You are on page 1of 97

Chair of Structural Analysis

In Cooperation with:

Int. Masters Program on Computational Mechanics

Klinikum Grohadern

Title:

Determination of muscle forces acting on the femur


and stress analysis
Masterthesis

Author:

Silke Renner, Technische Universitt Mnchen

Student number:

2840064

Date:

November 2007

Supervisors:

Univ. Prof. Dr.-Ing Kai Uwe Bletzinger


Prof. Dr. med. Bernhard Heimkes
M.Sc. Christoph Mller

Abstract
Abstract
.A two-dimensional biomechanical model of the juvenile hip joint, called the
Stemmkrpermodel, was developed in the one-legged stance. The one-legged stance is the
relevant phase during walking considering the loading of the femur.
Forces acting on the apophysis of the greater trochanter in the one-legged stance were
computed based on anatomical and radiological investigations and the information was
integrated into Pauwels biomechanical hip model. A basic assumption was the validity of the
Pauwels bone remodeling law, which states that growth plates are oriented at a right angle to
the acting resultant forces.
According to the developed biomechanical Stemmkrpermodel, the greater trochanter
apohysis has to absorb a considerable force, equivalent to almost twice the body weight, and
is subject to pressure stress from a craniolateral direction. Thus, the apophysis of the greater
trochanter is a pressure apophysis, which was still not entirely clear.
Using the musculoskeletal modelling system AnyBody, the muscle force magnitudes and
the inclination of the muscle force vectors included in the Stemmkrpermodel during the
gait cycle were determined. Comparing these determined values with the expected values, the
lightly modified Stemmkrpermodel showed good agreement with the expected values.
Thus, the Stemmkrpermodel can be said to be verified by AnyBody.
In addition, a stress analysis of the femur using the finite element software Ansys was
performed, comparing the biomechanical Pauwels model and the Stemmkrpermodel in
the one-legged stance.

-i-

Acknowledgement
Acknowledgement

This thesis could not have been completed without the help and support of many
people to whom I am very grateful.
My gratitude is sincerely expressed to Prof. Dr. med. Bernhard Heimkes from
Klinikum Grohadern for his guidance, medical material support and expert knowledge.
Thanks to my supervisor Univ. Prof. Dr.-Ing Kai Uwe Bletzinger from the Technical
University of Munich for his encouragement and long and useful discussion about my work.
Special thanks to M.Sc. Christoph Mller from CADFEM for the assistance in the
generation of the FE model of the femur and the supply of the softwares: Simpleware,
AnyBody and Ansys.

Munich, November 2007


Silke Renner

- ii -

Table of Contents
Table of Contents

1 INTRODUCTION.............................................................................................................................................. 1
1.1 MOTIVATION FOR THE PROJECT ..................................................................................................................... 1
1.2 STRUCTURE OF THE THESIS............................................................................................................................ 1
2 BACKGROUND ................................................................................................................................................ 3
2.1 TERMINOLOGY DESCRIBING THE MUSCULOSKELETAL SYSTEM ...................................................................... 3
2.2 HIP ANATOMY ............................................................................................................................................... 4
2.3 THE FEMUR: A HOLLOW BONE ....................................................................................................................... 5
2.3.1 The inner architecture .......................................................................................................................... 6
2.3.2 Classification according to the CCD angle .......................................................................................... 6
2.3.2 Antetorsion and Retrotorsion ............................................................................................................... 7
2.4 MUSCLES....................................................................................................................................................... 7
2.4.1 Structure of skeletal muscle .................................................................................................................. 7
2.4.2 Muscles of the lower extremity ............................................................................................................. 8
2.5 DIFFERENT KINDS OF JOINTS IN THE HUMAN BODY ........................................................................................ 9
2.5.1 Joints of the lower extremity................................................................................................................. 9
2.6 GAIT CYCLE................................................................................................................................................. 10
2.6.1 Stance and Swing phase...................................................................................................................... 10
2.6.2 Alternative Nomenclature................................................................................................................... 11
3 THEORY .......................................................................................................................................................... 13
3.1 MATHEMATICAL MODEL ............................................................................................................................. 13
3.1.1 Pauwels model .............................................................................................................................. 13
3.1.2 Modification aspects of the Pauwels model ................................................................................... 15
3.1.3 Heimkes Stemmkrpermodel .......................................................................................................... 17
3.1.3.1 Forces acting on the juvenile hip in the one-legged stance ........................................................................... 19

4 SUBJECT OF STUDY IN RESEARCH: FEMUR ....................................................................................... 20


4.1 BIOMATERIAL BONE.................................................................................................................................... 20
4.2 THE EARLY STEPS ........................................................................................................................................ 22
4.3 LOADING CONDITIONS SIMULATED IN THE LITERATURE .............................................................................. 24
4.4 CONSTRAINTS SIMULATED IN THE LITERATURE ........................................................................................... 25
5 METHODOLOGY........................................................................................................................................... 26
6 ANALYSIS USING ANYBODY..................................................................................................................... 28
6.1 BACKGROUND ANYBODY ........................................................................................................................... 28
6.1.1 Kinematical Analysis .......................................................................................................................... 28
6.2.1 Inverse Dynamics ............................................................................................................................... 29
6.1.2 Redundancy problem .......................................................................................................................... 31
6.1.4 Optimization in ANYBODY................................................................................................................. 32
6.1.4.1 Min/Max criterion......................................................................................................................................... 32

6.1.5 Programming language: AnyScript .................................................................................................... 35


6.1.6 ANYBODY Model Repository ............................................................................................................. 35
6.2 ANYGAIT MODEL ..................................................................................................................................... 36
6.2.1 Degrees of freedom............................................................................................................................. 36
6.2.2 Marker driven model .......................................................................................................................... 37
6.2.3 Force plates: Ground reaction forces ................................................................................................ 38
6.2.4 Boundary condition ............................................................................................................................ 39
6.3 EXPECTED DISPLACEMENT OF THE BODY CENTRE OF MASS DURING GAIT .................................................... 39
6.3.1 Calculation of the body centre of mass............................................................................................... 41
6.3.1.1 Force platform method ................................................................................................................................. 41
6.3.1.2 Full body model............................................................................................................................................ 43
6.3.1.3 Visualization of COM position during gait cycle ......................................................................................... 45

6.4 MUSCULOSKELETAL MODELLING IN ANYBODY ...................................................................................... 46


6.4.1 Direction of AnyViaPointMuscle pull ............................................................................................ 47
6.4.2 ViaPointMuscles of the Stemmkrpermodel .............................................................................. 48

- iii -

Table of Contents
6.5 DETERMINATION OF THE HIP JOINT FORCE ................................................................................................... 50
6.5.1 Comparison of the hip joint force: Heimkes/AnyBody........................................................................ 50
6.5.2 Correction of the hip joint force ......................................................................................................... 50
6.5.2.1 New nomenclature: Stemmkrpermodel ................................................................................................... 52

6.6 VERIFICATION OF THE STEMMKRPERMODEL ......................................................................................... 53


6.6.1 Verification of the force vectors ......................................................................................................... 54
6.6.2 Verification of the angles.................................................................................................................... 54
6.6.3 Visualization of the force parallelogram at the greater trochanter.................................................... 56
6.7 COORDINATE TRANSFORMATION ................................................................................................................ 57
7 GENERATION OF THE FE MODEL OF THE FEMUR ........................................................................... 58
7.1 SIMPLEWARE SOFTWARE ............................................................................................................................. 58
7.1.1 ScanIP: image processing software.................................................................................................... 58
7.1.2 ScanFE: mesh generation / material assignment module................................................................... 61
7.1.2.1 Correction of the material properties ............................................................................................................ 62

8 STRESS ANALYSIS USING ANSYS ............................................................................................................ 66


8.1 APPLIED LOADS ........................................................................................................................................... 66
8.1.1 Hip joint force..................................................................................................................................... 66
8.1.1.1 Determination of the hip joint load application point ................................................................................... 67

8.2 CONSTRAINTS.............................................................................................................................................. 67
8.3 POSTPROCESSING IN ANSYS ........................................................................................................................ 67
8.3.1 Stress tensor........................................................................................................................................ 68
8.3.2 Stress vector........................................................................................................................................ 68
8.3.3 Calculation of normal and shear stresses of a defined cutting plane ................................................. 69
8.4 ANALYTICAL SOLUTION .............................................................................................................................. 69
8.4.1 Bending stress of a cutting section ..................................................................................................... 69
8.4.2 Shear stress of a cutting section ......................................................................................................... 72
8.4.3 Visualization of the results in GiD...................................................................................................... 73
8.4.3.1 Total normal stress........................................................................................................................................ 73
8.4.3.2 Total shear stress .......................................................................................................................................... 74

8.5 EQUILIBRIUM CHECK ................................................................................................................................... 75


8.5.1 Internal equilibrium............................................................................................................................ 75
8.6 RESULTS: COMPARISON PAUWELS MODEL / STEMMKRPERMODEL ........................................................ 78
8.6.1. Transversal cut: FEM and analytical solution .................................................................................. 78
8.6.2. Cut through the femoral head/neck ................................................................................................... 81
9 SUMMARY AND DISCUSSION ................................................................................................................... 83
10 REFERENCES............................................................................................................................................... 84

- iv -

Table of Figures
Table of Figures
FIGURE 2.1: Anatomical body planes

FIGURE 2.2: Hip anatomy

FIGURE 2.3: Femur in ventral view and in dorsal view

FIGURE 2.4: Trabecular and cortical bone of the femur

FIGURE 2.5: Coxa normala (left); coxa vara (middle); coxa valga (right)

FIGURE 2.6: Antetorsion, normal torsion and retrotorsion

FIGURE 2.7: Structure of an arbitrary skeletal muscle

FIGURE 2.8: Different kind of joints in the human body

FIGURE 2.9: One complete gait cycle

10

FIGURE 2.10: All events of the stance / swing phase

11

FIGURE 2.11: Alternative nomenclature for the gait cycle

11

FIGURE 3.1: 16. Phase of walking after Fischer and Braune

13

FIGURE 3.2: Pauwels mathematical model

14

FIGURE 3.3: Femur end from the lateral view

15

FIGURE 3.4: Increase of the angle EY during growth (left);


Angle AY remains nearly constant during growth (right)

16

FIGURE 3.5: Scheme of the two force parallelograms (Pauwels, Heimkes)

18

FIGURE 3.6: Computed force vectors dependent on the body weight

19

FIGURE 4.1: Force-deformation curve of bones with an elastic


and plastic deformation region

20

FIGURE 4.2: Ultimate strength of the cortical and compact bone

21

FIGURE 4.3: Direction/magnitude of the hip joint force in dependency of the CCD-angle 21
FIGURE 4.4: Stress trajectories in a crane design compared with trabeculae in a femur

22

FIGURE 4.5: Lines of stress in the upper femur (Koch)

23

FIGURE 4.6: Percent of the literature according to the applied load

24

FIGURE 5.1: Flow chart of the followed steps within the project

26

-v-

Table of Figures

FIGURE 6.1: Mechanical model representing the musculosketal system

29

FIGURE 6.2: Movement chain of gait analysis

30

FIGURE 6.3: Simple arm model with three muscles

31

FIGURE 6.4: Simple arm model

33

FIGURE 6.5: Table showing the relative muscles activities

34

FIGURE 6.6: model with applied forces (F1=2F2) (left); relative activities
computed by the iterative method (right)

34

FIGURE 6.7: Two parts of the Repository

35

FIGURE 6.8: DOFs per bone and constraints per joint

36

FIGURE 6.9: Infrared reflecting markers on a human body

37

FIGURE 6.10: Visualized markers

37

FIGURE 6.11: Typical force plate

38

FIGURE 6.12: Scheme of vertical / lateral displacement during one gait cycle

40

FIGURE 6.13: Displacements of COM: saggital and horizontal view

40

FIGURE 6.14: Vertical ground reaction forces given by textfiles

42

FIGURE 6.15: Start and end position of the gait cycle in AnyBody

42

FIGURE 6.16: Definition of one step interval of the force plate method

43

FIGURE 6.17: Full body model (no muscles)

43

FIGURE 6.18: Vertical/Lateral displacement of the COM (AnyBody)

44

FIGURE 6.19: The generic three component Hill-type muscle-tendon model


described by Zajac

47

FIGURE 6.20: Muscle from origin to insertion

47

FIGURE 6.21: Via-point muscles

48

FIGURE 6.22: gluteus minimus (green) and gluteus medius (blue)

48

FIGURE 6.23: gluteus maximus (second part): insertion / origin / three via-points

49

FIGURE 6.24: Comparison of the hip joint force Heimkes/AnyBody

50

- vi -

Table of Figures
FIGURE 6.25: Activity patter of ilipsoas and rectus femoris

51

FIGURE 6.26: Comparison of the corrected hip joint force Heimkes/AnyBody

51

FIGURE 6.27: New nomenclature Stemmkrpermodel

52

FIGURE 6.28: Computed force vectors dependent on the body weight

53

FIGURE 6.29: Expected force values according to the body weight of 64.9 kg

53

FIGURE 6.30: Loading response: Computed muscle force vector values in kg

54

FIGURE 6.31: Global coordinate system of AnyBody and resultant force vector Rh

54

FIGURE 6.32: Loading response:


Angles to the perpendicular of the computed force vectors

55

FIGURE 6.33: Loading response: planar angle to the perpendicular in the frontal plane

56

FIGURE 6.34: Trochanter resultant in the frontal and lateral view at 9%

56

FIGURE 6.35: Global and local coordinate system of the femur

57

FIGURE 7.1: Simpleware software products

58

FIGURE 7.2: Slice of the femur in the x-z plane

58

FIGURE 7.3: cavities in mask1

59

FIGURE 7.4: The marrow of the femur of one slice

59

FIGURE 7.5: Three defined masks of the femur

60

FIGURE 7.6: Comparison of the femur before and after


applying the Recursive Gaussian Filter

60

FIGURE 7.7: Voxel model of the proximal femur

61

FIGURE 7.8: Displacement of the femur at 9% of the gait cycle


(loading according Heimkes)

62

FIGURE 7.9: assigned material properties from ScanFE; E-modulus versus density

62

FIGURE 7.10: Comparison of the conversion formula

63

FIGURE 7.11: E-modulus versus density (VAKHUM: femur)

64

FIGURE 7.12: blue elements (cortical bone); red elements (spongiosa bone)

65

FIGURE 7.13: Displacement of the femur at 9% of the gait cycle


(loading according Heimkes)

65

- vii -

Table of Figures
FIGURE 8.1: Hip joint force in the frontal plane

66

FIGURE 8.2: Link elements for determination of load application point

67

FIGURE 8.3: illustration of the stress vector defined on a cutting plane

68

FIGURE 8.4: shear/normal part of the stress vector

69

FIGURE 8.5: scheme of delaunay triangulation in a plane

70

FIGURE 8.6: point cloud (left); triangulated point cloud (middle); total area (right)

71

FIGURE 8.7: illustration of the procedure of the calculation of the bending moment

71

FIGURE 8.8: approximated inner and outer diameter of the cross section

72

FIGURE 8.9: total normal stress [Pa] in the hollow circle cross section

73

FIGURE 8.10: total normal stress in the real complex cross section

73

FIGURE 8.11: comparison compression/tension values over the cross-section

74

FIGURE 8.12: Heimkes: total shear stress (left); tau_yz (middle);


tau_xz (right); unit [Pa]

74

FIGURE 8.13: Pauwels: total shear stress (left);


tau_yz (middle); tau_xz (right); unit [Pa]

75

FIGURE 8.14: illustration of the interpolation normal stress values

76

FIGURE 8.15: internal equilibrium

77

FIGURE 8.16: stress in z-direction [MPa] of the Pauwels model

78

FIGURE 8.17: stress in z-direction [MPa] of the Stemmkrpermodel

78

FIGURE 8.18: Pauwels model: view from above

79

FIGURE 8.19: Stemmkrpermodel: view from above

79

FIGURE 8.20: comparison pure moment result

80

FIGURE 8.21: Ansys total shear force

80

FIGURE 8.22: defined cutting planes of the proximal femur

81

FIGURE 8.23: Heimkes: normal and shear stresses of the femoral head

81

FIGURE 8.24: Normal stresses [Pa] of the cut through the femoral neck

82

- viii -

1 Introduction

1 Introduction
1.1 Motivation for the project
Physiological loading of the bone or implant is of great importance for investigations
involving micro-motion, fracture fixation/healing and for implant design and its primary
stability. Considering hip joint replacement operations, the question may arise to what degree
a certain surgical approach modifies the load distribution within the femur and the maybe
resulting consequences of this decision.
Especially in biological processes such as fracture healing the fact that muscles are major
contributors to femoral loading plays an important role. Nevertheless, the knowledge of
musculoskeletal forces is still limited. The existing possibilities to measure muscle forces in
humans by invasive methods are excluded due to ethical considerations. However, the noninvasive methods such as electromyography are used to get an insight of the activation patter
of muscles, but for the prediction of the magnitudes of muscle forces they cannot be utilized.
Therefore, the only opportunity to estimate the complex distribution of muscle forces is
offered by computer analysis on the basis of optimization algorithms.
Pauwels was one of the first researchers, who included the effect of muscles on femoral
loading in his analytical analysis for the one-legged stance. His determined mathematical
model of the hip joint force can be found in nearly all biomechanical books and is accepted
and valid until today.
Prof. Dr. med. Heimkes from Klinikum Grohadern in Munich postulates in his developed
Stemmkrpermodel based on anatomical, radiological and computational results, that the
Pauwels model has to be modified. If muscle activity is considered, there is a general
consensus that the muscle forces tend to reduce the load acting within the bone. Thus, the
bending of the femur compared with the Pauwels model is expected to be reduced.
In this project, the musculoskeletal modelling system ANYBODY will be used to determine
the muscle forces of the so-called Stemmkrpermodel of Prof. Dr. med. Heimkes. The aim
is to verify the found magnitudes of muscle forces and their inclination angle to the vertical of
the Stemmkrpermodel.
Furthermore, a Finite Element Analysis of the loaded femur will be performed, comparing the
biomechanical model of Pauwels with the Stemmkrpermodelof Prof. Dr. med. Heimkes.
1.2 Structure of the thesis
In this thesis the main point of investigation is the femoral loading in the one-legged during
walking, but also the complete gait cycle will be investigated. Therefore, in chapter 2, some
general background information such as the terminology describing the musculoskeletal
system, the architecture of the femur, the definition of the gait cycle etc. will be described.
In chapter 3 the underlying theory of the mathematical Pauwels model, which determines the
loading of the proximal femur will be presented. Based on the Pauwels model the
Stemmkrpermodel from Prof. Dr. Heimkes will also be presented and the important facts
which lead to its development will be explained.

-1-

1 Introduction
Due to the fact that the femur is a subject of research since years, chapter 4 deals with the
bone as biomaterial and the early steps of investigations done on the femur starting by
Galileo. Additional, an overview of different loading conditions and applied constraints of
performed femur analysis will be summarized.
Chapter 5 illustrates in a flow chart the methodology of this project and the softwares which
will be used to reach the aim of the project.
The proceeding steps for the determination of the muscle forces (magnitudes and their
inclination to the perpendicular) involved in the Stemmkrpermodel and the verification
according to Prof. Dr. med. Heimkes computed/expected muscle forces will be done in
chapter 6 by using the software ANYBODY. In addition, some background information to the
software and the way of functioning of the gait model will be explained.
After having determined the load case of the femur a finite element model of the femur
considering inhomogeneous material distribution has to be built up from CT images. This
procedure is described in chapter 7.
In chapter 8 a stress analysis of the femur comparing the Pauwels model with the
Stemmkrpermodel in the one-legged stance will be performed.
In the last chapter 9 the found results will be discussed and also the realism of them will be
considered. Furthermore, recommendation for future work will be done.

-2-

2 Background

2 Background
2.1 Terminology describing the musculoskeletal system
For the description of the human body and its movement anatomical terminology is used. A
three dimensional coordinate system consisting of three anatomical planes is defined as
follows:

Coronal Plane (Frontal Plane):


vertical plane running from side to side
divides the body in anterior and
posterior portions
Sagittal Plane (Lateral Plane):
vertical plane running from front to
back
divides the body into right and left side
Axial Plane (Transversal Plane):
horizontal plane
divides the body into upper and lower
parts

Figure 2.1: Anatomical body planes [1]

Median Plane:
sagittal plane through the midline of
the body
divides the body into right and left
halves

The three anatomical body planes are used to determine an anatomical position and the axes
of motion.
To describe the positions of structures relative to other structures or locations in the body
directional and spatial terminology is necessary:

superior or cranial:
inferior or caudal:
anterior or ventral:
posterior or dorsal:
medial:
lateral:
proximal:

distal:

toward the head end of the body; upper


away from the head; lower
toward the front
toward the back
toward the midline of the body
away from the midline of the body
refers to structures nearer to the trunk
(e.g. the knee is proximal to the foot)
refers to structures further from the trunk
(e.g. the foot is distal to the knee)

-3-

2 Background
In addition some important terms of movement will be explained next:

Flexion / extension: increasing angle with frontal plane / decreasing angle with frontal
plane
Abduction / adduction: moving away from or toward the sagittal plane
Protraction / retraction: moving forward or backward along a surface
Elevation / depression: raising or lowering a structure
Medial rotation / lateral rotation: movement around an axis of a bone
Pronation / supination: placing palm backward or forward (in anatomical position)
Circumduction: combined movements of flexion, extension, abduction, adduction
medial and lateral rotation circumscribe a cone
Opposition: bringing tips of fingers and thumb together as in picking something up

2.2 Hip anatomy


In anatomy, the hip is the bony projection of the femur which is known as the greater
trochanter, and the overlying muscle and fat.
The following figure 2.2 is illustrating the hip anatomy:

Figure 2.2: hip anatomy [2]


The hip comprises the joint between the upper end of the femur and the pelvis or acetabulum.
This joint is a so-called ball-and-socket joint, where the femoral head forms the ball portion
and the round area of the lower pelvis known as acetabulum forms the socket portion. Thus,
the femoral head fits inside the acetabulum and both bones are covered by a protective layer,
which acts as a cushion, called cartilage.
A diseased hip shows degeneration of the cartilage, the consequences are swelling,
inflammation and pain. Sometimes even bone will begin to contact with bone.
In turn, cartilage is covered by a synovial fluid, which has the function of a lubricant to reduce
wear during joint motion.
The primary function of the hip joint is to support the body weight in the static (e.g. standing)
and dynamic (e.g. walking) posture.

-4-

2 Background
2.3 The femur: a hollow bone
The femur or the so-called thigh bone is the longest, most voluminous and strongest bone in
the human skeleton. The following figure 2.3 illustrates the ventral and dorsal view of the
femur and presents the most important terms and definitions:

head

greater trochanter

greater trochanter

neck

lesser trochanter
corpus femoris

condyles

Figure 2.3: Femur in ventral view (left) and in dorsal view (right) [3]
The femur can be divided into three main sections:
1. the upper extremity (proximal extremity)consists of:

a rounded head which articulates with the acetabulum of the hip


bone to form the hip joint
a relatively narrow neck
two protuberances for muscle attachment, the greater trochanter and
the lesser trochanter

2. the body or shaft (corpus femoris):

it is almost cylindrical in form


it is a little bit broader above than in the centre
it is slightly arched: convex in front and concave behind

3. the lower extremity (distal extremity):

it is larger than the upper extremity and consists of two oblong


eminences known as the condyles

-5-

2 Background
2.3.1 The inner architecture
The next figure 2.4 shows the two major kinds of bone structure of the femur, the trabecular
(spongiosa/spongy) bone and cortical (compact) bone:

Trabecular bone

Cortical bone

Figure 2.4: Trabecular and cortical bone of the femur [4]


The outside of the shaft of the femur consists of cortical (solid) bone and this type of bone
forms the outer shell of all bones of the human body. Trabecular bone gives supporting
strength to the ends of the weight-bearing bone and is found at the expanded heads of long
bones.
Considering the area near the hip joint of the femur, it is filled with a micro-framework of
very fine small struts of bone (spongiosa). This fine framework near the joint has the function
to distribute the load from the hard bone wall over a larger area and to act as a shock absorber
or dash pot.
During standing on one leg, the neck of the femur has to transmit about 2.5 to 6 times the
body weight BW as axial loading due to lever relationships. To cope with this high load a
well adapted design has to exist. The cortical bone has a higher density and stiffness than the
spongiosa bone and is therefore better adapted to higher local stresses.
2.3.2 Classification according to the CCD angle
The angle formed between the neck and shaft of the femur is described as the CCD (CaputCollum-Diaphysis) angle . According to this CCD angle a division between three cases can
be made as shown [5]:

Figure 2.5: Coxa normala (left); coxa vara (middle); coxa valga (right)
-6-

2 Background
Coxa normale is defined at a CCD angle between 120 and 135. At birth the CCD angle is
normally between 135 and 140 and should gradually reduce with development to the adult
norm of 125. If the CCD angle reduces too much coax vara (CCD angle < 120) is resulting
and if the reduction is inadequate coxa valga (CCD angle > 135) is resulting.
Coxa vara and coxa valga are deformities of the hip which are of clinical importance, but out
of the scope of this thesis.
2.3.2 Antetorsion and Retrotorsion
Looking in the direction of the mechanical axis of the femur (shaft axis) from proximal to
distal, the projection of the neck of the femur is not covered by the axis of the condyles.
Antetorsion is an increase in the angle of the head and neck of the femur relative to the
femoral condyles and in healthy people the antetorsion is around 12 also called normal
torsion. In turn, retrotorsion is a decrease in the angle of the head and neck of the femur
relative to the femoral condyles.
For purposes of clarity, the following figure 2.6 illustrates the superimposition of normal
torsion, retrotorsion, and antetorsion [6]:

antetorion 30
normal torsion 12
retrotorsion 8

Figure 2.6: Antetorsion, normal torsion and retrotorsion


The resulting clinical consequences are out of the scope of the thesis and will not be discussed
here.
2.4 Muscles
Muscles are the actuators of the human body. Muscles can be separated into three different
types:
Skeletal muscles
Cardiac muscles
Smooth muscles
For the movement of the body segments the skeletal muscles are responsible and their
structure will be shortly explained in the next section.
2.4.1 Structure of skeletal muscle
Each person has around 600 skeletal muscles which have a wide range in size. A whole
skeletal muscle is considered as an organ of the muscular system. It consists of skeletal
muscle tissue, connective tissue, nerve tissue, and blood or vascular tissue.

-7-

2 Background
The following figure 2.7 shall just give an impression how the structure of an arbitrary
skeletal muscle looks like:

Figure 2.7: Structure of an arbitrary skeletal muscle [7]


In some muscles the fibres are parallel to the long axis of the muscle, in some they converge
to a narrow attachment, and in some they are oblique. Each muscle fibre is a single cylindrical
cell and one individual muscle can contain hundreds, or even thousands of these fibres
bundled together and wrapped in a connective tissue (Epimysium) covering [7].
Typically a muscle spans a joint and is attached to bones by tendons at both ends. One of the
bones remains nearly fixed or stable while the other end moves as a result of muscle
contraction. Muscle fibres are oriented either in the direction of the tendon or at an acute
angle (pennation angle) to the tendon.
The primary function of skeletal muscle is contraction. Before a skeletal muscle fibre can
contract, it has to receive an impulse from a nerve cell. Therefore, skeletal muscles have a
supply of blood vessels and nerves. Generally, an artery and at least one vein accompany
each nerve that penetrates the epimysium of a skeletal muscle. Just this small insight in the
structure of skeletal muscles reveals the complexity of this topic and will not be explained
more in detail in this project.
2.4.2 Muscles of the lower extremity
This project deals just with the muscles of the lower extremity, especially the muscles acting
on the femur. Therefore, a brief overview of the function of the muscles that move the thigh
and the leg will be given:
Thigh flexors: Iliacus and Psoas major
Thigh adductors: Adductor magnus, Adductor longus and Gracilis
Extensor of the thigh: Gluteus maximus and Tensor fascia latae
Thigh abductors: Gluteus medius and Gluteus minimus
Thigh extenders: Biceps femoris, Ssemimembranosus and Semitendinosus
Knee extenders Quadriceps femoris:
Rectus fermoris, Vastus lateralis, Vastus medialis and Vastus intermedius
Further details about the anatomical location, origin and insertion points will be explained in
the muscle modelling section in chapter 6.

-8-

2 Background
2.5 Different kinds of joints in the human body
The following figure 2.8 [5] gives an overview of the existing kinds of joints in the human
body with their according degrees of freedom (DOF):

a)

e)

d)

c)

b)

f)

a.)
b.)
c.)
d.)
e.)
f.)
g.)

plane joint
hinge-joint
Radgelenk
condylar joint
elipsoid joint
saddle joint
ball and socket
joint

g)

Figure 2.8: Different kind of joints in the human body

2.5.1 Joints of the lower extremity

In this project the lower extremity is of special interest. Therefore, just the according joints
are considered more in detail:
1. hip joint: ball and socket joint (3 DOF)
The hip joint is the joint between the femur and the acetabulum of the pelvis and its
primary function is to support the weight of the body in both static (e.g. standing) and
dynamic (e.g. walking or running) postures.
Seven different kinds of movement are possible in the hip joint:
Abduction and adduction of the femur
Internal (medial) and external (lateral) rotation of the pelvis, thigh or
spine
Circumduction of the femur or pelvis
Flexion and extension on or from the spine (Wirbelsule) or on or from
the thigh (Oberschenkel)
2. knee joint: condylar joint (2 DOF)
The condylar joint is a joint allowing primary movement in one plane flexion,
extension) with small amounts of movement in another plane (rotation).
3. ankle joint: hinge joint (1DOF)
The hinge (ginglymus) joint allows movement in one plane (flexion, extension) and is
termed uniaxial.

-9-

2 Background
2.6 Gait cycle
The gait cycle is sometimes also called the walking cycle. The term gait cycle describes the
whole activity, from the heel which is first put on the ground (heel strike HS) and the
following heel contact with the same feet (HS) [8].
The next figure 2.9 represents one complete gait cycle:

Figure 2.9: One complete gait cycle


From the above presented figure 2.9 it is visible that the complete gait cycle can be divided
into two main phases:
stance phase
swing phase
The stance phase (ca.60%) takes longer than the swing phase (ca.40%). During the stance
phase of one leg, the other leg is in the swing phase, which is the shorter phase. This results in
the double limb support phase where both legs are on the ground.
Double limb support occurs for two periods of 12% of the gait cycle in a normal walk. Double
limb support occurs between heel strike of the limb and toe off of the contralateral limb.
Single limb support where just one foot is on the ground occurs for two periods of 38% of the
gait cycle in a normal walk
In general, as walking speed increases double support time is reduced until it is eliminated
and the gait changes to running.
2.6.1 Stance and Swing phase

The stance phase consists of five events based on the movement of the foot. The first event of
the stance phase is the contact of the heel on the ground, heel strike (HS), and ends with the
event where the toes of the same leg quitting the ground, toe-off (TO). In the heel strike the
centre of mass is at lowest position.
The three events between the heel strike and the toe-off are in order:

Foot-flat (FF): plantar surface of the foot is on the ground


Midstance (MS): the swinging foot passes the stance foot, body centre of
gravity is at highest position
Heel-off (HO): the heel loses the contact with the ground
- 10 -

2 Background
The swing phase starts in the moment when the stance phase is ending. It takes from the toes
lift-off to the heel contact of the same leg and is described by three events:

Acceleration: begins as soon as the foot leaves the ground


Midswing: same event as the event midstance of the stance phase, just
the feet are interchanged
Deceleration: action of the muscles slowing down the leg and stabilize
the foot for the next heel strike

The next figure 2.10 summarizes all the events of the stance phase and the swing phase [9]:

Figure 2.10: All events of the stance / swing phase


2.6.2 Alternative Nomenclature

An alternative nomenclature describing the gait cycle is developed by the famous gait analyst
Perry and her associates at Rancho Los Amigos Hospital in California (Cochran, 1982) shown
in the next figure 2.11:

Figure 2.11: Alternative nomenclature for the gait cycle


In this project this alternative nomenclature [10] is more reasonable and will be used in the
proceeding sections.

- 11 -

2 Background
The different phases are given in percentage of one gait cycle:
1.
2.
3.
4.
5.
6.
7.
8.

Initial contact (0%)


Loading response (0 - 12%)
Midstance (13 - 31%)
Terminalstance (32 - 50%)
Pre-swing (51 - 62%)
Initial swing (63 - 75%)
Midswing (76 86%)
Terminal swing (87 100%)

- 12 -

3 Theory

3 Theory
3.1 Mathematical model
In this section the biomechanical Pauwels model and the developed Stemmkrpermodel by
Prof. Dr. med. Heimkes will be explained. These two models are the base of investigation in
this project. Both models are two dimensional and illustrated in the frontal plane in the onelegged stance phase of the gait cycle (loading response).
3.1.1 Pauwels model
Braune and Fischer, two famous gait analysts, divided in their experiments one walking
period into 31 phases. The next figure 3.1 shows the 16. Phase of one walking period, the so
called one legged stance:

Figure 3.1: 16. Phase of walking after Fischer and Braune


The red point in the above figure represents the partial body weight, consisting of the trunk,
the head, both arms and the swinging leg, which has to be balanced by the hip joint of the
stance leg.
According to Pauwels the hip joint force during standing or slow walking mainly depends on
the torque around the hip joint centre H, caused by the partial body weight G5 (body weight
minus the weight of the stance leg).
Pauwels used gait data from Fischer to determine the center of gravity (d5) of the weight G5
during standing. He also located the point T where the abductor muscles M have their
insertion point at the greater trochanter. The origin point A at the pelvic bone of these muscles
was found in anatomical studies.

- 13 -

3 Theory
Using all the available data made it possible to calculate the resultant hip joint force R from
the equilibrium of forces and moments acting around the hip joint centre H [11]:
Fx = 0 :
Fz = 0:
MH = 0:

Mx + R x = 0
Mz + Rz + G5 = 0
M * m = G5 * d5

(3.1)
(3.2)
(3.3)

The following figure 3.2 represents Pauwels simplified static one legged stance model in
the frontal plane (corresponding to the 16.phase from Braune and Fischer) [12]:

Figure 3.2: Pauwels mathematical model


All other muscle forces which may be active during the standing phase to stabilize the body
concerning flexion and stretching of the hip joint are negligible.
Pauwels proved that this model can be applied for the one legged stance and the period when
the swinging leg passes the frontal plane during slow walking. Dynamic forces are not
considered in this model.
Pauwels also recognizes the problem of bending in the leg. He searched for a muscle which
decreases the bending and found the muscle tensor fascia latae and gluteus maximus (tractus).
- 14 -

3 Theory
3.1.2 Modification aspects of the Pauwels model
In the previous described biomechanical Pauwels model theoretical aspects of the growthinducing forces acting on the capital epiphysis were studied, but no computations were done
of the forces acting on the apophysis of the greater trochanter. Actually, until today it is still
not completely clear whether the apophysis of the greater trochanter is a traction apophysis
or a pressure apophysis.
The aim of the investigations was to analyze the forces acting on the trochanter apophysis in
the one-legged stance and integrate them into the biomechanical hip model of Pauwels.
The development of a two-dimensional vectorial model of the load on the juvenile hip in the
one-legged stance was based on two kinds of data:
1. Anatomical data: The anatomical data were received from 16 anatomic specimens
from newborns to children aged 14 years. Additional 6 specimens from adults were
taken to determine the normal cross-sectional areas of the muscles inserting on the
greater trochanter.
2. Radiological data: The radiological data were received from 1350 hip joints of healthy
children in a cross-sectional radiological study. In this study 11 biomechanical
relevant angles and length were measured.
In the following points the anatomic and radiological results are listed:
1. Anatomic results:

Insertion areas of the greater trochanter apophysis


o lateral surface is trapezoidal
o ventral surface is triangular

Muscles insertion of the greater trochanter apophysis:


o gluteus medius muscle has a ribbon-like insertion on the lateral surface
o gluteus minimus muscle has the insertion on the triangular ventral surface
o vastus lateralis muscle originates both from the lateral trapezoid and from the
ventral triangle
The following figure 3.3 shows the femur end from the lateral view with the insertion
areas of the greater trochanter apophysis [13]:
Insertion area of the muscle
Gluteus medius

Insertion area of the muscle


Gluteus minimus

Periost

Origin area of the muscle


Vastus lateralis
ventral

dorsal

Figure 3.3: Femur end from the lateral view


- 15 -

3 Theory

Normal cross-sections of muscles inserting on the apophysis of the greater trochanter


(mean physiologic muscle cross-section of six anatomic specimens):
o gluteus medius muscle : 27.78 cm2
o gluteus minimus muscle : 10.05 cm2
o vastus lateralis muscle : 25.45 cm2
o vastus intermedius muscle : 10.69 cm2

Inclination of the greater trochanter growth plate (the proximal end of the femur was
cut into 3 to 6 mm thick slices in the coronal plane):
o the larger part of the apophyseal growth plate (90.4% of the total area on the
average) is oriented parallel to the femoral neck
o the smaller, dorsal part of the apophyseal growth plate (9.6% of the total area
on the average) is almost oriented perpendicular to the shaft

2. Radiological results:

most important results:


o the epiphyseal angle EY increases steadily during growth inclination of the
capital growth plate in the coronal plane
o the apophyseal angle AY remains nearly constant during growth inclination
of the greater trochanter growth plate in the coronal plane
The next figure 3.4 illustrates the position of the angles EY and AY in the femur and
their change in angle during growth [14]:

EY

AY

years

years

Figure 3.4: Increase of the angle EY during growth (left); angle AY remains nearly constant
during growth (right)
- 16 -

3 Theory

From the anatomic and radiological results the following conclusions can be drawn, on which
subsequent computations are based:
1. Forces acting on the greater trochanter apophysis can be calculated from the muscle
forces exerted by the gluteus maximus and the tensor fascia latae in tensing the
iliotibial tract, the gluteus medius and minimus muscles, and parts of the knee
extensors that are connected to the greater trochanter apophysis by the vastus lateralis
muscle.
2. The results of the normal cross-sections of the muscles indicate that the muscles
gluteus medius and gluteus minimus are connected to the counteracting vastus lateralis
muscle by a common tendinous junction at the apophysis of the greater trochanter (socalled vastogluteal muscular sling).
3. Approximately 90% of the area of the greater trochanter growth plate is a plane
oriented nearly parallel to the femoral neck a force RT is acting on the greater
trochanter apophysis (resultant of all muscular forces acting on the greater trochanter
apophysis) which can be postulated to act perpendicular to the greater trochanter
growth plate.
Considering all the facts of the anatomic and radiological investigations it can be stated that
the greater trochanter apophysis is a pressure apophysis.
After having computed the forces acting on the greater trochanter apophysis in the one-legged
stance these forces can be integrated into the biomechanical hip model of Pauwels. The result
is the Stemmkrpermodel, which will be presented in the next section.
3.1.3 Heimkes Stemmkrpermodel
The Stemmkrpermodel consists of two vectorial force parallelograms, which are reflecting
the loading of the femur in the one-legged stance:
1. The Pauwels hip parallelogram G5/M/R with the hip joint resultant R described in
section 3.1.1.
2. The trochanter parallelogram M/Mfsc/RT with the trochanter resultant RT, which
reflects the loading of the greater trochanter.
Therefore, the femur is stressed by two resultant forces R and RT.
The direction and magnitude of the partial body weight G5 is known and the direction of the
hip resultant force R can be measured by the epiphyseal angle EY. The magnitude of R and
the direction of M can be calculated with the magnitude of M received from measurements
and computation of the ratio of the lever arms.
The computation of the trochanter parallelogram forces M/Mfsc/RT acting on the trochanter
apophysis included the following steps:

All muscles pulling the greater trochanter apophysis in a cranial direction are
summarized in the resultant force M. The muscles included in M are gluteus maximus,
tensor fascia latae, gluteus medius and gluteus minimus. Due to the fact that the
resultant force M is nearly equivalent to the muscle resultant force M of Pauwels
parallelogram. Thus, M is known.
The direction of the trochanter resultant RT is also known because it is perpendicular
to the apophyseal angle AY.
- 17 -

3 Theory

The direction of all distal muscle forces acting on the greater trochanter apophysis,
Mfsc, can be assumed to act parallel to the femoral axis. The resultant force Mfsc
contains the muscles gluteus maximus, tensor fascia latae, vastus lateralis, vastus
medialis and vastus intermedius.

Thus, if the angle and absolute value of M is known and also the angle of RT and Mfsc, then
the absolute values of RT and Mfsc can be calculated by using the following formulas [14]:
RT = M

sin(M + M fsc )

(3.4)

sin(RT M fsc )

M fsc = M

sin[180 (RT + M )]
sin(RT M fsc )

(3.5)

The next figure shows the scheme of the Stemmkrpermodel with the two vectorial force
parallelograms:
M
G5
R

1. gluteus medius

Mpt

2. gluteus minimus

Msc

T
Msc
RT

Mfsc
Mfsc

3.

gluteus maximus

4.

tensor fascia latae

1.

gluteus maximus

2.

tensor fascia latae

3.

vastus lateralis

4.

vastus intermedius

Mpt
Msc

Mfc

Figure 3.5: Scheme of the two force parallelograms (Pauwels, Heimkes)

- 18 -

Msc

Mfc

3 Theory

3.1.3.1 Forces acting on the juvenile hip in the one-legged stance


Based on all previous described investigations and computations a table reflecting all the
resultant forces of the Stemmkrpermodel depending on the body weight at the end of the
growing period near the adult age in the one-legged stance can be presented [14]:

Figure 3.6: Computed force vectors dependent on the body weight


The magnitude of the forces and the inclination to the perpendicular of the force vectors of the
above figure 3.6 should be verified in this project.

- 19 -

4 Subject of study in research: femur

4 Subject of study in research: femur


4.1 Biomaterial Bone
Bone tissue appears in a lot of microstructure forms, which have different material densities.
Bone tissue with a density above 1.5 g/cm3 is referred as cortical (compact) bone and below a
density of 1 g/cm3 as spongiosa bone [15].
The behaviour of bone tissue under compressive loading concerning the mechanical
properties and the failure limit shows a significant dependency on the density of the material.
Bone tissue is mechanically a two-phase material. First it behaves elastic and after a certain
point plastic as the next figure 4.1 illustrates:

Figure 4.1: Force-deformation curve of bones with an elastic and plastic deformation region
The failure criterions can be determined by two points (see figure 4.1):
transition between the elastic (reversible) and the inelastic behaviour (begin of the
permanent damage): region between the points B and D
yield point
stress at the ultimate failure point C
tensile strength

The mechanical material properties of bones are depending on the direction of loading. Thus,
bones have an anisotropic structure. During body activities forces and moments are
transmitted in different directions to the bony structure and produce tensile stress and strain,
compressive stress and strain as well as shearing stress and strain. These forms of loading can
appear in all combinations. The largest strength of the bone is presented by loading in the
longitudinal direction (compression). A load case transversal to the central axis of the bone
(shear stress) shows the lowest strength. Thus, the stiffness of the bone is largest in the
longitudinal direction and lowest perpendicular to the central axis of the bone.

- 20 -

4 Subject of study in research: femur


Spongiosa bone based on the elastic region has an E-modulus of approximately 1Gpa and the
much stiffer cortical bone 20Gpa. For comparison, steel has an E-modulus value of 100Gpa.
The failure limit of the bone material is different for tension and compression. Considering
the cortical bone the failure limit for compression is circa 30-50% higher than for tension. As
an evidence of the limit loading average values for the cortical and spongiosa bone were
found in the literature [15] as the next figure 4.2 shows:

Compressive stress
Tensile stress

Cortical bone Spongiosa bone


200MPa
20MPa
135MPa
14MPa

Figure 4.2: Ultimate strength of the cortical and compact bone


Muscle forces can significantly influence the stresses and stress distributions in bones.
Especially, shearing stresses can be reduced or completely eliminated. The consequence of
additional muscle forces is an increase in the compressive stresses. Therefore, the tolerance
limit of bones for compressive stresses is much larger than for shearing stresses.
The direction of the force transmission to the bone is of great importance regarding the limit
load exceeding. If the load is not transmitted in the direction of the longitudinal axis of the
bone, the complete mechanical load is increasing on the bone surface.
Joint forces, which are not acting along the longitudinal axis of the bones, are in general
compensated by muscle forces. Thus, also the CCD angle of the femur plays an important role
considering the hip joint force as illustrated [11]:

Figure 4.3: Direction/magnitude of the hip joint force in dependency of the CCD-angle
In this project, inhomogeneous material properties of the femur will be considered, which
obey Hooks law and the according procedure is described in chapter 7. For simplification,
isotropic material behaviour will be assumed.

- 21 -

4 Subject of study in research: femur

4.2 The early steps


The load situation within the femur has been a subject of study in research for many years.
In 1638, Galileo studied the mechanics of long bones and analyzed the cross anatomical
structure of the femur [17]. His work based on the assumption that there exists a relationship
between mechanical principles and the anatomical shape of the bones.
Culmann, a German civil engineer, who has developed a technique called graphical statics for
determining the direction of internal structures of complex systems, studied the drawing of the
Swiss anatomist von Meyer, who set the beginning of serious research directed in uncovering
the influence of the mechanical environment on trabecular structures.
The drawings of von Meyer were based on the observations of cadaver specimens. The
analyzed stress pattern of Culmanns curved column structure showed remarkable similarity to
the trabeculae pattern of that von Meyers of the femur. Therefore, Culmann and von Meyer
postulated that trabeculae are oriented along principle stresses.
Around 1892, Wolff quoted the work of Culmann in which the advantages offered by the
orientation of the trabecular structure could be interpreted by mechanical means. With his
ideas about the relation of stimulus and anatomy Wolff brought outspread support to the idea
of von Meyer. Wolff denoted his work as Trajectorial Theory of trabecular alignment and
adapted von Meyers drawings according to the established fact in mechanics that internal
stress directions must always intersect perpendicular for any load case.
The next figure 4.4 compares the drawing of the internal structure of the proximal femur
(trabeculae) adapted by Wolff with the stress trajectories in a crane design according to
Culmann:

Figure 4.4: Stress trajectories in a crane design (left) compared


with trabeculae in a femur (right)
Although Wolff described the adaption of bone to mechanical stimulus (Wolffs law or
Law of bone remodelling), it was the American anatomist Koch (1917), who first
performed an analysis of the femur by calculating the cross-sectional area and moment of
inertia of seventy-five locations of the femur to quantify the stresses and strains.

- 22 -

4 Subject of study in research: femur


A maximum shear force was present at the femoral head and decreased towards the lesser
trochanter; the maximal bending moments occurred at the level of the lesser trochanter. Koch
compared shear stress with density and trabeculae directions with principle stress directions.
It can be seen from Kochs lines of stress result that the upper femur is composed of two
distinct systems of trabeculae arranged in curved paths as follows:

Figure 4.5: Lines of stress in the upper femur (Koch) [18]

One with its origin in the medial (inner) side of the shaft and curving upward in a
fan-like radiation to the opposite side of the bone
The second, having origin in the lateral (outer) portion of the shaft and arching
upward and medially to end in the upper surface of the greater trochanter, neck and
head.

These two systems intersect each other at right angles.


With this analysis Koch confirmed Wolffs statements and he also recognized that bone
density is highest in areas of highest shear stress. However, Kochs analysis was only capable
of representing femoral curvatures in the frontal plane and he neglected to include muscle
activities. The exclusion of muscle activity led to an underestimation of the joint contact force
and therefore also to an underestimation of the femoral loads. Nevertheless, Kochs work is
considered as the classical approach to femoral stress calculations which were the bases of
numerous research studies on this subject.
The first researcher, who included the effect of muscles on femoral loading in his analytical
analysis, was Pauwels (1950) as described in the previous chapter. However, the three
dimensional characteristic of the bone was ignored. Although the Pauwels model has been
criticized [19], it is still accepted and can be found nearly in every biomechanical book until
today.
Summarized the important points are:
1. Trabeculae orientation according to principle stress directions
2. Highest density in areas of highest shear stress
3. Change in the loading of the bone causes adaptation of bone structure

- 23 -

4 Subject of study in research: femur


The next points are still the basis of many current research projects:
4. Bone cells may be regulated by local stresses
5. Results show no correlation of stress to cell activity
6. Relationship of bone structure to mechanics was derived without considering
physiological mechanisms

4.3 Loading Conditions simulated in the literature


In the literature a lot of different systems of forces can be found. One of the reasons for this
variability is the fact that different load situations are simulated, for example, two-leg stance,
single-legged stance, or different phases of gait.
The most common set-up found in literature is the one that simulates a single force at the
femoral head, neglecting everything else (e.g. Crowninshield, Huiskes et al., Koch, Rohlmann
etc.). The second most common set-up in addition to the first most common set-up simulates
the action of the abductors muscles in a single force (e.g. Crowninshield, Crick et al.,
Okumura et al., Rohlmann etc) [20]. However, a large variability considering the magnitude
and the direction of the abducting force exists. The third most common set-up is the one
which includes the hip joint force, the abducting force and the iliotibial tract (e.g. Finlay et al.,
Prati et al., Rohlmann et al. etc). In several works there exists incomplete information about
the magnitude and direction of these forces. In very few studies, only numerical ones,
simulated a system of more than three forces (e.g. Ferre et al).
In the following figure 4.6 the percentage of works in the literature that apply different setups, as described earlier, is illustrated [20]:
Three muscles or more
Hip force, abductors, iliotibial tract

Only hip joint force

Hip force + abductors

Figure 4.6: Percent of the literature according to the applied load


In summary, only a few researchers included muscle forces as an important influence on the
load conditions in the femur in their investigations. In all of the investigations, nearly no one
tried to use more than three to four muscle groups in modelling and most of the works are
restricted to two-dimensions. A complete description of the internal loads of the femur was
not found, which incorporates the forces from all muscle attachments to the bone and also a
documentation of a complete free body diagram of the femur was not found. However,
methods to quantify the load condition in the femur are currently available, but none of the
found publications considered all the forces acting on the femur.

- 24 -

4 Subject of study in research: femur

4.4 Constraints simulated in the literature


In a large number of cases existing in the literature it is not clear what sort of constraints is
applied to the femur.
The other large number of cases of constraints found in the literature is to lock the knee, thus
assuming that this distal lock applies the constraints physiologically exerted by the knee
muscles. Therefore, most finite element models are fully constrained distally, with the forces
being applied proximally in the desired direction [20]. The corresponding experimental
situation is when the femur is distally cemented and shear plates are used to avoid any
undesirable horizontal force component.
Thus, the distally constrained femur can be said to be the preferable standard due to the fact
that it is easily reproducible experimentally and also in finite element models.

- 25 -

5 Methodology

5 Methodology
To reach the aim of this project the following flow chart was constructed and will be
followed:

CT Scanning
of the femur

Musco-skeletal model

Segmentation of
the CT data

Material
properties

Reconstruction
of the solid
geometry

Verification of the
Stemmkrpermodel

Smoothing the
geometry

Importing
geometry

Gait analysis

Computation of the
desired muscle forces

Load case
determination

Meshing

FEM anaylsis

Figure 5.1: Flow chart of the followed steps within the project
The first step will be to get familiar with the software ANYBODY, which will be used to
calculate the muscle forces during walking. The loading response is of special interest due to
the fact that the Pauwels model and the Stemmkrpermodel are referring to this phase of the
gait cycle. The aim is the verification of the force magnitudes and the inclination to the
perpendicular of the considered force vectors of the Stemmkrpermodel in the loading
response with ANYBODY.

- 26 -

5 Methodology
However, it is also of interest how the trochanter resultant force RT of the
Stemmkrpermodel is behaving in the other phases of the gait cycle. After the
determination of the desired muscle forces for the Stemmkrpermodel the load case for the
Finite Element Model of the femur is also defined.
The second step will be to build up a finite element model of the femur. Therefore, an existing
right leg femoral bone will be scanned by CT. From the CT images the finite element model
will be reconstructed.
The final step is the finite element analysis according to the determined load case by
ANYBODY. Of special interest is the loading response phase of the gait cycle, where the two
different load conditions of the Pauwels model and the Stemmkrpermodel will be analysed
and compared.

- 27 -

6 Analysis using ANYBODY

6 Analysis using ANYBODY


ANYBODY is a software system for modelling the mechanics of the human body developed
by the Aalborg University in Denmark. It can handle models with hundreds of muscles on
ordinary personal computers. As calculation output forces in individual muscles, joint forces,
metabolism, elastic energy in tendons, antagonistic muscle actions, and many other useful
properties of the working human body are accessible.
Application fields of ANYBODY are automotive, medical/rehabilitation, aerospace,
occupational health and sports.

6.1 Background AnyBody


6.1.1 Kinematical Analysis
An unconstrained segment in space has six degrees of freedom. Therefore, if the model has n
segments, the model will have a total of 6n degrees of freedom unless some of them are
constrained somehow. The purpose of a kinematical analysis is to determine the position of all
segments at all times. Thus, kinematical analysis is about solving 6n equations with 6n
unknowns.
Adding joints, n jnt (number of joints), to the model is a way to constrain the degrees of
freedom. However, to provide all 6n constraints by adding joints would make the system
unmovable. Therefore, usually a body model will have enough joints to keep the segments
together and few to let the model move. After the joints, n jnt , have taken their part of degrees
of freedom, AnyBody utilizes so-called drivers, n drv (number of drivers), which are added to
the system to resolve the remaining unknowns in the system up to the required number of 6n.
Joints can also be understood as kinematical measures equipped with drivers. For instance, a
spherical joint is a distance between two points on two different segments that is driven to be
zero.
Solving the system of 6n equations, the system is said to be kinematically determinate and
usually this is necessary to perform a kinematical analysis. However, there are also some
cases where the system is solvable although the number of equations is different to 6n. In few
cases the system cannot be solved even though there are 6n equations available. Both cases
are connected with redundant constraints.
According to the following formula:
Determinacy d = 6n n jnt n drv

(6.1)

Three cases according to the determinacy can be distinguished:

d = 0: kinematically determinate (suitable for inverse dynamics)


d > 0: kinematically indeterminate (forward dynamics needed)
d < 0: kinematically over-determinate (statically indeterminate)

It is an important point to keep track of the number of constraints and the number of degrees
of freedom is quite important.

- 28 -

6 Analysis using ANYBODY

6.2.1 Inverse Dynamics


Essentially there are existing two types of problems in rigid body dynamics:

Direct Dynamics Problem: Known forces are applied to a mechanical system and the
objective is to determine the motion of the system.
Inverse Dynamics Problem: The motion of a mechanical system is known and the
objective is to determine the forces that cause the motion of the system.

Due to the fact that the direct measurement of the tension in muscles, the forces and moments
transmitted by the joints of the human body, and the activation of the peripheral and central
nervous system is difficult or sometimes impossible the Inverse Dynamic approach is used
in the software ANYBODY.
In the analysis using the inverse dynamic approach the human locomotor system is modelled
as a kinematic chain, consisting of single bone segments which are connected by frictionless
joints. The skeletal bones are represented by rigid body segments. Each segment has six
degrees of freedom (three translational and three rotational), whereas through the connection
of two segments according to the kind of joint a certain number of degrees of freedom is
restricted.
On each single segment there are acting muscle forces, forces as a result of gravity and
inertia, joint contact forces and external forces (e.g. ground reaction force) as the next figure
6.1 represents [21]:

Figure 6.1:
Left: Mechanical model representing the musculosketal system (foot, tibia and femur),
the resultant volume forces of a segment act at the segment centre of mass
Middle: Free body diagram resulting from the sectional cut A-A and B-B, at the cutting
section muscle forces and joint contact forces
Right: Combination to a resultant joint force and resultant joint moment

- 29 -

6 Analysis using ANYBODY


For every segment, according to the Newtonian laws, the following equations have to be
fulfilled:
Equilibrium of forces:

Equilibrium of moments:

r F

(6.2)

= mb &r&b
i

(6.3)

= b & b

mb : Mass of the segment b


r&&b : Acceleration of the segment b
Fi : All forces acting on the segment b

b : Moment of inertia regarding the centre of gravity of the segment b


& b : Angular acceleration of the segment b
rb : Vector from the segment centre of gravity to the force point of application
The computational scheme proceeds distal to proximal and is based on three assumptions:

a complete kinematics analysis has been carried out


the geometric and mechanical parameters of each segment are known
external forces are known (e.g. ground reaction forces)

Gait analysts are able to measure four components in the movement chain which are
highlighted in the next figure 6.2:

Electromyography

Tension in
muscles

Joint forces
and moments

Equation of
motion

Segment masses and


moments of inertia

Velocities and
accelerations

Anthropometry of
Skeletal segments

Segment
displacements

Ground
Reaction forces

Figure 6.2: Movement chain of gait analysis [9]

- 30 -

6 Analysis using ANYBODY


Electromyography (EMG) is not able to measure the muscles tension but can give an idea
about the activation pattern of the muscles. Via skin-or needle electrodes relayed electrical
signal shows the examiner an impression about sequence, force and intensity of the muscle
activity during the gait cycle. The emitted EMG signals can be quantified manually,
electronically or with a computer program and reflect a quite objective figure of the muscle
activity. Thus, the electromyo-graphical investigations represent, which muscle in which
moment during walking is active.
As seen in the above figure 6.2, segment anthropometry may be used to generate the segment
masses, whereas the displacements of the segments may be double differentiated to yield
accelerations. Ground reaction forces are used with the segment masses and accelerations in
the equations of motion which are solved in turn to give resultant joint forces and moments.
Some gait analysts measure all of those four highlighted components.
In inverse dynamics as mentioned above, the external loads on the body and the motion is
assumed to be known (e.g. from motion capturing devices). The output of the computation is
the internal forces. If the internal forces are just joint moments and joint reaction forces,
this becomes then in most cases a straightforward procedure, involving the solution of a
system of linear equilibrium equations.
However, for the computation of individual muscle forces, inverse dynamics leads to a
redundancy problem, which will be described in the next section.

6.1.2 Redundancy problem


Considering, for example, the elbow joint:
shoulder
biceps brachii
brachialis
upper arm

elbow joint

brachioradialis

lower arm

Figure 6.3: Simple arm model with three muscles


3 muscles for 1 DOF (rotation of lower arm around elbow joint) redundant system
A musculoskeletal system is usually a redundant system meaning that the number of muscles
nM is greater than the number of degrees of freedom nDOF.
Therefore, a desired motion can be achieved by an infinite number of activation patterns of
muscles. In nature, the central nervous system (CNS) overtakes the task to determine how
much force each individual muscle has to provide. Constructing an algorithm to determine the
activation of each muscle therefore entails guessing the motives behind the CNSs function.
ANYBODY overcomes the redundancy problem with an optimization strategy. Mathematical
optimization, applied to the distribution of muscle forces, is a well-known technique for
simulation of the muscle recruitment.
- 31 -

6 Analysis using ANYBODY

6.1.4 Optimization in ANYBODY


Several optimization criteria can be found in the literature. For example, the so-called
polynomial criterion, minimization of weighted sum of muscles loads raised to some power,
has become popular [21]. This criterion leads to non-linear optimization problems which need
to be solved by time-consuming and sophisticated numerical methods. Therefore, a criteria
leading to linear optimization problem is generally more attractive in terms of more ease
implementation and efficiency.
Numerical efficiency in the development of the optimization algorithm has been a motivation
factor with several aims. Efficient computations are convenient considering large models, for
example full body models, analyst work with. Secondly, numerical efficiency is important for
parameter optimization which can be used for several purposes such as model parameter
estimation, design optimization of man-driven artefacts, and simulation of human motion
patters.
In the following the min/max muscle recruitment criterion used in ANYBODY is presented.
Large body models contain of more than 100 muscles and are still solved in fractions of a
second on a standard PC for one configuration of an inverse dynamics simulation.

6.1.4.1 Min/Max criterion


Based on the assumption that muscles are recruited according to an optimality criterion leads
to the task choosing the right one.
The min/max muscle recruitment [22] can be expressed mathematically as an optimization
problem with an objective function called G.
Minimize objective function:
G( f
f

fi

(M )

(M )

(6.4)

: vector of all muscle forces


: ith muscle force

Subjected to:
(6.5)

Cf = d
fi

(M )

0,

i 1,..., n

(M )

(6.6)

C is the coefficient matrix and d the right hand side of equation (6.5) consists of external
forces, inertia forces, and passive elasticity in the tissues of the body.
Due to the fact that muscles cannot push an additional equation (6.6) was formulated.
In the AnyBody Modeling System, a min/max criterion is used for the objective function G:

G( f

(M )

) = max

fi(M )
Ni

- 32 -

(6.7)

6 Analysis using ANYBODY


To change the non-smooth min/max problem into a linear problem an artificial variable, , is
introduced. This variable has the function as an upper bound on all muscle activities:
f i(M )
,
Ni

i 1,..., n ( M )

with 0 f i ( M ) N i

(6.8)

Muscle activity is defined as a relative measure of the muscle force, f i ( M ) / N i , where N i is the
momentary strength of the muscle i.
Via the so-called bound formulation a linear programming problem can be obtained with
muscle forces and joint reactions as free variables. Due to the fact that joint reactions are free
in sign and without side constraints, they can be eliminated from the equation system. Thus,
the result is a linear program with as many unknowns as there are muscles.
Due to the fact that the min/max formulation only cares about the maximal activity of the
muscles another problem arises. Therefore, only a subset of the muscles is actually
represented in the objective which leads to indeterminacy. It is also observed that groups of
sub-maximally activated muscles may not be determined uniquely.
There exist three categories of such sub-maximal muscle groups:
Counter-working muscles
Parallel muscles, i.e., muscles with the same function
Independent sub-systems (one limb in a multi-limb model)

The problem which arises from counter-working muscles can be solved by adding penalties to
the objective G:
n( M )
f (M )
G = + i
(6.9)
Ni
i =1

But these penalties dont handle the remaining two other sub-maximal muscle groups.
This problem called for an iterative solution scheme, where each iteration step eliminates
muscles which are uniquely determined and removes their contribution to the support of the
external load from the right hand side of the equation. The next iteration can then determine
the sub-maximal muscles. This procedure continues until there are no muscles left in the
system.
Considering that an analysis involves many time steps and each time step involves the
determination of thousand muscles it is obvious that this is a very demanding numerical task.
To get a better understanding of how the procedure works a small example will be given.
The following figure 6.4 shows a simple model of the arm with an applied load:

Figure 6.4: Simple arm model [22]

- 33 -

6 Analysis using ANYBODY


Next a table [22] is presented (figure 6.5) which shows the relative muscles activities in
percent for three cases for the angle = 0:

1. raw min/max solution having


counter-working muscles
2. elimination by linear penalty
(as mentioned above)
3. iterative solution method

Figure 6.5: Table showing the relative muscles activities

Case 1: simultaneous activity of the elbow extensors and flexors


Case 2: solution is still not satisfactory, not all elbow flexors are active
Case 3: all three mono-articular elbow flexors are activated equally, they are so-called parallel
muscles
Having a look at the next figure 6.6, a sub-maximal activated sub-system exists in one arm
when a higher force is applied to one arm than to the other. The corresponding relative
activities computed by the iterative method for one revolution of the external forces (F1 = 2F2)
are shown as well:

Figure 6.6: model with applied forces (F1=2F2) (left); relative activities computed by the
iterative method (right)

It is visible from both activation profiles that they have the same form but different
magnitudes by a factor of two (according to the applied forces). This proofs that the iterative
min/max solution method does handle the two sub-systems uniquely, which would not have
been the case we the raw min/max solution method.

- 34 -

6 Analysis using ANYBODY

6.1.5 Programming language: AnyScript


AnyScript is the modelling language of ANYBODY for the development of multibody
dynamics models. It is an object oriented modelling language with a fixed set of available
data types (classes). This means it is not possible to define own data types and write, like in a
programming language, algorithms. An AnyScript file is purely consisting of definitions
(declarations) of objects and the internal data structure in ANYBODY maps these definitions
completely. Syntactically it is not unlike Java or C++, but perhaps it resembles most of all
JavaScript.
An AnyScript model is roughly divided into two main sections:

The body model containing the definition of the mechanical system


The study section containing lists of analyses and other operations which
can be done on the model (e.g. kinematical analyse)

6.1.6 ANYBODY Model Repository


Developing accurate human body models is a complex challenging task. Therefore a so-called
library ANYBODY Model Repository [23] of models exist that scientists and other
advanced users have developed and made available in the public domain.
The ANYBODY Model Repository uses elements of the AnyScript language such as include
files, parameters, and equipping parts with their own interface to make it possible:

To change the model pieces to fit to a given task preferably without tampering with
the interior workings of the parts that are used.
To combine existing body parts to larger models.
To find parts and attach them to construct bits of an own new model.

The Repository files are divided into two main groups:

Repository

ARep

BRep

Figure 6.7: Two parts of the Repository

ARep: Application Repository

The Application Repository contains various devices, environments, and working situations
of different AnyScript models. For example, existing applications are models lifting a box or
riding a bicycle. For analyzing these application models the main files can be loaded in the
ANYBODY Modelling System and different studies can be performed.

- 35 -

6 Analysis using ANYBODY


BRep: Body Repository

The Body Repository contains AnyScript models with no specification of movements,


forces, supports and no attachments such as environmental devices. Due to the fact that the
entire body model consists of hundreds of muscles which is heavy computational, the BRep
directory is structured to link applications to subsets of the body model such as the lower
extremities in this project.

6.2 AnyGait model


Movement analysis has now become an important clinical tool to diagnose gait disorders. For
this purpose a universal gait model was developed [23]. However, an estimation of muscle
forces during a gait cycle will give precise information for the treatment of gait pathologies,
particularly, in surgical cases.
The available Gait3D model in the ARep branch of the Repository consists just of the
lower extremity with 7 rigid segments (thigh, shank and foot for both legs and the pelvis) and
35 muscles in each leg. The model contains three different kinds of joints which are
connecting the segments:
1. HIP: spherical joint it is a point constraint, also known as a ball-and-socket joint
2. KNEE: revolute joint it is an ideal hinge joint that only allows rotation about one
axis
3. ANKLE: universal joint it allows rotation about two perpendicular axis

6.2.1 Degrees of freedom


The next table lists the degrees of freedom per bone and the constraints per joint:

Segments # bones DOFs per bone total


Foot
2
6
12
2
6
12
Shank
2
6
12
Thigh
1
6
Pelvis
6

42

Joint type # joints Constraints per joint total


Hip
2
3
6
2
5
10
Knee
2
4
8
Ankle

24

Figure 6.8: DOFs per bone and constraints per joint

42 DOFs minus 24 joint constraints 18 drivers are needed that the model is
kinematically determinate (d = 0).

- 36 -

6 Analysis using ANYBODY

6.2.2 Marker driven model


As mentioned in the previous section the gait model needs 18 drivers. The Gait3D model is
driven by markers and therefore it needs 18 marker coordinates.
The centres of the joints are needed for describing the movements of the patients. Due to fact
that markers cannot be put directly at the joints location inside the body of the patient,
external markers are used and their positions are measured throughout each investigated
activity.
The next figure 6.9 illustrates an exemplary system of infrared reflecting markers which are
used on the trunk and the single segments of the legs [24]:

Figure 6.9: Infrared reflecting markers on a human body

A Vicon system with six cameras and a sample rate of 50 Hz was used to measure the
position of body markers which were smoothed by 5th order splines. The coordinates of these
markers were recorded in a fixed laboratory coordinate system shown in the above figure.
The marker positions relative to the palpable bony landmarks were measured on the patient.
The location of joint centres and additional reference points to these landmarks, used for
calculation of rotations, were determined using individual CT data [24]. Thus, the calculation
of the coordinates of joint centres and reference points relative to the laboratory coordinate
system from the measured marker positions are possible.
However, the motion data for the Gait3D model are grabbed
from the book by Vaughan et al: Dynamics of Human Gait
[9], which are available through the public domain [25]. In
the Gait3D model only as many driver coordinates as
necessary are selected to avoid kinematical overdeterminacy.
Segments for each of the markers are created and driven
according to the data from the motion capture experiment.
Therefore, text files with the coordinate data from the
motion capture system for every marker (grey spheres) for
each time step is read in (see figure 6.10). Corresponding
markers are defined on the bones (blue spheres). The model
is driven by requiring coincidence between the free floating
markers (grey) and the markers on the bones (blue) for
selected DOFs.
Figure 6.10: Visualized markers

- 37 -

6 Analysis using ANYBODY

6.2.3 Force plates: Ground reaction forces


The most common force-measuring instrument used in gait analysis is the force plate. Force
plates are available in many configurations, sizes, and with various performance
characteristics.
For the Gait3D model in ANYBODY, the ground reaction forces (GRF), acting from the
floor to the supporting leg(s), were measured by two force plates.
To get a better understanding how these force plates work the figure 6.11 is presented:

Figure 6.11: Typical force plate

The force plate has four output sensors as seen in the figure 6.11 above. The typical
transducers used in piezoelectric-based force plates are four identical three component force
transducers; one placed at each corner of the plate. Walking direction is the positive ydirection.
The four output sensors produce the following eight outputs:
Fx1 + Fx2
Fx3 + Fx4
Fy1 + Fy4
Fy2 + Fy3
Fz1, Fz2, Fz3, Fz4

Channel 1: force in x-direction measured by sensor 1 and sensor 2


Channel 2: force in x-direction measured by sensor 3 and sensor 4
Channel 3: force in y-direction measured by sensor 1 and sensor 4
Channel 4: force in y-direction measured by sensor 2 and sensor 3
Channel 5-8: force in z-direction measured by sensor 1-4

These eight channels represent the four individual vertical forces measured, two shear forces
in the x-direction and two shear forces in the y-direction. To receive six ground reactions and
moments, the data are further reduced as follows:
Fx = (Fx1 + Fx 2 ) + (Fx 3 + Fx 4 )

(6.10)

Fy = (Fy1 + Fy 4 ) + (Fy 2 + Fy 3 )

(6.11)

Fz = Fz1 + Fz 2 + Fz 3 + Fz 4

(6.12)

M x = [(Fz1 + Fz 2 ) + ( Fz 3 Fz 4 )] b

(6.13)

M y = [(Fz 2 + Fz 3 ) + ( Fz1 Fz 4 )] a

(6.14)

M z = [(Fx 3 + Fx 4 ) + ( Fx1 Fx 2 )] b + (Fy1 + Fy 4 ) + ( Fy 2 Fy 3 ) a

(6.15)

The loads transmitted between the force plate and a body in contact with it can be determined
by a resultant force and a resultant moment.

- 38 -

6 Analysis using ANYBODY


When only compressive forces in the z-direction act on the force plate, the couple that can be
transmitted to the force plate would be in the x-y plane. This couple is referred as the free
moment.
To perform a 3D gait analysis also the point of application of the resultant force and the free
moment has to be known [26] and can be calculated from the measured force and moment
components expressed in the Cartesian coordinate system:

ax =
ay =

M y + Fx az 0
Fz
M x + Fy az 0

Fz
Tz = M z Fy ax + Fx ay

(6.16)
(6.17)
(6.18)

The point of the resultant force application is also known as the centre of pressure (COG).
The ground reaction data used in the Gait3D model are also grabbed from the book by
Vaughan et al: Dynamics of Human Gait [9].

6.2.4 Boundary condition


In general the question arises, especially for the latter calculation of the centre of mass, why
does the model work without the upper body.
The following boundary conditions are applied to the model:

The feet are loaded with the measured forces from force plate experiments [9].
The pelvis is supported by six reaction forces (they do not supply any motion!).
The reaction force provides the missing reactions between the lower extremity and the
upper body.

Theoretically it is possible to perform gait analysis without force plate data. Therefore, the
upper body has to be modelled which requires upper body movements and a high accuracy.

6.3 Expected displacement of the body centre of mass during gait


The body centre of mass (COM) is the point in the human body about which all the body
segments balance each other:

All linear forces acting on the human body are balanced, i.e. F = 0
All rotary forces acting on the human body are balanced, i.e. M = 0

- 39 -

6 Analysis using ANYBODY


During the gait cycle the body centre of mass translates along the walking direction but also
moves in a sinusoidal pattern in the lateral and vertical direction [10] as illustrated as follows:

Figure 6.12: Scheme of vertical / lateral displacement during one gait cycle

Considering the sagittal plane, the body centre of mass moves vertically through two full
oscillations during each gait cycle. Therefore, the vertical displacement curve has two peaks
and two troughs. The two minimum heights of the COM occur at loading response (5%) and
at preswing (55%) and the two maximum heights of the COM occur at midstance (30%) and
again at midswing (80%). The total excursion of the vertical displacement curve at average
walking speed is approximately 5cm.
Considering the transverse plane, the body centre of mass also oscillates laterally in a
sinusoidal pattern with a total excursion at average walking speed of approximately 4cm [].
Only one full lateral oscillation of the body centre of mass occurs during one gait cycle and
the minimum right position of the COM is at the end of midstance (30%). The maximum left
position of the COM is at midswing (80%).
The following figure 6.13 illustrates the displacements in the saggital and in the transverse
(horizontal) plane during one gait cycle:

Figure 6.13: Displacements of COM: saggital and horizontal view

- 40 -

6 Analysis using ANYBODY


It is also important to mention at this point, that the effect of walking speed plays a role on the
body centre of mass displacement. Investigations on the COM displacement in vertical and
lateral directions have been made and it was found out that the vertical displacement increases
as walking speed increases [27]. In opposite to the decreasing lateral displacement, the
walking speed increases. Thus, slow walking speeds require significant balance response due
to the large lateral displacement of the COM. It was also shown that vertical displacement of
the COM is highly correlated with energy consumption and therefore energetically and
metabolically costly.
The locomotor system has several methods to try to reduce the amplitude of the COM. These
methods are the six so-called determinants of gait, but they will not be discussed here.

6.3.1 Calculation of the body centre of mass


As mentioned in one of the previous sections, in the Gait3D model the measured reaction
forces provided the missing reactions between the lower extremity and the upper body.
Therefore, it was possible to just model the lower extremity.
In ANYBODY there exists a class called AnyKinCom, which is able to calculate the centre
of mass for an arbitrary number of desired segments.
However, this class cannot be used due to the fact that just the lower segments (pelvis, thigh
and shank) are available in the model. Another way has to be found.

6.3.1.1 Force platform method


A number of methodologies for calculating the body centre of mass exist. Some of these
methods uses kinematic data obtained from markers that are placed on the body and others
uses data obtained from force platforms.
Due to the fact that the Gait3D model is driven by force plate data the force platform
method will be used. A step is defined as the interval from initial contact of one foot to the
initial contact of the other foot [28].
The vertical acceleration a z (t ) is computed from the summed vertical ground reaction
forces Fz (t ) , less the body weight m g , and divided by the body mass m :
a z (t ) =

Fz (t ) m g
m

(6.19)

The vertical velocity of the body centre of mass is calculated by integrating the acceleration
over a single step:
t

1
v z (t ) = v0 + a z ( ) d = v0 + ( Fz (t ) m g ) d
m0
0

(6.20)

where v0 is the integration constant at the beginning of the step cycle. The integration
constant was determined by requiring the average vertical body centre of mass velocity to be
zero.

- 41 -

6 Analysis using ANYBODY


The vertical position of the body centre of mass z B (t ) is found by integrating the vertical
velocity over a single step:
t

z B (t ) = z 0 + v z ( ) d

(6.21)

where z 0 is the integration constant at the beginning of the step cycle, which is set to zero
since only the vertical displacement of the body mass centre is of interest.
In the Gait3D model two force platforms are used and both vertical forces given by the
textfiles and the total vertical force of both of them are presented:

Vertical forces given by the two force plate textfiles


1000

vertical force: force plate 1

900

vertical force: force plate 2

800

total vertical force

700

force [N]

600
500
400
300
200
100
0
-100

0,5

1,5

time [s]

Figure 6.14: Vertical ground reaction forces given by textfiles

In the following figure 6.15 the start position and the end position during walking of the
human model in ANYBODY of one gait cycle is shown:

Figure 6.15: Start and end position of the gait cycle in AnyBody

- 42 -

6 Analysis using ANYBODY


Comparing the definition of the step interval from initial contact of one foot to the initial
contact of the other foot, on which the vertical displacement calculations are based on, with
the start and end position of the AnyBody model, it can be seen that the AnyBody model
starts with the initial contact of the right foot but does not end with the initial contact of the
other foot (again the right foot).
According to the definition of the step interval there should be force data available of three
double support intervals and two single support intervals illustrated as follows:

Figure 6.16: Definition of one step interval of the force plate method

As it can be observed from figure 6.14 there exists just one double support phase. Therefore,
the calculation of the vertical displacement of the COM with the force platform method is not
possible because the force under the other leg during two of the three periods of double
support is not recorded. The force platform method is also error prone, if the integration
factors are not chosen accurately.

6.3.1.2 Full body model


Due to the fact that the Gait3D model just consists of the lower extremity as mentioned
earlier the AnyKinCom class of AnyBody to calculate the COM cannot be used. Therefore,
a full body model with no muscles from the BRep was loaded into the Gait3D model as the
next figure 6.17 illustrates:

Figure 6.17: Full body model (no muscles)

- 43 -

6 Analysis using ANYBODY


In the mannequin.any file, variables were set to control the rotation of the thorax with
respect to the pelvis. These values are overtaken and modified from the StandingModel
mannequin file from the ARep. The human model contains 29 body segments, which are put
in the class AnyKinCom to compute the COM.
The AnyKinCom class calculates the position from full body kinematics for every time
step. Therefore, the total body centre of mass for one time step is the weighted sum of the
centre of mass of every segment of the body:

COM =

m j pi, j

(6.22)
j

mj: mass of segment j


pi,j: ith component (i = x, y, z) of the position vector of its centre of mass
Then a kinematical analysis was performed to get the position values for every time step.
The received result values can just be considered as approximated position values because of
the missing movement information of the arms. Therefore, the arm movement is neglected in
the analysis, which is assumed to play a secondary role in the determination of the COM.
Different arm postures were tried out, giving nearly all the same lateral/vertical displacements
as presented:
Vertical / Lateral displacement of the COM
Brep: Full body model no muscles
1,20

displacement [m]

1,00
0,80

vertical displacement
lateral displacement

0,60
0,40
0,20
0,00
0%

20%

40%

60%

80%

100%

120%

gait cycle [%]

Figure 6.18: Vertical/Lateral displacement of the COM (AnyBody)

As mentioned earlier the two peaks of the sinusoidal vertical displacement curve should be at
midstance (13 31%) and midswing (76 86%). The two troughs should appear at loading
response (0 12%) and at preswing (51 62%).The above presented sinusoidal pattern of the
vertical displacement of figure 6.18 shows good agreement according to the peaks and
troughs. The amplitude of the vertical displacement curve received from AnyBody using a
male human model is 3.4cm.
- 44 -

6 Analysis using ANYBODY


Only one full lateral oscillation of the body centre of mass should occur during one gait cycle
and the minimum right position of the COM should be at the end of midstance (30%). The
maximum left position of the COM should be at midswing (80%). Comparing with the above
figure 6.18 the result can be assumed not to be completely correct.
Thus, also marker coordinates for the upper body have to be used from motion capture
systems to drive the upper body to give correct results. It can be stated that the vertical
displacement of the COM is not as strong influenced by the arm movement as the lateral
displacement of the COM.

6.3.1.3 Visualization of COM position during gait cycle


The position coordinates (x, y and z) of the COM for every time step of the gait cycle were
copied from the output data received from the AnyKinCom class, including 29 body
segments, into a textfile. The first column of the textfile is the time followed by the columns
with the x, y and z coordinate values.
First, a segment with mass = 0 and principle moments of inertia = 0 is created with the
AnySeg class including the visualization of a node for the COM:
AnySeg Segment = {
r0 =.ComData(Main.Study.tStart);
//initial position
Mass = 0;
Jii = {0.0,0.0,0.0};
//principle axis of inertia
AnyRefNode node = {
sRel = {0,0,0};
//relative position vector
AnyDrawNode drw = {
//visualization
ScaleXYZ = {0.03,.03,.03};RGB = {1,0,0};};
};};

In the next step two objects, AnyKinLinear and AnyKinRotational, are created to
measure the translation, the position of one reference frame with respect to the global
reference frame, and the rotations.
AnyKinLinear LinCom = {
AnyRefFrame &ref1 =..GlobalRef;

// the reference frames that


define the vector

AnyRefFrame &ref2 = .Segment;


};

AnyKinRotational RotCom = {
AnyRefFrame &ref1 =..GlobalRef;
AnyRefFrame &ref2=.Segment;
Type=RotVector;
};

- 45 -

// measures 3D
orientation/rotation

6 Analysis using ANYBODY


Then a kinematic measure is used to for driving the motion of the COM by adding a
kinematic driver to the AnyKinLinear object.
AnyKinEqInterPolDriver LinDrvCom = {
AnyKinLinear &ref1 = .LinCom;
Type=Bspline;
BsplineOrder = 8;
FileName =.datafile;
};

// interpolates between data points


spaced in time
// measure that should be used

//reference to the name of the data


file to use

The AnyKinEqInterPolDriver interpolates between the given data points; it is a very useful
driver for the import of sampled kinematic data.
In addition, an AnyEqSimpleDriver is used, which drives the AnyKinRotational object.
This kind of driver is often used to set constant values or constant velocities.
Therefore, the model is again kinematical determinate.

6.4 Musculoskeletal Modelling in ANYBODY


A muscle model is a description of how a muscle behaves under different operation
conditions. A traditional muscle model takes an activation signal and a present muscle state as
input and produces a force as output. AnyBody, instead of taking an activation signal as input
produces the muscle active state as output by using inverse dynamics. Thus, it is necessary to
know the muscle properties involved. AnyBody has three different muscle models included,
based on the classical work by A.V.Hill in 1938, which are ranging from simple to more
complicated physiological behaviour:
1. AnyMuscleModel:

assuming constant strength of the muscle regardless of


its working condition

2. AnyMuscleModel3E:

a three element model taking serial and parallel elastic


elements into account along with fibre length and
contraction velocity

3. AnyMuscleModel2ELin:

a bilinear model taking length and contraction velocity


into account

The AnyGait3D model is using the muscle model AnyMuscleModel3E. Therefore, just
this muscle model will be considered more in detail.
To represent a muscle contraction in a mechanical formulation the so-called three component
Hill model is often used in simulations. This model consists of the following three
components:

contractile element (CE) : representing the active force of the muscle fibres
two non-linear elastic elements:
o parallel elastic element (PE): representing passive properties of surrounding
tissue
o series elastic element (T): representing the tendon and other elastic tissues

- 46 -

6 Analysis using ANYBODY


The AnyMuscleModel3E is a muscle-tendon actuator model based on the three component
Hill-model proposed by Zajac in 1998, including a pennation angle .
The next figure 6.19 illustrates the scheme of the AnyMuscle3E model:

Figure 6.19: The generic three component Hill-type muscle-tendon model described by Zajac

Received from experimental studies muscle architectural properties of the muscle force-fibre
length, the muscle force-fibre velocity and tendon force-tendon length relationships, together
with peak isometric force, optimal muscle fibre length and pennation angle at optimal fibre
length, tendon slack length and maximum shortening velocity are used as input data.
In the AnyBody Modeling system, muscles mechanically consist of two separate
computational models, the strength model such as the three previous mentioned models and
the kinematical model.
Three different kinematical models are existing in AnyBody, which are defining the muscles
path from origin to insertion depending on the posture of the body:
1. AnyViaPointMuscle
2. AnyShortestPathMuscle
3. AnyGeneralMuscle
In this project just the kinematical model AnyViaPointMuscle will be used and explained in
the following two sections.

6.4.1 Direction of AnyViaPointMuscle pull


Muscles can be just defined by two points (a straight line), from origin to insertion. In these
cases the direction of the pull is from the origin point to the insertion point as illustrated:

Figure 6.20: Muscle from origin to insertion

- 47 -

6 Analysis using ANYBODY


There are also existing muscles that are modelled not only with the insertion and origin point.
These muscles or its tendon is bent out of a straight line by a bony process or ligament so that
it runs over a pulley-like arrangement, the so called Via-points in AnyBody.
The direction of muscle pull is then naturally bent out of line as illustrated in the next figure
6.21:

Figure 6.21: Via-point muscles

6.4.2 ViaPointMuscles of the Stemmkrpermodel


Anatomically, via point muscles are mostly found in the lower extremities and the spine.
Therefore, all the muscles of the lower extremity which will contribute to the verification of
the Stemmkrpermodel are modelled as ViaPointMuscles. ViaPointMuscles are passing
through at least a set of two points on their way from origin to insertion. As many Via points
as needed can be used, they just have to be attached to a segment or the global reference
frame of the model. The approximation between these constraining points is a piecewise
straight line. The first and the last point of the set defining the muscle path are fixing the
muscle rigidly to them and transfer forces in its longitudinal direction to them. Conversely,
the muscle passes through the interior via points transfer only forces to interior via points
along a line that bisects the angle formed by the muscle on the two sides of the via point.
The muscles of the Gait3Dare modelled according to the data by Scott Delp [25].
The only two muscles with just origin and insertion point are Gluteus minimus and
Gluteus medius. Both muscles are divided into three parts with the same insertion point
around the greater trochanter as figure 6.22 shows:

Origin points

Insertion points

Figure 6.22: gluteus minimus (green) and gluteus medius (blue)

- 48 -

6 Analysis using ANYBODY


All the other muscles which contribute to the Stemmkrpermodel are modelled with at least
one Via-point:

gluteus maximimus:
o part 1: two via-points
o part 2: three via-points
o part 3: one via-point
tensor fascia latae: one via-point
vastus lateralis: two via-points
vastus medialis: two via-points
vastus intermedius: two via-points
rectus femoris: two via-points

The next figure 6.23 presents the second part of the gluteus maximus muscle (three viapoints):
origin

via-points

insertion

Figure 6.23: gluteus maximus (second part): insertion / origin / three via-points

For the determination of the muscle forces of the Stemmkrpermodel different things were
tried out. For example, it was tried out to shift the origin point of the muscle vastus lateralis
up to the greater trochanter, where it anatomically originates [13]. Due to the fact that the
vastus lateralis is a pennation muscle, which means if one elongates it, more muscle fibres are
added, which in turn created a much larger muscle force than the original model.
Therefore, it was decided to use the original muscle model according to the data from Scott
Delp.
However, when all the segments of the different muscles were considered, the calculated
muscles resultant, for example Msc, which should pull in downward direction to the knee,
was bended in an outward direction of the femur, which is physiologically not possible. Due
to this fact some segments of the ViaPoint muscles were not considered.
For the muscles vastus lateralis, vastus intermedius and vastus medialis an assumption was
done, which considers that the vastus lateralis, vastus intermedius and vastus medialis are one
muscle, which originates at the greater trochanter region.
- 49 -

6 Analysis using ANYBODY

6.5 Determination of the hip joint force


According to Heimkes Stemmkrpermodel the hip joint force is the resultant of the
summation of the abductor muscles M and the partial body weight wh.
It is also possible to get the hip joint force directly from AnyBody. In AnyBody the ground
reaction forces are used with the segment masses and accelerations in the equations of motion,
which are solved in turn to give resultant joint forces and moments.

6.5.1 Comparison of the hip joint force: Heimkes/AnyBody


First of all the computed hip joint force Rh according to Heimkes Stemmkrpermodel is
compared with the output force vector of the hip joint getting from ANYBODY. The
following figure shows the comparison of the magnitudes of Rh:

Comparison: calculated hip joint force R (Stemmkrpermodel)


and output hip joint force AnyBody
4000

hip contact force [N]

3500
3000
R: ANYBODY

2500

R: Stemmkrpermodel
2000
1500
1000
500
0
0%

20%

40%

60%

80%

100%

gait cycle [%]

Figure 6.24: Comparison of the hip joint force Heimkes/AnyBody

The two above presented curves of figure 6.24 are showing in shape a quite good agreement.
However, the magnitude of the computed hip joint force according to the
Stemmkrpermodel is much lower until approximately 70% of the gait cycle, especially
between 13% and 20% of the gait cycle. Thus, a muscle or several muscles are assumed to be
missing in the calculation of the hip joint force of the Stemmkrpermodel.

6.5.2 Correction of the hip joint force


To correct the hip joint force of the Stemmkrpermodel the missing muscle or muscles have
to be found. In a dissertation [29] found, the muscles that are involved in the calculation of the
hip joint force in the one-legged stance were investigated and also the percentage of each
single muscle force involved was given according to the sum of all muscle forces.

- 50 -

6 Analysis using ANYBODY


Therefore, the muscle activity of the muscles ilipsoas and rectus femoris are checked in
AnyBody:
Activity patter of rectus femoris and ilipsoas
1,00E+00
8,00E-01
ilipsoas

activity

6,00E-01

rectus femoris

4,00E-01
2,00E-01
0,00E+00
0%

20%

40%

60%

80%

100%

-2,00E-01
gait [%]

Figure 6.25: Activity patter of ilipsoas and rectus femoris

The activation patter of the muscles ilipsaos and rectus femoris is looking like it could be the
missing part of the difference in the calculation of the hip joint force. Thus, the muscles
ilipsoas and rectus femoris are also included into the calculation of the hip joint force and the
result is presented as follows:

Comparison: calculated hip joint force R


(Stemmkrpermodel modified)
and output hip joint force AnyBody
4000
3500

hip contact force [N]

R: ANYBODY
3000
R: Stemmkrpermodel
2500
R: Stemmkrpermodel +
ilipsoas + rectus femoris

2000
1500
1000
500
0
0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

gait cycle [%]

Figure 6.26: Comparison of the corrected hip joint force Heimkes/AnyBody

- 51 -

6 Analysis using ANYBODY


As it can be observed from the above figure 6.26, the output hip joint force from AnyBody
and the calculated hip joint force of the Stemmkrpermodel plus the muscles ilipsoas and
rectus femoris are now fitting quite well. This in turn means that a small correction of the
Stemmkrpermodel has to be done and also a new nomenclature of it will be introduced in
the next section.

6.5.2.1 New nomenclature: Stemmkrpermodel


The following new nomenclature is introduced for the Stemmkrpermodel and will be used
in the proceeding of this project:

wh
Fmh

Fmh: all hip equilibrates


gluteus minimus
gluteus medius
gluteus maximus
tensor fascia latae
rectus femoris
ilipsoas
wh: partial body weight
Rh: hip joint force
sum of Fmh and wh

Fmt
Rh

Fmk
RT

Fmt: all hip equilibrates acting on the trochanter


gluteus minimus
gluteus medius
gluteus maximus
tensor fascia latae
Fmk: knee muscles + additional muscles
gluteus maximus
tensor fascia latae
vastus lateralis
vastus medialis
vastus intermedius
RT: trochanter resultant force
sum of Fmt and Fmk

Figure 6.27: New nomenclature Stemmkrpermodel

As it can be seen from the above described new nomenclature, the first subscript index of the
muscle forces F is m standing for muscles. The second subscript index indicates either
which muscles are acting or the location they are acting.

- 52 -

6 Analysis using ANYBODY

6.6 Verification of the Stemmkrpermodel


Resulting from the done investigations and computations described in chapter 3 a table of
resultant force vectors acting on the femur in the one-legged stance (presenting the loading
response phase) with the new nomenclature is presented:

Figure 6.28: Computed force vectors dependent on the body weight

The above presented table of the resultant force vector values should be verified by the
computed muscle force vector values received from the AnyBody modelling system.
The following table shows the expected resultant force vector values according to the human
body model with a body weight of 64.9 kg, which is used in the Gait3D model of
AnyBody:

Figure 6.29: Expected force values according to the body weight of 64.9 kg

Due to the correction of the Stemmkrpermodel the expected force magnitude value of Rh
is expected to be higher and in turn also the force magnitude value of RD. The angles of these
muscle forces may also change. All other muscle force magnitudes should be approximately
the same as figure 6.29 shows.

- 53 -

6 Analysis using ANYBODY

6.6.1 Verification of the force vectors


The following figure presents a summarized table of the muscle force vectors Fmt, Rh, Fmk, RT
and RD in kg:

1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
11%
12%

Fmt = 150.57

Rh = 201.19

Fmk = 209.63

RT = 111.63

RD = 262.85

[kg]
36,79
57,92
85,15
123,13
155,32
147,61
134,94
118,48
134,03
128,42
118,58
105,3

[kg]
90,3
125,03
155,17
198,96
238,97
241,23
233,94
225,83
247,87
247,06
239,19
228,82

[kg]
70,43
97,02
128,79
169,39
194,5
210,27
217,33
216,89
215,61
208,85
201,4
194,09

[kg]
44,64
57,77
74,32
96,15
110,42
121,83
132,04
138,58
133,25
128,50
124,96
122,97

[kg]
121,42
161,57
197,3
244,64
278,1
303,3
315,57
322,97
330,41
328,07
322,28
317,56

Figure 6.30: Loading response: Computed muscle force vector values in kg

The calculated resultant muscle forces of the Stemmkrpermodel showed good agreement
with the expected values. However, the values of Rh and RD are higher than expected due to
the additional muscles ilipsoas and rectus femoris.

6.6.2 Verification of the angles


The next figure 6.31 illustrates the global AnyBody coordinate system and the hip joint
resultant force vector Rh :
z

d
y

Inclination to the
perpendicular:
x

Figure 6.31: Global coordinate system of AnyBody and resultant force vector Rh

- 54 -

6 Analysis using ANYBODY


To verify the inclination to the perpendicular the resultant force vectors have to be converted
into an angle representation. The calculation will be illustrated for the resultant force vector R
at 9% of the gait cycle (loading response):

X ( R)
R = Y ( R)
Z ( R)

d=

X ( R) 2 + Y ( R) 2

= arctan

d
Z ( R)

R = (223.64, -499, -2369.28)T

d = 223.64 2 + (499) 2 = 546.83

= arctan

546.83
= 13
(2369.28)

(6.23)

(6.24)

The inclination to the perpendicular, = 13, of the resultant force vector R calculated from
the computed values from AnyBody is a little bit lower than the expected value of 17. Thus,
it can be said that the inclination to the perpendicular of the resultant force vector R at 9% of
the gait cycle is verified, but with a small deviation.
The next figure 6.32 shows the summarized table of all calculated inclination angles of Fmt,
Rh, Fmk, RT and RD during the loading response phase, which means from 0%-12% of the gait
cycle:

1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
11%
12%

Fmt = 23

Rh = 17

Fmk = 8

RT = 52

RD = 6,4

12,76
18,12
22,81
25,08
25,78
24,47
24,14
23,24
26,37
25,4
24,56
23,69

5,15
8,19
12,22
15,14
16,29
14,36
13,16
11,23
13,0
11,9
10,68
9,17

25,81
28,56
28,53
28,67
28,93
28,66
28,25
27,81
27,15
26,35
25,85
24,83

51,93
58,52
62,2
67,3
74,03
65,69
58,85
53,22
57,42
56,39
54,03
49,59

14,67
17,69
19,21
20,49
20,93
20,64
20,23
19,48
18,53
17,49
16,85
15,76

Figure 6.32: Loading response: Angles to the perpendicular of the computed force vectors
The angles of the computed force vectors are in general too high, especially the angles to the
perpendicular of Fmk and RD.

- 55 -

6 Analysis using ANYBODY


Due to the fact that the Stemmkrpermodel is actually a 2D model, the planar angle to the
perpendicular in the frontal plane is calculated and summarized again in a table:

Fmt = 23 Rh = 17
1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
11%
12%

12,06
17,76
21,8
23,69
24,32
23,31
23,29
22,67
25,35
24,47
23,77
23,02

4,87
7,67
11,19
13,74
14,79
13,09
12,14
10,43
11,89
10,96
9,89
8,56

Fmk = 8
3,8
5,43
5,48
5,67
6,0
6,14
6,26
6,67
7,15
7,57
7,62
7,78

RT = 52 RD = 6,4
23,36
40,45
50,63
59,78
69,47
57,18
47,3
39,29
48,44
47,51
44,36
39,21

2,06
3,37
3,76
4,18
4,52
4,72
4,82
5,09
5,48
5,63
5,7
5,8

Figure 6.33: Loading response: planar angle to the perpendicular in the frontal plane
As it can be seen from the above figure 6.33 the planar angle of the frontal plane are now in
well agreement with the inclination angles to the perpendicular of the 2D
Stemmkrpermodel and therefore also verified like the magnitudes of the force vectors.

6.6.3 Visualization of the force parallelogram at the greater trochanter


During the loading response of the gait cycle, the greater trochanter apophysis should be
subject to pressure stress from a craniolateral direction. The visualization of the trochanter
force resultant acting on the greater trochanter apophysis ensured the craniolateral force
direction during the whole loading response (0%-12%) of the gait cycle.
The next figure 6.34 shows the visualized force parallelogram acting on the greater trochanter
apophysis in the frontal view and the lateral view at 9% of the gait cycle to present the
craniolateral force direction:

Figure 6.34: Trochanter resultant in the frontal and lateral view at 9%


- 56 -

6 Analysis using ANYBODY


Considering the complete gait cycle the trochanter resultant is highest in the loading response
like expected.

6.7 Coordinate Transformation


The calculated muscle force vectors are all given in the global coordinate system of AnyBody.
This global coordinate system is fixed to the ground where the human body model is
walking on. For the force application in ANSYS the force vectors have to be transformed
from the global coordinate system to the local coordinate system of the femur. The local
coordinate system of the femur is located at the centre of gravity of the bone:

y
x
Figure 6.35: Global and local (technical coordinate system) coordinate system of the femur
For the transformation of the force vector from the global coordinate system of AnyBody to
the local coordinate system of the femur a rotation matrix is necessary. AnyBody has for
every segment an output option Axes which is the rotation matrix. The rotation matrix of
the femur describes the orientation of the local coordinate system L with respect to the
global coordinate system G, it can also be calculated when the location of the unit vectors
of system L are expressed in system G. The three unit vectors are the columns of the
rotation matrix:

e xx e yx e zx

Axes = e x e y e z = e xy e yy e zy
e xz e yz e zz
Therefore, a muscle force vector F defined in system G can be rotated into system L
according to the following formula:

FL = ( Axes ) FG
1

(6.25)

By using this formula all muscle force vectors are now available also in the local coordinate
system of the femur.
- 57 -

7 Generation of the FE model of the femur

7 Generation of the FE model of the femur


7.1 Simpleware software
Simpleware provides three software options for the conversion of 3D images to CAD, Rapid
Prototyped and Finite Element models. A core image processing platform called ScanIP
forms the basis of the software including modules for CAD integration and mesh generation.
The next figure 7.1 illustrates the relationship between the different simpleware products:

Figure 7.1: Simpleware software products


In this project ScanIP and ScanFE are used to construct the FE model of the femur from
the CT-data of the femur.

7.1.1 ScanIP: image processing software


A left femur bone was taken and scanned by a CT scanner. In a CT scanning process the
amount of X-rays absorbed by the femur bone is measured and at each sampling point within
the volume these values are converted into greyscale pixel values. The unit of these pixel
values is called Hounsfield unit (HU).The femur bone is scanned by 661 slices and the
images are saved in a format called DICOM(Digital Imaging and Communications in
Medicine) which is the standardized format for exchanging digital pictures in medicine.
The 661 DICOM images are imported in ScanIP and the different slices can be visualized
by their greyscale pixel values in different views. The following figure 7.2 shows one part of
the slice of the femur in the x-z plane:

Figure 7.2: Slice of the femur in the x-z plane


- 58 -

7 Generation of the FE model of the femur


First a segmentation has to be done to distinguish between the compact bone region and the
spongiosa bone region of the femur.
The segmentation is performed by an operation called FloodFill which creates a so-called
mask1 consisting of the pixels of greyscale values chosen. The range for the greyscale value
was chosen from 1 (to preserve the marrow) to the highest greyscale value of the femur.
Therefore, the femur without the marrow is defined as one mask. The problem of the
FloodFill operation is that there were still remaining small holes like presented in the
following figure 7.3:

Figure 7.3: cavities in mask1


To get rid of these small holes of the mask1 while keeping the marrow as a cavity of the
femur again the Floodfill operation as cavity fill will be applied. Due to the fact that all
pixels which are not belonging to the exterior of the object will be filled by this operation the
marrow is connected with the outside of the bone by a padding operation.
Then another mask2 for the interior of the bone (marrow) is created by unpadding the
marrow and another FloodFill operation on mask1. In order to connect the marrow
structures in Mask 2, a morphological close is applied.
The result is presented in the next figure 7.4 containing mask1 and mask2:

Figure 7.4: The marrow of the femur of one slice


To separate the compact bone region and the spongiosa bone region of the femur another
mask for the spongiosa region has to be defined. This mask will then be subtracted from
mask1 and the result will be a new mask containing the compact bone region. A FloodFill
operation with a different range of greyscale pixel values is done to receive one mask with the
spongiosa region of the femur. The operation was not performing well and manual corrections
had to be done in every slice. As a result the outer shell of the femur (compact bone) was
thicker than the original outer femur shell not to get problems in meshing the compact bone
region.

- 59 -

7 Generation of the FE model of the femur


The received three different masks are shown in the following figure 7.5:

Figure 7.5: Three defined masks of the femur


Having a closer look to the surface of the rendered femur model it can be observed that the
surface has still some artefacts. To smooth the data a Recursive Gaussian Filter is applied and
the comparison before and after applying the filter is presented:

Figure 7.6: Comparison of the femur before and after applying the Recursive Gaussian Filter
The femur with the manually corrected masks was exported as STL file to ANSYS ICEM and
meshed with tetra-and hexahedral elements.
The original femur shape containing one mask including the marrow was exported to
ScanFEas SFH files.

- 60 -

7 Generation of the FE model of the femur

7.1.2 ScanFE: mesh generation / material assignment module


ScanFE offers a robust approach to the conversion of segmented 3D image data into multipart volumetric and/or surface meshes. The generated meshes can be imported into a range of
commercial FE programs such as ANSYS.
The model was resampled in ScanIP not to get too many elements. The exported material
model is visualized as voxel model in ScanFE as shown:

Figure 7.7: Voxel model of the proximal femur


With the help of the CT images an inhomogeneous material distribution of the femur can be
considered.
According to the Beer-Lambert law the attenuation of the x-ray intensity by penetrating the
human bone is proportional to the density of the bone. In turn, the density of the bone can be
set in relation to the values of absorption (in Hounsfield units [HU]) of the CT images and
again the E-modulus of the bone is a function of the density of the bone.
Therefore, for every element or group of elements an own value of the density and E-modulus
can be assigned.
The program ScanFE also has the possibility to assign material properties throughout the
femur based on the Hounsfield number. It can be set of how many different material types the
femur model shall consist (upper limit 256 different material types). For the femur model 100
different material types were chosen.
The mass density, the youngs modulus and the poissons ration were calculated according to
the following formula:

mass density : = a + b GS

a = 1000 and b = 0.385

= 1000 + 0.385 GS
GS: greyscale value

young ' s mod ulus : E = a + b c + d e


E = 0.4

a = 0, b = 0.4, c = 3 and d = e = 0

poisson' s ratio : = a + b c + d e

a = 0.4, b = 0, c = 0 and d = e = 0

= 0 .4
The coefficient values of a, b, c, d and e were taken from the ScanFE documentation [30].
- 61 -

7 Generation of the FE model of the femur

7.1.2.1 Correction of the material properties


After the first FEM analysis with 100 different material types (loading according Heimkes) at
9% of the gait cycle was performed, a look at the displacements of the femur revealed that
there is something wrong with the femur model (maximum displacement: 0.675411 m) as
illustrated:

DMX = 0.675411 m

Figure 7.8: Displacement of the femur at 9% of the gait cycle (loading according Heimkes)
Due to the large maximum displacement, the material properties (density and E-modulus)
have been regarded more in detail and the E-modulus values [Mpa] are plotted versus the
density values [kg/m3] as follows:
E-Modulus versus density
160
140

E-Modulus [MPa]

120

ten different materials

100

hundred different
materials

80
60
40
20
0
0

500

1000

1500

2000

density [kg/cm3]

Figure7.9: assigned material properties from ScanFE; E-modulus versus density


The calculated E-modulus values from ScanFE seemed to be very low. Therefore, they had
to be compared with E-modulus values of the femur recorded in the literature.

- 62 -

7 Generation of the FE model of the femur


However, in the literature there exists no common conversion from the original CT data to the
values of E-modulus utilized for the FEM model as the next figure 7.10 illustrates:

Figure 7.10: Comparison of the conversion formula [31]


Curve A: valid for the complete bone
Curve B: spongiosa part of the bone
Curve C: cortical part of the bone
Curve D: two parts, spongiosa and cortical bone
The comparison of the E-modulus values [Mpa] computed from ScanFE and the E-modulus
values from the literature confirmed that the E-modulus values received from ScanFE are
too low. Therefore, a correction has to be done.
A European project about Virtual Animation of the Kinematics of the Human for Industrial,
Educational and Research Purposes (VAKHUM) was found [32], which provides also
datasets of the femur.

- 63 -

7 Generation of the FE model of the femur


The E-modulus [Mpa] versus density [kg/m3] of the VAKHUM femur model is plotted:

E-modulus versus density


45000
40000
E-modulus [Mpa]]

35000

VAKHUM: femur

30000
Polynomisch
(VAKHUM: femur)

25000
20000

y = 0,00000377545225x3 0,00001140543178x2 +
0,01228931065013x - 4,28249233909230

15000
10000
5000
0
0

500

1000

1500

2000

2500

density[kg/m3]

Figure 7.11: E-modulus versus density (VAKHUM: femur)


To correct the E-modulus values a program was written in C++.
The lowest and the highest density values of the femur of the VAKHUM data were taken and
also the highest and lowest grey scale values (GS) from the available output data from
ScanFE. Then, these values were set in the following linear equations to recalculate the
coefficient values a and b:
= a + b GS
(7.1)

566 = a + b (692)
2174 = a + b (1292)

a = 1126.85 and b = 0.81


According to the recalculated coefficients a and b the density values of the femur were
recalculated. In the next step, a polynomial E-modulus versus density curve fitting (see figure
7.11) was done and with the received coefficient values the new E-modulus was calculated:
E [ MPa ] = 0.00000377545225 3 0.00001140543178 2 + 0.01228931065013 4.2825

(7.2)

The assignment of the material properties of the femur according to this described procedure
is expected to present now the separation between the cortical and spongiosa bone quite well.

- 64 -

7 Generation of the FE model of the femur


To ensure that the separation of the compact and spongiosa bone of the femur is correct a
transversal cut in approximately the middle of the femur was done and the elements with the
lowest E-modulus (representing spongiosa bone) and highest E-modul (representing cortical
bone) were plotted as follows:

Figure 7.12: blue elements (cortical bone); red elements (spongiosa bone)
The maximum displacement of the FEM femur model with 100 different material types
(loading according Heimkes) at 9% of the gait cycle was also checked again:

DMX = 0.018 m

Figure 7.13: Displacement of the femur at 9% of the gait cycle (loading according Heimkes)

The FEM model shows realistic results and therefore the stress analysis in Ansys can be
performed now.

- 65 -

8 Stress analysis using ANSYS

8 Stress analysis using ANSYS


8.1 Applied loads
In the stress analysis performed with Ansys, the Pauwels model is compared with the
Stemmkrpermodel at 9% of the gait cycle. This step of the loading response is chosen due
to the fact that the trochanter resultant has its highest magnitude there.
Therefore, the loads, which are applied for the Stemmkrpermodelare the following:

abductor muscles Fmt


hip joint force Rh
trochanter resultant RT

The applied loads of the Pauwels model differ in the absence of the vastus
lateralis/intermedius/medialis muscle forces:

abductor muscles Fmt


hip joint force Rh
tractus forces (gluteus maximus and tensor fascia latae)

8.1.1 Hip joint force


Comparing the output of the hip joint force given by AnyBody with the calculated hip joint
force according to Heimkes, a difference in the direction can be observed. Considering, for
example, the planar angle to the perpendicular in the frontal plane, the planar angle of the
output hip joint force is approximately 30 and the calculated planar angle according to
Heimkes is just 13. Thus, theses values have to be compared with data from the literature.
Bergmann [24] investigated nine different activities (slow/normal/fast walking, up/down
stairs, standing up, sitting down, standing on 2-1-2 legs and knee bend) which are assumed to
cause high hip joint loads. The angle Ay to the perpendicular in the frontal plane of the hip
contact force was also investigated and illustrated:
z

Figure 8.1: Hip joint force in the frontal plane [24]


For all the investigated activities the hip joint force directions in the frontal plane are very
similar and their variation is remarkable small as it can be seen from the above figure 8.1.
Small forces act more from medial than large ones.
- 66 -

8 Stress analysis using ANSYS


The found peak force is in a small range of 12-16 for all activities except standing on one
leg, where the peak force has an angle of 7 to the perpendicular.
The other question which is arising is, if the very high force values of the output hip joint
force especially between 15% and 19% of the gait cycle are realistic. In another study [33] the
hip reaction forces calculated in AnyBody using the gait analysis data from Vaughan were
compared with data from literature. It was concluded that the calculated hip joint values from
AnyBody, except the large lateral force at HS, are possible due to individual variation
between patients.
Due to the fact that at 9% of the gait cycle the left leg already quitted the ground (toe-off) the
force values from AnyBody can be taken. Thus, the calculated force values according to
Heimkes are applied.

8.1.1.1 Determination of the hip joint load application point


The joint reaction force R at the hip should hit exactly the femur head centre. Therefore, the
load application point at a surface node, which is fulfilling this condition, has to be found.
Due to the fact that the femur head is not exactly spherical, an approximation has to be done
for every time step.
From the middle plane of the femur head approximately the centre node was chosen and the
opposite direction of the load vector is applied to have an outer node of the femur. Then, from
the centre node and the outer node of the femur a link element was created as shown in the
next figure:

Figure 8.2: Link elements for determination of load application point


Considering the above presented figure 8.2, the load application point can now be determined
easily. Therefore, the hip contact force R is applied to the node, which is nearest to the point,
where the link element is penetrating the bone surface.

8.2 Constraints
In the FE model the femur is locked distally, which is the preferable standard in the literature.

8.3 Postprocessing in Ansys


In order to calculate the normal and shear stresses on defined cutting planes some
postprocessing steps in Ansys have to be performed and are described in the next sections.

- 67 -

8 Stress analysis using ANSYS

8.3.1 Stress tensor


Stress is a second-order tensor with nine components, but can be fully described with six
components due to symmetry for elastic isotropic materials.
From ANSYS the 2nd order stress tensor can be obtained:

xx xy xz

[ ] =

yx yy yz
zx zy zz

8.3.2 Stress vector


In order to calculate the effective stresses a stress vector t has to be generated. The stress
vector belongs to the stress tensor and a plane of arbitrary orientation (normal vector of the
plane) as desired for example for a cut through the femoral head.
The following figure 8.3 illustrates an arbitrary cutting plane in 2D and the stress vector:

Figure 8.3: illustration of the stress vector defined on a cutting plane


Unit vector normal to the plane:
n = ni g i

n = g ij ni n j

(8.1)

Stress vector t:
(8.2)

t = n
i

ij

ij

ij

t g i = g i g j n j g = n j g i g j g = n j g i

j
j

t i = ij n j
ti are the coordinates of the stress tensor t in the covariant basis.
Since the base vectors in ANSYS (Cartesian coordinate system) form a set of orthogonal and
normal vectors the coefficients of the stress tensor related to them are physical quantities.
Therefore, no further calculations are required.
However, the components of a tensor depend on the chosen coordinate frame, but tensors
themselves do not depend on a chosen coordinate frame.
- 68 -

8 Stress analysis using ANSYS

8.3.3 Calculation of normal and shear stresses of a defined cutting plane


For the investigation of the femoral head and neck, normal and shear stresses are of
significant interest. For this aim the stress vector t has to be divided into vectors of the
corresponding directions as the next figure 8.4 shows:

t shear
tnormal
t

Figure 8.4: shear/normal part of the stress vector


Value of the normal stress:

=t n = n n =
= n i g i ij g i g j n j g j = n i ij n j g i g i g j g j = ni ij n j

(8.3)

Value of the shear stress:

= p n=
= p i g i ij g i g j n j g j = p i ij n j g i g i g j g j = p i ij n j
with: p unit vector in direction of the investigated shear stress

(8.4)

or:

= t2 2

(8.5)

The value of normal and shear stresses depends on the investigated cutting plane. Thus, for
the different cuts through the femur calculations for the normal and shear stresses in ANSYS
have to be done.

8.4 Analytical solution


The analytical solution is developed by the help of GiD personal postprocessing library in
C++. This part of the project is achieved independent of Ansys pre- and postprocessing data.

8.4.1 Bending stress of a cutting section


The analytical solution of the bending stress is calculated for an arbitrary cutting section
through the femur. Before applying the analytical bending stress formula the second moment
of inertia and the total area of the cross sections are calculated numerically. The components
of the external forces being perpendicular and parallel to the cutting sections are assumed to
be uniformly distributed. Thus, the procedure is not completely analytical.

- 69 -

8 Stress analysis using ANSYS


The classic formula for determining the bending stress in a member is [34]:

b =

M c
Im

(8.6)

It is simplified for a beam of any cross section to:

b : bending stress
M: moment at the neutral axis
c: perpendicular distance to the neutral axis
I m : second moment of inertia with respect to the applied moment axis
The second moment of inertia can be calculated numerically as:

I m = c 2 ( x, y ) dA
A

# nodes

2
i

( x, y ) A

(8.7)

In order to find the discretisized areas A, the node cloud of the transversal cut at z=0.2m
through the body of the femur is taken. This point cloud is triangulated with the delaunay
triangulation algorithm. The elemental areas after triangulation are taken as the discrete
areas, and used to calculate the second moment of inertia and the total area of the complex
cross section.
Delaunay triangulation (DT) for a set P of points in the plane is a triangulation DT(P) such
that no point in P is inside the circumcircle of any triangle in DT(P). Delaunay triangulations
maximize the minimum angle of all the angles of the triangles in the triangulation; they tend
to avoid "sliver" triangles [wikipedia].
The illustration of the delaunay triangulation in a plane is as follows:

Figure 8.5: scheme of delaunay triangulation in a plane


As it can be observed from the above figure 8.5,none of the points in the group enters the
circumcircles of the candidate elements.

- 70 -

8 Stress analysis using ANSYS


Summarized, a point cloud of a transversal cut through the body of the femur, the
triangulation of this point cloud and the calculated total area is presented:

Figure 8.6: point cloud (left); triangulated point cloud (middle); total area (right)
The bending moment is calculated by considering the moment effect of the forces applied in
the unconstrained region of the femoral head. The procedure is accomplished for both
biomechanical models: the Stemmkrpermodel and the Pauwels model. The resultant
moment of the forces in the unconstrained part of femur to the centre of the cross section
determines the bending direction.
Due to the fact that the bending moment is zero along this axis, the neutral axis is chosen as
the moment direction. Therefore, the c values and second moment of inertia are calculated
with respect to this line of action.
Beside of that, only the planar components of the applied moment are decisive on this
direction. The component of the moment being parallel to the cross section normal drives the
torsion effect causing the shear stress but not the bending stress.
The next figure 8.7 illustrates the above described procedure:

Figure 8.7: illustration of the procedure of the calculation of the bending moment

- 71 -

8 Stress analysis using ANSYS


The same procedure is also performed with a hollow circle cross section with an
approximated outer and inner diameter as follows:

Figure 8.8: approximated inner and outer diameter of the cross section
To proof that the results visualized in GID are trustable, the numerically calculated second
moment of inertia is compared with the analytical moment of inertia for the hollow circle:
analytical:
I=

( D 4 d 4 ) = 3.27491e 8mm 4
64

(8.8)

numerical:

I m = c 2 ( x, y ) dA
A

# nodes

( x, y ) A = 3.15195e 8mm 4

(8.9)

From the analytical and numerical calculated second moment of inertia it can be said that the
results are quite confidential.

8.4.2 Shear stress of a cutting section


As mentioned also in the previous section the solution is apart from the calculation of the
polar moment of inertia and the calculation of the cutting section area is analytical. The
calculated values will be compared with the postprocessed FEM results.

total = shear + torsion

F T r
= s +
A
J

J = r dA re2 Ae
2

shear : shear stress due to the pure shear force


torsion : shear stress due to the torsion effect
T: torsion with respect to the centre of gravity of the cross section
r: distance to the centre of gravity point
Fs : shear force parallel to the cross section
J: polar moment of inertia w.r.t. the centre of gravity of the cross section

- 72 -

(8.10)

8 Stress analysis using ANSYS

8.4.3 Visualization of the results in GiD


8.4.3.1 Total normal stress
The total normal stress of the transversal section at z = 0.2m is calculated as the sum of the
normal stress and the bending stress:

total = normal + bending =

F M F M c
+
= +
A W
A
Im

(8.11)

For the Stemmkrpermodel and the Pauwels model with the hollow circle cross section the
following results of the total normal stress of the transversal section through the body of the
femur are expected:
total normal stress [Pa]

total normal stress [Pa]

Figure 8.9: total normal stress [Pa] in the hollow circle cross section: Pauwels(left);
Heimkes(right)
The total normal stress is also shown for the real complex cross section of the femur:
total normal stress [Pa]

total normal stress [Pa]

Figure 8.10: total normal stress in the real complex cross section: Pauwels(left);
Heimkes(right)

- 73 -

8 Stress analysis using ANSYS


The next figure 8.11 illustrates a comparison of the total stress values [MPa] over the full
cross-section at z = 0.2m versus the number of defined colours of the Pauwels model and the
Stemmkrpermodel:
Compression/tension of the full cross-section
at z=0.2m

total stress [MPa]

80,00
60,00

Pauwels

40,00

"Stemmkrpermodel"

20,00
0,00
-20,00

10

-40,00
-60,00
-80,00

number of colours of the cross-section

Figure 8.11: comparison compression/tension values over the cross-section


As it can be obviously seen from the above figure the Pauwels model has higher tension
values, but lower compression values than the Stemmkrpermodel.

8.4.3.2 Total shear stress


In the following the expected total shear stress and the two components of it of the transversal
cutting section at z = 0.2m of the homogeneous Stemmkrpermodel is visualized according
to the analytical solution:

Figure 8.12: Heimkes: total shear stress (left); tau_yz (middle); tau_xz (right); unit [Pa]
:
- 74 -

8 Stress analysis using ANSYS


For the same transversal cut the expected total shear stress and the two components of it are
presented for the homogeneous Pauwels model:

Figure 8.13: Pauwels: total shear stress (left); tau_yz (middle); tau_xz (right); unit [Pa]

8.5 Equilibrium check


The static equilibrium is simply checked by regarding the reaction forces given by Ansys.
For the Stemmkrpermodel and the Pauwels model the following force component values
are expected to be verified by the reaction forces:
Heimkes:

Fx = 52.69 N

Fy = 413.3 N

Fz = -1118.14 N

Reaction forces (Heimkes): TOTAL VALUES


VALUE -52.69 -413-3 1118.1

Pauwels:

Fx = 31.7 N

Fy = 383.0 N

Fz = -1061.19 N

Reaction Forces (Pauwels): TOTAL VALUES


VALUE -31.700 -383.00 1061.2

static equilibrium of the femur model

8.5.1 Internal equilibrium


The internal equilibrium of the femur model is also checked at two different cross-sections, at
a transversal cross section through the body of the femur and at a cross section through the
femoral head.

- 75 -

8 Stress analysis using ANSYS


First, the internal equilibrium of a transversal cross section was checked in Ansys with the
following surface commands:
sucr,cut1,cplane,3
sumap,sz,s,z
sueval,f1,sz,intg

// create cutting plane at the location of the working plane


// mapping of z-direction stress to the surface and name it SZ
//get axial force by integrating axial stress and name it f1

The calculated axial force values for different transversal cross sections, for example of the
Stemmkrpermodel are:
z = 0.2 m:
z = 0.25 m:
z = 0.35 m:

Fz = -3239.27 N
Fz = -3229.74 N
Fz = -3227.94 N

This calculated internal axial force value of the transversal cross sections should be equal to
the above applied external force components in the axial direction. The axial summed external
force components have a value of -3225.14 N. Thus, the internal equilibrium is fulfilled for
the axial force with a small error..
It can be stated that the physical stresses normal and parallel to some critical planes found by
Ansys are fitting together with the analytical solutions.
To verify statical equilibrium of the external forces with the internal stresses, some additional
operations are performed in C++.
In this way, the accuracy of the solutions in terms of the stresses was ensured quantitatively.
The verification is done by comparing the analytical bending moment exerted by the external
muscle forces with the internal stress traction vector moments. For this accomplishment some
physical area values were necessary.
For this purpose, the nodal normal stress values obtained from the previous postprocessing
works are interpolated to obtain the elemental stress values:

1
1
1
1
1
1
+
+
= 1 + 2
+3
d1 d 2 d 3
d1
d2
d3

(8.12)

1
i Ai
d1

d3

3
d2

Figure 8.14: illustration of the interpolation normal stress values

- 76 -

8 Stress analysis using ANSYS


Where, the 1, 2,3 values are standing for the nodal total normal stress resultants. The force
equilibrium of the internal-external balance is verified similarly.
According to the following formulas the errors can be calculated:

r r
Bending _ error r Fextern
r
Force _ error Fextern

onplane

normalplane

i ci Ai
i

(8.13)

i Ai
i

This procedure is applied to different cross sections with different models studied for the
project. The algorithm is implemented in femur_cuts.dsw.
The following figure 8.15 presents the verified internal equilibrium for a cut through the
femoral head and a transversal cut through the body of the femur visualized in GiD:
bending-analytical/ansys:3.82068/3.99402[Nm]
normalforce-analytical/ansys:-1910.42/-1908.76[N]

bending-analytical/ansys: 144.036/141.739 [Nm]


normalforce-analytical/ansys: -3225.14 /-3244.82 [N]

Figure 8.15: internal equilibrium; (top) cut through the femoral head; (bottom) cut through
the body of the femur

- 77 -

8 Stress analysis using ANSYS

8.6 Results: Comparison Pauwels model / Stemmkrpermodel


The results presented in this chapter are all corresponding to the loading response at 9% of the
gait cycle.

8.6.1. Transversal cut: FEM and analytical solution


For the purpose of comparison, the transversal cut is done like in the analytical solution
through the body of the femur at z=0.2m.
First, the FEM results of the stress in z-direction of the Pauwels model and the
Stemmkrpermodel are presented in the view from above:

Figure 8.16: stress in z-direction [MPa] of the Pauwels model

Figure 8.17: stress in z-direction [MPa] of the Stemmkrpermodel


Comparing the results of the analytical solution with the FEM results, it can be also observed
that the compression values of the Stemmkrpermodel are higher and the tension values are
lower compared with the Pauwels model. However, the FEM stress values of both models are
much higher than those of the analytical solution.
The explanation of these higher values could be that the outer layer elements of both FEM
models are not carrying the highest values like expected. It seems like the scanned femur has
an outer porous layer.
To ensure that the higher stress values of the FEM femur models are caused by the
inhomogeneous material distribution, a new analysis is performed with homogenous material
distribution.

- 78 -

8 Stress analysis using ANSYS


Next, the FEM results of the homogenous models according to Pauwels and Heimkes are
compared again with the analytical solution:

Figure 8.18: Pauwels model: view from above


left: FEM results with corresponding colour legend at the bottom
right: analytical solution with corresponding colour legend

Figure 8.19: Stemmkrpermodel: view from above


left: FEM results with corresponding colour legend at the bottom
right: analytical solution with corresponding colour legend
Both FEM models show nearly perfect agreement with the analytical solution of the
transversal cut through the body of the femur at z=0.2m. Thus, the higher stress values of the
FEM models including inhomogeneous material distribution is therefore explained.

- 79 -

8 Stress analysis using ANSYS


Next, the pure moment results by subtracting the axial stress are obtained from Ansys for
the homogenous Pauwels model and the Stemmkrpermodel from the view from above:

Figure 8.20: comparison pure moment result; (left) Stemmkrpermodel; (right) Pauwels
As it can be obviously seen from the above figure 8.20 the pure moment results of the
Stemmkrpermodel are slightly higher.
Inspite the total shear stress is not of special interest for this transversal cut, a comparison
between the FEM solution of the homogenous Pauwels model and the Stemmkrpermodel
is done:

Figure 8.21: Ansys total shear force; (left) Pauwels; (right) Heimkes
The maximum of the total shear force in the Stemmkrpermodel is slightly higher than in
the Pauwels model, but considering the minimum value of the total shear force it is vice
versa.

- 80 -

8 Stress analysis using ANSYS

8.6.2. Cut through the femoral head/neck


In this section two cuts will be done through the proximal femur. One through the head and
the other one through the neck as illustrated:
cutting plane through the femoral neck
cutting plane through the femoral head

Figure 8.22: defined cutting planes of the proximal femur


Due to the fact that the stresses existing in the femoral head are only influenced by the hip
joint force, just the FEM results of the homogenous Stemmkrpermodel are presented:

Figure 8.23: Heimkes: normal and shear stresses approximately in the middle of the head
[Pa]
The distribution of the normal stresses look like expected, just compression. However, the
normal stresses of a cut approximately through the femoral head can not be compared with the
analytical solution because it is based on the assumption that the hip joint force is distributed
uniquely. The same is valid for the analytical solution of the shear stress, but the FEM result
of the shear force is like it is expected from the literature, with the largest shear value in the
middle decreasing with the increasing radius of the spherical head. Here, it should be
mentioned that the cut is not exactly through the hip joint centre.

- 81 -

8 Stress analysis using ANSYS


Next, a cut through the femoral neck for the inhomogeneous Pauwels model and the
Stemmkrpermodel is done and the normal stress is investigated. The following figure 8.24
presents the normal stress in comparison:

Figure 8.24: Normal stresses [Pa] of the cut through the femoral neck;
(right) Pauwels; (left) Heimkes
It can be concluded from the above figure 8.24 that the normal stresses of a cutting section
through the femoral neck of the Pauwels and the Stemmkrpermodel are approximately the
same.

- 82 -

9 Summary and discussion

9 Summary and discussion


The determined magnitudes of the muscle force vectors and their inclination to the
perpendicular corresponding to the Stemmkrpermodel using AnyBody showed good
agreement to the expected values based on anatomical, radiological and computational results.
However, the Stemmkrpermodel was slightly modified by the addition of the two muscles
ilipsoas and rectus femoris. Actually, if these additional two muscles have to be included in
the Stemmkrpermodel is dependent on the antetorsion of the femur and therefore
dependent on the individual patient.
However, it should also be mentioned here, that the calculated hip joint force according to the
vectorial force parallogram of the modified Stemmkrpermodel was differing in angle with
the output hip joint reaction force given by AnyBody. In turn, the output hip joint reaction
force received from AnyBody was also a little bit differing compared to the hip joint contact
force values and angles given in the literature. The computation of the hip joint force in
AnyBody due to inverse dynamics is based on the ground reaction forces input data. The
accuracy of this data plays a crucial role for the computations of the joint forces and moments.
However, the hip joint force is not influencing the muscle force parallelogram at the
apophysis of the greater trochanter.
The length of the generated FE model based on CT images of the femur was with a small
deviation the same as the length of the femur used in AnyBody. However, the shape of the
femur for every individual patient is different.
The application point of the muscle forces acting on the apophysis of the greater trochanter
according to the Stemmkrpermodel is assumed at one point, which is of course
anatomically not correct. Therefore, in the FE model the forces are applied at a single node at
the greater trochanter, which does not play a role for the investigation of a transversal cut
through the body of the femur due to the St. Venant principle.
In the stress analysis performed in Ansys the change in the bending in a transversal cut
through the body of the femur was shown comparing the Pauwels model with the
Stemmkrpermodel. Considering the applied muscle forces of both of these models, the
difference of them is the absence of the vastus lateralis, vastus intermedius and vastus
medialis muscle in the Pauwels model. It was found, that the bending of the
Stemmkrpermodel of a transversal cut through the body of the femur can be said to be
nearly the same (slightly higher through the higher compressive normal forces) as the
Pauwels model. Thus, it can not be concluded that the bending of the body of the femur,
which was the main investigation point of the stress analysis is reduced in the
Stemmkrpermodel compared to the Pauwels model.

- 83 -

10 References

10 References
1. Wikimedia Commons Category: Anatomy,
http://www.commons.wikimedia.org/wiki/Anatomy (accessed august 2007)
2. Jean-Pierre Kassi (2004) Musculoskeletal Loading and Pre-clinical Analysis of Primary
Stability after Cementless Total hip Arthroplasty in vitro, Technische Universitt Berlin,
Dissertation
3. D. Weichert (2006) Zur Mechanik des menschlichen Bewegungsapparates, Institut fr
Allgemeine Mechanik, Scriptum
4. Visual Histology Chapter 6: Bone, http://www.visualhistology.com (accessed august 2007)
5. Rauber / Kopsch (1998) Schenkelhalswinkel [CCD Winkel], Universitt Hamburg,
Scriptum
6. Ask Dr. Chris (1997) Femoral Anteversion & Tibial (Malleolar) Torsion,
http://www.univie.ac.at/cga/faq/torsion.html
7. Web Books Structure of skeletal muscle,
http://www.web-books.com/.../Skeletal_Structure.htm
8. Dominiek Beckers and Joe Deckers Ganganalyse und Gangschulung, Springer-Verlag,
1997
9. Christopher L Vaughan, Brian L Davis, and Jeremy C OConnor Dynamics of Human Gait,
Kiboho Publishers, 1992
10. Jacquelin Perry Ganganalyse, Urban und Fischer, 2003
11. B. Kummer Biomechanik-Form und Funktion des Bewegungsapparates, Deutscher rzte
Verlag, 2005
12. [Bild Pauwels] Michael Krieg (1996) Berechnung von Muskel und Gelenkkrften mit Hilfe
eines dreidimensionalen Computermodels einer menschlichen Hfte im Einbeinstand,
Eberhard-Karls- Universitt zu Tbingen, Dissertation
13. B. Heimkes, P. Posel, and M. Bolkart (1992) The transgluteal approaches to the hip,
Archives of Orthopaedic and trauma Surgery, 111, pp 220-223
14. B. Heimkes, M.D., P. Posel, M. D., W. Plitz, M.D., and V. Jansson, M.D (1993) Forces
Acting on the Juvenile Hip Joint in the One-Legged Stance, Journal of Pediatric Orthopaedics,
13, pp 431-436
15. G.-P. Brggemann Grundlagen der Biomechanik des muskulo-skeletalen systems, Sport
und Buch Strauss, 2005-4.,verb. Auflage
16. Wageningen University (2006) Wolffs law, http://www.ezo.wur.nl (accessed august
2007)
- 84 -

10 References

17. Georg N. Duda Influence of Muscle Forces on the Internal Loads in the Femur during
gait, Shaker Verlag, 1996
18. Grays anatomy The femur, http://www.theodora.com/anatomy (accessed august 2007)
19. M. Mser & W.Hein (1987) Krfte an der Hfte-das Untergurtmodel; Teil1: Kritik am
Pauwels-Modell-der Zweibeinstand, Orthop. Traumatol., 34, pp. 83-92
20. Luca Cristofolini (1997) A critical Analysis of Stress Shielding Evaluation of Hip
Protheses, Critical Reviews in Biomedical Engineering by Begell House, Inc., 25(4&5): 409483
21. Erik Forster (2003) Predicting Muscle Forces in the Human Lower Limb during
Locomotion, Universitt Ulm, Dissertation
22. M. Damsgaard, S.T. Christensen, and J. Rasmussen (2001) An efficient numerical
algorithm for solving the muscle recruitment problem in inverse dynamics simulations,
Institute of Mechanical Engineering,Aalborg University
23. AnyBody Technology Repository, http://www.anybody.aau.dk/repository (accessed Mai
2007)
24. G. Bergmann, G. Deuretzbacher, M. Heller, F. Graichen, A. Rohlmann, J. Strauss, and G.
N. Duda (2001) Hip contact forces and gait patterns from routine activities, Journal of
Biomechanics, 34, pp. 859-871
25.
International
Society of
Biomechanics
http://www.isbweb.org (accessed Mai 2007)

Biomechnical

data

resources,

26. Scott Z. Barnes, Necip Berme Measurment of Kinetic Parameters Technology, Gait
Analysis, Rebecca L. Craik & Carol A. Oatis
27. Michael S. Orendurff, MS; Ava D. Segal, BAS; Glenn K. Klute, PhD; Jocelyn S. Berge,
MS; Eric S. Rohr, MS; Nancy J. Kadel, MD (2004) The effect of walking speed on center of
mass displacement, Journal of Rehabilitation Research & Development, 41 (5a), pp. 829-834
28. Steven A. Gard, Steve C. Miff, and Arthur D. Kuo (2004) Comparison of kinematic and
kinetic methods for computing the vertical motion of the body center of mass during walking,
Human Movement Science, 22, pp 597-610
29. Dirk-Thorsten Dunkelberg (1989) Die Muskel- und Gelenkresultierende am Hftgelenk,
Anatomisches Institut der Albert-Ludwigs-Universitt Freiburg i. Br., Inaugural-Dissertation
30. Simpleware LTD. ScanIP, ScanFE and ScanCAD Tutorial guide, 2007 version 2.1
31. Christoph Rene Burchard (2005)Femorale Umbauprozesse nach totaler Hftendoprothese
und Finite-Element-Analyse 12 Jahre nach Operation, Inaugural-Dissertation, Phillips
Universitt Marburg
32. The VAKHUM project Database, http://www.vakhum.com (accessed august 2007)
- 85 -

10 References

33. (2007) Validation of Models Generated in the AnyBody Modelling Programme


34. Alfred Bge Mechanik und Festigkeitslehre, Vieweg & Sohn Verlagsgesellschaft mbH,
1981

- 86 -

You might also like