1
Musculoskeletal WebinarMusculoskeletal Simulation and Device
Design (webinar will start at 9am PST)
David Wagner, PhDOzen EngineeringAugust 28, 2009
Please visit:http://www.ozeninc.com/default.asp?ii=273for upcoming webinars
You will be connected to audio usingyour computer's microphone andspeakers (VoIP). A headset is
recommended.
Or, you may select "Use Telephone"after joining the Webinar.
- Dial 312-878-0211- Access Code: Shown in
window- Audio PIN: Shown after
joining the meeting
- Webinar ID: 416-995-435
Welcome to the WebinarWelcome to the Webinar. Please make sure
your audio is working
Feel free to use
computer speakers
or telephone
Type any questions
you have here
2
!"#$%#$&'$##('$&)%'$*+%!"#$%#&'()(#*+&'',-#.&/0',/1#2345!6!%7#84%"9:;%7#"<35!8<"1=3$!%!%71#=<8>%!8$:#"92243=1#?$3@<;%7#$%6#"$:<"#!%#'43=><3%#*$:!A43%!$#A43#,$-.-"4BC$3<#234698="#$%6#A43#/0123(!$4'#()##-,*0/5%$%6#,678017+
92%:;2<=2%1>;2:=)%?0:@A21%A0B@C06A%?0;%70@;%:0/5<67
4D2%*,#%400BA%92%EA2The CAE Tools
Why Is Ozen Engineering Hosting Webinars?
- Get our name out there
- Another avenue that compliments dissemination throughpeer reviewed papers
- Resources available for research => would like to seesimilar dedication at the commercial level
- Improve collaboration between ‘people’ (researchers,students, teachers, companies) who use (or want to use)musculoskeletal simulation
- Answer the question: “What engineering problems can weaddress with the existing state of the art software andmethods currently available?”
3
Welcome to the WebinarWelcome to the Webinar. Please make sure
your audio is working
Feel free to use
computer speakers
or telephone
Type any questions
you have here
Summary
Optimization results and future questions
An example application (proximal radius fracture plate)
A Proposed workflow for incorporating musculoskeletal modeling
Modeling the human body – Musculoskeletal simulation of activitiesof daily living
Quick review of the previous webinar (uses of simulation)
4
Uses of Simulation in the Orthopedic Industry
Replicating Physical Test Research (Internal/University)
Kim et al. 2008, SBC2008-193023
Li et al. 2008, SBC2008-192776
Design of Orthopedic Devices and ProstheticsASME Summer Bioengineering Conference (2008)
Finding out what went wrong
Finite-elementanalysis offailure of theCapital HipdesignsJanssen et al.2005
Benefits of Simulation
The use of computational simulation can be beneficial if it:• accurately represents and replicates the physics of the system• increases the number of possible design iterations (within a fixed
time)• decreases the cost associated with each design iteration• improves the fidelity of analysis as related to making design
decisions• is integrated in the design process
5
Replicating Standardized Physical Tests
For example…ASTM F384 -06 Standard Specifications and Test Methods for Metallic Angled OrthopedicFracture Fixation Devices (no associated ISO standard)
• Methods for bending fatigue testing• Fatigue life over a range of maximum bending moment levels• Estimate the fatigue strength for a specified number of fatigue cycles• Not intended to define levels of performance of case-specific
ASTM F1264 Standard Specification and Test Methods for Intramedullary Fixation Devices• performance definitions• test methods and characteristics determined to be important to in-vivo performance
of the device (bending fatigue test, static torsion test, static four-point bend test)
• It is not the intention of this specification to define levels of performance or case-specific clinical performance of these devices, as insufficient knowledge to predictthe consequences of the use of any of these devices in individual patients forspecific activities of daily living is available
Components of a Device Analysis
Similar process flow used by Duda et al. 1998, Taylor M.E. et al. 1996, Speirs et al. 2007, TaylorR.E. et al. 2008, and Wagner et al. 2009
6
From Kojic 2008
Comparison of Fracture Fixation Devices
Fixed PlateInternal compressionresulting from screw +fixation plate geometry
Intramedullary nailBending stiffness:Kb = ExI
E, Young’s Modulus of ElasticityI, the second moment of inertia
for bending of the nail crosssection
Torsional stiffness:Kt = ExIt
G, Shear ModulusIt, the second moment of inertia
for torsion
From Kojic 2008
Example Analysis - Fixed Plate Boundary Conditions
FixedConstraint
~ approximatingof axial loadduring humanwalking (singlestance phase of70 kgindividual)
7
From Kojic 2008
Example Analysis Results - Effective Stresses
No slipconditionmodeledbetweenscrews, plate,and bone =>i.e. bondedcontacts
MPa
From Kojic 2008
Example Analysis Results - Fixed Plate Stresses
Stainless steelused for plateand screws
E = 2.1x105 Mpa
Poissons ratio = 0.3
Maximum effective stressless than critical values forstainless steel. However,cyclic loading leading tomaterial fatigue must alsobe considered
8
From Kojic 2008
Example Analysis - Intramedullary Nail
Same bone geometry,material properties, and
boundary conditions as inthe neutralization plateanalysis
From Kojic 2008
Example Analysis - Intramedullary Nail StressesEffective stress concentrations in the nail near the screw regions => However, stress valuesare significantly lower than the corresponding neutralization plate regions (~80 MPa).Implication is that risk of intramedullary nail failure is significantly lower when compared toneutralization plate.
9
From Kojic 2008
Example Analysis - Intracapsular Fractures
Parallel Screws Dynamic Hip Implant
Comparison of implant designs for internal fixation of intracapsular fractures of thefemoral neck
From Kojic 2008
Example Analysis - Parallel Screws BCs
Positive correlationbetweenintraoperativestability and
femoral neckfractures that havehealed (versus didnot heal),Rehnberg et al.1989
Fixed BoundaryCondition
FR: Pelvis to femur head reaction force, 199 daNFA: Force generated by gluteal muscles, 137 daNBody weight: 70 daN
10
Components of a Device Analysis
Similar process flow used by Duda et al. 1998, Taylor M.E. et al. 1996, Speirs et al. 2007, TaylorR.E. et al. 2008, and Wagner et al. 2009
Components of a Device Analysis
Similar process flow used by Duda et al. 1998, Taylor M.E. et al. 1996, Speirs et al. 2007, TaylorR.E. et al. 2008, and Wagner et al. 2009
11
Summary
Optimization results and future questions
An example application (proximal radius fracture plate)
A Proposed workflow for incorporating musculoskeletal modeling
Modeling the human body – Musculoskeletal simulation of activitiesof daily living
Quick review of the previous webinar (uses of simulation)
Can we usesimulation in amore ‘pro-active’way to developbetter products?
Doing More with Simulation (one idea)
12
• Help understand what is going on inside the human body
• We use simulation for many other engineering analyses,why not for the human body as well
• Design/redesign ‘safe’ working environments
• Teaching
• Functional assessments (neuromusculoskeletal system)
• Create/Mimic realistic movement
• Sometimes the only way to understand and learn moreabout complex systems (like people!)
Simulation for !Biomechanics" - Why?
• Musculoskeletal Analysis– AnyBody– LifeModeler– Opensim/SIMM/SimTK– Madymo (TNO)– ESI Group– Marlbrook– Motek– Visual3D (c-motion)
• Digital Manikins– RAMSIS (Human
Solutions)– Jack (UGS/Siemens)– HumanBuilder/Delmia
(Dassault)– HumanCAD (NexGen)– SANTOS (U. Iowa)– Some others
• Motion Capture– BodyBuilder (Vicon)– Simi – Qualisys – SIMM (Motion Analysis)– XSENS– Many others
• CAE tools (FE/CAD)– ANSYS – LS-DYNA (ANSYS)– Abacus (Dassault)– AutoCAD (AutoDesk)– NASTRAN & ADAMS (MSC)– COMSOL
• Other tools– Matlab (Mathworks)– Mathematica
Simulation Software for !Biomechanics"
13
The Holy Grail…
Task + Environment + Population
UniqueSimulation
from Parkinson and Reed (2008)
Working Within the Confines of the Current Technology
• Library of activities– Can’t rely (yet) on the musculoskeletal models to ‘adapt’ to new
task/environment conditions => particularly for novel (~non-cyclic)tasks
• Global Assessments vs. Better Products/Designs– Models that match measured results are great, but models that
exhibit realistic trends may be sufficient (and as useful)
• Better incorporation/understanding of variability– E.g. Within subject variability as indicator of model performance
• Will we ever be able to use Musculoskeletal Simulationwithout a corresponding validation study– Can’t ALWAYS be expected to conduct a validation study for a new activity– Must have confidence in the tools (e.g. Finite Element Models)
14
Expanding the Use of Activities of Daily Living with a
Library of Musculoskeletal Simulations
• Long-term stability of hip-implants have been
evaluated using normalwalking, sit to stand, stairclimbing, and combinationsof those activities.
• Traditionally used aspass/fail tests to identifywhether a particular designperforms to a set ofminimum specifications
• Significantly Underutilized
Musculoskeletal Models Used Here80
14.6
35
5.2
549
121
709
782804
17
121
121
(b)
Popular class of musculoskeletalmodels based on rigid bodydynamics:
• Bones and objects from theenvironment are rigid
• Muscles and ligaments aremass-less actuators
• Soft tissue – “wobbly“masses are not taken intoaccount (mass isconcentrated in bones)
• Phenomenological musclemodels
• ‘Easily’ scalable
Suited for simulating internal body forces (muscle,joint, ligament) for prescribed activities
Static 2D
Dynamic 3D (AnyBody
Modeling System)
15
},..,1{ ,0
],[ where,
)()(
)()(
MMi nif !"
==MRfffdCf
MuscleforcesJoint
reactions
Internalforces
Appliedforces
The matrix C is rectangular. This means that there areinfinitely many solutions to the system of equations.How to pick the right one?
Formulating Dynamic Equilibrium
Using Optimization to Get a Solution
!
Minimize
G(f (M))
Subject to
Cf = d
fi(M )
" 0, i # {1,..,n(M )}
Objective function. Differentchoices give different muscle
recruitment patterns.
What should be used for ?
!
G(f(M))
16
Musculoskeletal Models for Commercial Use
No ‘gold-standard’, just like with other pieces of engineeringsoftware
Commercially available (including open source) softwarepackages demand a knowledgeable user
Not traditionally incorporated in current design/engineeringmethodologies
Always room for improvement (I.e. improved validation, betteraccuracy, scaling to populations or patient specific, etc.)
Still must demonstrate where/how this arena of modeling canimprove specific processes (I.e. $$$)
Summary
Optimization results and future questions
An example application (proximal radius fracture plate)
A Proposed workflow for incorporating musculoskeletal modeling
Modeling the human body – Musculoskeletal simulation of activitiesof daily living
Quick review of the previous webinar (uses of simulation)
17
Bridging the Gap with Simulation
Physical Testing“Simulated”
Physical Testing“Simulated” In-
Vivo Performance
Implant Evaluation (see previous webcast for a more
substantive description of this process, 7-24-09)
18
Implant Optimization
Summary
Optimization results and future questions
An example application (proximal radius fracture plate)
A Proposed workflow for incorporating musculoskeletal modeling
Modeling the human body – Musculoskeletal simulation of activitiesof daily living
Quick review of the previous webinar (uses of simulation)
19
Task: Wheel ChairPropulsionExertion
Device: TraumaImplant (proximalradius fracture)
Proposed Analysis
Necessary Pieces
Titanium alloy(Ti-6Al-4V)
20
Titanium Alloy (Ti-6Al-4V)
Geometry Parameterization
A 100º angular plate position
(PlatePositionAngle) is
defined to be located on the
proximal lateral side of the
radius. An increase in the
angular position corresponds
to a counterclockwise rotation
of the plate about an axis
defined by looking down the
radius shaft from the proximal
toward the distal end.
21
Design of Experiments (Parametric Exploration)
90 Designs, Full Factorial for:
45
19043
14541
410039
35537
21035
Plate Depth (mm)Plate Position
(degrees)Plate Size (degrees)
30 Designs shown for 2mm Plate Depth =>
Design Parameter Boundaries
22
Max Stress v. Plate Variables
Maximum Stress Calculation
Vs.
Top Surface Entire Plate
23
Top Surface Equivalent Stress
10
Plate Position (degrees)190
10045
35
2 mm Plate 3 mm Plate 4 mm Plate
707 MPa
106 MPa
612 MPa
58 MPa
542 MPa
37 MPa
Plate Size (degrees)
Data available for Fatigue Life, Max Principal Strain, Max Displacement, etc.=> criteria a design engineer could use to make develop better products
Summary
Optimization results and future questions
An example application (proximal radius fracture plate)
A Proposed workflow for incorporating musculoskeletal modeling
Modeling the human body – Musculoskeletal simulation of activitiesof daily living
Quick review of the previous webinar (uses of simulation)
24
Optimization Flow (Desired)
- Minimize Maximum Stress in Plate- Minimize Maximum Displacement in Plate- Minimize Plate Mass
- Plate Orientation
wrt bone- Plate Thickness- Plate Size (angular)
Optimization Flow
- Minimize Maximum Stress in Plate- Minimize Maximum Displacement in Plate- Minimize Plate Mass
- Plate Orientation
wrt bone- Plate Thickness- Plate Size (angular)
25
Optimization Performed in ModeFRONTIER
250 design runs~8 minutes per run (34 hour total run time)MOGA-II optimization algorithm
Multi Objective GeneticAlgorithm - an efficient multi-
objective genetic algorithm(MOGA) that uses a smart multi-search elitism. This new elitismoperator is able to preservesome excellent solutions withoutbringing premature convergenceto local-optimal frontiers
Results - Input - Plate Position Angle
26
Results - Input - Plate Size Angle
Results - Input - Plate Thickness
27
Scatter Plot of Design Points
Plot to stare at…
28
Parallel Coordinate Charts (another way to sift through the data)
Parallel Coordinate Charts (another way to sift through the data)
29
Another Optimization Run
80 degrees
Trade Off Between Mass and Plate Stress
Pareto Designs Marked
30
Only Pareto Points Plotted
More Possibilities…Robust Design => ApplyingUncertainty to Parameters
OR…SensitivityAnalysis to determinerelationship betweenvariation inparameters andoutcome metrics
31
*06=<:C6F%!G26%#6F>622;>6F%=0%H2<;6%I0;2
• 04#:<$3%#?43<#$D49=#&'()(1#?46<E/F'0G,/1#43#&%HI46H1
2:<$"<#84%=$8=#9"J#!%A4K4L<%!%8M84?
###NOPQR#STUVOWWX
• GA#H49#C49:6#:!@<#=4#>$5<#$#6<?4#A43#H493#"2<8!Y8#%<<6"#43#A43
$%H#4=><3#Z9<";4%"#3<:$=<6#=4#=>!"#C<D!%$31#2:<$"<#84%=$8=#9"
$=J#?9"89:4"@<:<=$:K4L<%!%8M84?
Contacting Ozen to Learn More
Thank you for your attentionThank You For Your Attention
E43#A93=><3#!%A43?$;4%1#2:<$"<#84%=$8=J
F[,'#,'\G',,/G'\1#G'*M
]U]P#,M#&/^_,(#&`,M#(_G0,J#UPS
(_'')`&-,1#*&#aOPQX
NOPQR#STUVOWWX
!%A4K4L<%!%8M84?
CCCM4L<%!%8M84?
• .:<$"<#:<=#9"#@%4C#!A#H49#>$5<#$%H#Z9<";4%"#4%#"2<8!Y8#=42!8"#23<"<%=<6#><3<
N?9"89:4"@<:<=$:K4L<%!%8M84?R
• b<D!%$3"#3<:$=<6#=4#?9"89:4"@<:<=$:#"!?9:$;4%#<5<3H#?4%=>
N>c2JddCCCM4L<%!%8M84?d6<A$9:=M$"2e!!fUSTR
• F93#I9"!%<""J
• *4::$D43$;4%#N!M<M#73$%="1#=3$!%!%7R
• *4%"9:;%7
• (4BC$3<
• (4?<#/<8<%=#'<C"#Ng9:H#UT1#UPPaR>c2JddCCCM8:<?"4%M<69d%<C"344?d$3;8:<"dUPPadh9:HdFL<%i7!Bi3<"M2>2X
32