vibration engineering, Leuven, Belgium 4. Gabriella G, Paul L, Edwin van den E, Arjan van L, Cees van S, John C (2007) Real time control of restraint systems in frontal crashes. In: SAE World Congress and Exhibition, Detroit, MI USA 5. Ewout van den L, Bram de J, Frans V, Maarten S, Ellen van N, Dehlia W (2004) Continuous restraint control systems: safety improvement for various occupants tiofidelity rating of MADYMO three-year-olds child FE human model. In: The 18th international technical conference on the enhanced safety of vehicles, Nagoya, Japan 6. Institute for Traffic Accident Research and Data Analysis (1999) Investigative research report on effect of wearing CRS, ITARDA pp 13–29 7. Kathleen DeSantis K, Roger AS, Gaston A, Stanley B, Michael K (1996) Techniques for developing child dummy protection reference values. NHTSA biomechanics reports 8. TNO Automotive (2004) MADYMO theory manual version 6.2 9. ECE Regulation No. 44. Uniform provision concerning the approval of restraining devices for child occupant of power-driven vehicles 10. Workgroup Data Processing Vehicle Safety (2004) Crash analysis criteria description version 1.6.1

Chapter2 Dynamic Response and Damage Estimation of Infant Brain for Vibration Takayuki Koizumi, Nobutaka Tsujiuchi, Keisuke Hara, and Yusuke Miyazaki Abstract The purpose of this paper is to clarify the mechanical generation mechanism of acute subdual hematoma, which is a severe injury in infants, by performing experiments and finite element analysis. The acute subdural hematoma in infants is caused by accidents such as falling or abuse such as shaking. This paper describes the shaking events. In the experiments, we used a 6-month-old anthropometric dummy and a vibration exciter, which can set the parameters. The dummy was fixed to the exciter at the chest, which it was given the vibration. The head model of the dummy is transparent, and the brain behavior can be visualized. In finite element analysis, we used a model that has been converted to the 6-month-old head through the adult head by the method of free-form deformation (FFD) and scaling. Also, we performed the simulation of shaking events as input acceleration and angular velocity of the head obtained in the experiments. We measured the stretch ratio of the bridging veins, which connect the skull and the brain, then compared this with the ratio to the threshold (1.5). In this study, we examined the effect on the infant head of shaking action, along with the risk. Keywords Infant • Acute subdural hematoma (ASDH) • Shaken baby syndrome • Finite element analysis • Material properties • Frequency 2.1 Introduction Currently in Japan, the leading cause of death in children under a year old is accident, and this has not changed since 1960 [1]. Intentional injury such as abuse experienced by children has also become a problem, and physical abuse involving infant head trauma is considered the highest risk to life. Among severe cases of infant head trauma, acute subdural hematoma (ASDH) is prominent. The mortality rate due to this is high, and the survivors suffer from heavy permanent damage. When assessing ASDH, it is not easy to ascertain whether its cause is abuse by shaking or an accident such as a fall. Judgment of abuse or accident in medical institutions that relies on experience and intuition lacks a scientific basis. Shaken baby syndrome is well known as a form of head injury caused by abuse. However, it is not clear whether the shaking action itself is fatal from previous study. Therefore, it is necessary to clarify the generating mechanism of ASDH in infants and to provide a scientific basis to make the judgment. ASDH occurs by the relative rotational motion between the skull and the brain with a rupture of bridging veins. However, with the traditional dummy that has a rigid head, it is not possible to visualize the relative rotational motion between the skull and brain during shaking. Therefore, in this study, we use an infant anthropometric dummy that has a realistically shaped physical model of an infant head to visualize the relative motion between the skull and the brain. Also, we perform experiments to evaluate the T. Koizumi • N. Tsujiuchi • K. Hara ( ) Department of Mechanical Engineering, Doshisha University, 1-3, Tataramiyakodani, Kyotanabe-city, Kyoto 610-0321, Japan e-mail: tkoizumi@mail.doshisha.ac.jp; ntsujiuc@mail.doshisha.ac.jp; dum0517@mail4.doshisha.ac.jp Y. Miyazaki Department of Information Environment, Graduate School of Information Science and Technology, Tokyo Institute of Technology, 2-12-1, Ookayama, Meguro-ku, Tokyo 152-8552, Japan e-mail: y-miyazaki@mei.titech.ac.jp R. Allemang et al. (eds.), Special Topics in Structural Dynamics, Volume 6: Proceedings of the 31st IMAC, A Conference on Structural Dynamics, 2013, Conference Proceedings of the Society for Experimental Mechanics Series 43, DOI 10.1007/978-1-4614-6546-1 2, © The Society for Experimental Mechanics, Inc. 2013 11

12 T. Koizumi et al. behavior of the dummy head for vibration input that assumes a shaking act with a vibration exciter. The vibration exciter can set the parameters such as frequency and amplitude. In addition, we constructed a finite element model of the infant head and simulated the shaking. The model faithfully reproduced the shape. By performing the simulation, we examined the stretch ratio of the bridging veins and relative motion between the brain and skull when shaking. 2.2 Anthropometric Dummy 2.2.1 Six-Month-Old Anthropometric Test Dummy In this experiment, we used a CRABI 6-month-old (CRABI 6-Mo) [2], whose height is 67 cm and weight is 7.8 kg. The neck is shaped in soft rubber, and attributes such as bending and stretching are close to the actual biology. The dummy reproduced the highest biological characteristics of the infant in current commercial dummies. The x axis is aligned in the anterior posterior direction, with positive x indicating the anterior direction. The y axis is aligned in the lateral directions, with the positive y indicating the dummy’s left side. The z axis is aligned in the superior–inferior direction, with positive z indicating the superior direction. 2.2.2 Transparent Skull Model In this study, we used the transparent skull model to visualize the behavior of the brain [3]. Figure 2.1 shows the model. This model reproduces the shape of the head of an infant. The skull is composed of polycarbonate with high transmittance. The brain model is composed of left and right cerebrum, cerebellum and brain stem, and manufactured with silicon gel, the dynamic viscoelastic properties of which are equivalent to the real brain. This brain model was inserted into the skull model and then the gap between brain and skull was filled with water as cerebrospinal fluid (CSF). Flax and tentorium were modeled with a polyurethane sheet, the Young’s modulus of which was equivalent to the real one; then it was affixed to the inner surface of the skull. Thus, the brain movement was confined. We constructed the infant dummy, which can visualize brain behavior, by attaching the transparent skull model to CRABI 6-Mo. Figure 2.2 shows the dummy. 2.3 Experiments The shaking vibration by humans is not constant. Therefore, in this study, we performed the experiments using a vibration exciter to set the input parameters such as frequency and amplitude. Fig. 2.1 Head physical model

2 Dynamic Response and Damage Estimation of Infant Brain for Vibration 13 Fig. 2.2 Infant dummy (CRABI 6-Mo) Fig. 2.3 Appearance of shaking experiment by vibration exciter 2.3.1 Methods Figure 2.3 shows the experiment aspect. The dummy was fixed on a board it was to transmit vibration of the vibration exciter to the chest. We set up an angular velocity sensor and acceleration sensors on the head of the dummy and attached an acceleration sensor on the chest; then, we measured the acceleration and angular velocity. We applied white markers to the inner surface of the skull and the brain surface to measure the relative movement between the skull and brain. We set to bv1-bv3 each combination of markers of brain and skull. Figure 2.4 shows the combination of markers. Head behavior was taken at a sampling rate of 500 fps or 200 fps with two high-speed cameras (D-III: Detect). The displacement time history of markers was measured by the digital image correlation method. Then, we measured the relative displacement of the markers of bv1-bv3 by converting the three-dimensional displacement method using the Direct Linear Transformation (DLT) method. In addition, we smoothed each three-dimensional displacement by the three-point moving average method. We calculated the stretch ratio (œ) between two points as the evaluation strain parameters of bridging veins of the attached positions by the three-dimensional displacement. The stretch ratio (œ) is defined by the following equation using the distance (l0) of the bv1-bv3 in the video frame and the distance (l) of the bv1-bv3 in the initial position. We defined the value obtained by the following equation as the bridging vein stretch ratio. œD l l0

14 T. Koizumi et al. Fig. 2.4 Makers positions 2.3.2 Input Vibration In previous studies, we concluded that the vibration in the vertical direction has less impact on the infant head. Therefore, we have performed experiments for only the X axial, which is the main component of the shaking vibration. Input vibrations were a total of 12 patterns, combining 3 amplitude patterns [30.0 mm, 40.0 mm, 50.0 mm] in the X axial with 4 frequency patterns [1.5 Hz, 2.0 Hz, 2.5 Hz, 3.0 Hz]. To reproduce the human act of shaking infants, input amplitudes are set to the value around the one with which people shake the infant dummy. In addition, input frequencies were set at a lower value than the maximum high-risk frequency (around 3.0 Hz) obtained by previous studies. In the previous study, the dummy’s head, which was a rigid model, was swung the most on the 3.0-Hz frequency. 2.4 Results and Discussion Results are reported in Fig. 2.5 in terms of the bridging vein stretch ratio of bv1-bv3 obtained by the experiments and time history and in Fig. 2.6 in terms of the maximum ratio of results and amplitudes. The maximum of the bridging vein stretch ratio tends to increase with the increasing amplitude at any frequency in Fig. 2.6. Therefore, the breaking risk of bridging veins is high as amplitude increases. For 40 mm or more amplitude on 3.0-Hz frequency, the stretch ratio exceeds the 1.5 threshold, which is the breaking value of bridging veins reported by Lee and others [4]. Particularly, for 50.0-mm amplitude, the stretch ratio is significantly higher than the threshold. For 30.0-mm amplitude, the stretch ratio is significantly below the threshold even though the frequency is 3.0 Hz, which is high-risk. Therefore, bridging veins are likely to break above 40.0-mm amplitude on 3.0-Hz frequency, and the risk is especially high in 50.0-mm amplitude, while there is no risk in 30.0 mm amplitude. In other words, we can prove mechanically that ASDH is likely to occur for severe vibration. In Fig. 2.6, the stretch ratio is significantly below the threshold in any amplitude of 2.5 Hz and under. In other words, bridging veins don’t break for 2.5 Hz and under. Therefore, we can conclude that ASDH due to breaking bridging veins does not occur with the vibrations resulting from cradling. 2.5 Simulation 2.5.1 Construction of Model and Material Properties The finite element model of an infant head has been developed very little, even though the head model of adults has been developed in many studies. For constructing a finite element model of the infant head, it is not appropriate to obtain it from scaling the adult head model. A highly accurate infant head model cannot be constructed because the characteristics of skull shape are different for infants and adults. In this study, we used the finite element model of a 6-month-old head, which is newly constructed. Figure 2.7 shows the new model and Table 2.1 shows the dimensions. We retrieved the three-dimensional shape date of the head by CT images of a

2 Dynamic Response and Damage Estimation of Infant Brain for Vibration 15 2 1 0 0 0 1 1 2 0.5 BV1 BV2 BV3 BV1 BV2 BV3 BV1 BV2 0.5 1.5 Time [s] Time [s] Time [s] 1.5 0 1 0.5 1.5 2.5 1.5 0.5 BV stretch ratio BV stretch ratio BV stretch ratio 2 1 0 1.5 0.5 3 2 1 0 2.5 1.5 0.5 Fig. 2.5 Bridging vein stretch ratio responses with high speed camera images Fig. 2.6 Max bridging vein stretch ratio responses with high speed camera images Fig. 2.7 Finite element model Table 2.1 Dimension of 6-Mo head [mm] Length Breath Height 154.9 119.4 147.3 particular 4-month-old head. Then, we converted the shape of the finite element model of an adult head into the 4-month-old head model based on the shape date with the free-form deformation transformation (FFD) method. The 6-month-old infant head finite element model that is used in this study is constructed by scaling the dimensions of a 4-month-old head which have been converted to the shape of the CRABI 6-Mo head. The head finite element model is composed of a skull including structure and anterior fontanel, cerebrospinal fluid (CSF), brain (left and right cerebrum, cerebellum and brainstem), and membrane (dura mater, pia mater, flax and tentorium).

16 T. Koizumi et al. Table 2.2 Material properties (1) Part Material property Density ¡ [kg/m3] Young’s modulus E [GPa] Poisson’s ratio Right cerebrum Linear Visco Elastic 1,040 KD2.19 Left cerebrum Cerebellum Brain stem Piamater Elastic 1,133 1.15 10 2 0.45 Duramater 1,133 3.15 10 2 0.45 Falx 1,133 3.15 10 2 0.45 Tentorium 1,133 3.15 10 2 0.45 Sagittal sinus 1,133 3.15 10 2 0.45 CSF Linear Visco Elastic 1,060 KD2.19 Ventricle Skull diploe Elastic 2,150 421 10 3 0.22 Structure 4.2 10 3 Mandible Elastic 2,150 4.6 0.05 Inner table Elastic 2,723 15 0.21 Outer table Bridging vein Elastic 1,133 9.62 10 3 0.45 Table 2.3 Material properties (2) Part Density [kg/m3] Bulk modulus [Gpa] Short time shear modulus [GPa] Long time shear modulus [GPa] Decay constant [s 1] Brain 1,040 2.19 2,710 10 9 891 10 9 166 CSF 1,060 2.19 500 5.0 10 7 500,000 Ventricle Table 2.4 Properties of the modified material (1) Part Material property Density ¡[kg/m3] Young’s modulus E [GPa] Poisson’s ratio Brain Linear Visco Elastic 1,040 KD2.19 Falx Elastic 1,133 3.15 10 2 0.45 Tentorium 1,133 3.15 10 2 0.45 Skull Elastic 1,200 2.45 0.33 Table 2.5 Properties of the modified material (2) Part Density [kg/m3] Bulk modulus [Gpa] Short time shear modulus [GPa] Long time shear modulus [GPa] Decay constant [s 1] Brain 1,040 2.19 2,704 10 9 886 10 9 166 CSF 1,060 2.19 500 5.0 10 7 500,000 Tables 2.2 and 2.3 shows the material properties of each tissue. Suture and skulls of infants are soft as ossification is not complete compared with adults. Thus, the material properties used the average value of the measurement results of 1–12 months, which are obtained in the three-point bending test of Margulies et al. [5]. Material testing at 6 months has not been carried out. Therefore, the suture material property used is for 11 months. The CSF property used is a viscosity elastic property, almost like water. In addition, other properties given are the adult properties. We can verify the operational state of the load and the deformation of the intimal structure such as skull and brain by using a finite element modal. 2.5.2 Simulation and Result In this study, we have carried out a simulation to verify the frequency response of the brain. In the simulation, we constructed the finite element model of the dummy’s head. We developed the model by modifying the material properties because the dummy’s head and the finite element model are equal for shape and structure. The dummy’s head is composed skull, brain, flax, tentorium and CSF. Therefore, the simulation model is standardized of dummy’s configuration. Tables 2.4 and 2.5 shows the properties of the modified material.

RkJQdWJsaXNoZXIy MTMzNzEzMQ==