Advanced Structured Materials

Volume 14

Series Editors
Andreas Öchsner
Lucas F. M. da Silva
Holm Altenbach

For further volumes:

http://www.springer.com/series/8611
Andreas Öchsner Lucas F. M. da Silva
•

Holm Altenbach
Editors

Analysis and Design

of Biological Materials
and Structures

123
Prof. Dr.-Ing. Andreas Öchsner Prof. Dr.-Ing. Holm Altenbach
Faculty of Mechanical Engineering Institut für Mechanik
Department of Solid Mechanics and Design Fakultät für Maschinenbau
University of Technology Malaysia (UTM) Otto-von-Guericke-Universität Magdeburg
Skudai, Johor Universitätsplatz 2
Malaysia 39106 Magdeburg
e-mail: [email protected] Germany
e-mail: [email protected]
Prof. Dr. Lucas F. M. da Silva
Faculdade de Engenharia da
Departamento de Engenharia Mecânica
Universidade do Porto
Rua Dr. Roberto Frias s/n
4200-465 Porto
Portugal
e-mail: [email protected]

ISSN 1869-8433 e-ISSN 1869-8441

ISBN 978-3-642-22130-9 e-ISBN 978-3-642-22131-6
DOI 10.1007/978-3-642-22131-6
Springer Heidelberg Dordrecht London New York

Library of Congress Control Number: 2011942744

Ó Springer-Verlag Berlin Heidelberg 2012

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broad-
casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this
publication or parts thereof is permitted only under the provisions of the German Copyright Law of
September 9, 1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.

Cover design: WMXDesign GmbH, Heidelberg

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Preface

The area of biological materials and structures is a growing research field of

immense importance. The possibility to predict or even to influence the ‘lifetime’
of parts of the human body or to offer adequate replacements in the case of failure
and injury directly influences our entire well-being. In many cases, this becomes
more and more important at higher age when joints or other parts must be replaced
in order to guarantee an adequate mobility and function of our body. To adopt the
mechanical performance of structural parts of our body or to offer alternatives if
they do no more function properly in order to meet the general biological life
expectancy is a great challenge which requires joint efforts of many academic
disciplines. On the other hand, many materials and structures seen in living
organisms, i.e. human body, animals and plants, are optimized to an extend which
is still difficult to achieve for artificial materials and structures. Thus, learning
from nature is nowadays a new direction to overcome limitations of classical
engineering materials.
This collection of manuscripts is subdivided in four sections. The first part
covers the determination and analysis of the stress and strain state in biological
materials and structures based on numerical, i.e. finite element method, and
experimental procedures. The second part has its focus on properties of the
materials. This is important in order to identify possible applications of new
materials or to provide the input data for simulations and predictions of the
behavior of materials and structures. The third part covers different aggregate
states, i.e. liquid and gaseous. Thus, the theoretical and mathematical modeling is
different from the investigations of solids as presented in the previous sections.
The blood flow is examined as a representative of a fluid while gases resulting
from fruits and vegetables represent the investigated third aggregate state. The
monograph concludes with two related problems taken from the physics of the
human eye and the analysis and simulation of the human gait.
The editors wish to thank all the authors for their participation and cooperation
which made this volume possible. Finally, we would like to thank the team of

v
vi Preface

Springer-Verlag, especially Dr. Christoph Baumann, for the excellent cooperation

during the preparation of this volume.

November 2011 Andreas Öchsner

Lucas F. M. da Silva
Holm Altenbach
Contents

Part I Stress and Strain Analysis

Assessment of the Structural Integrity of C3–C5 Cervical

Porcine Vertebrae Model Based on 2D Classic CAD, 3D Scanner
and 3D Computed Tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
J. A. Beltrán-Fernández, L. H. Hernández-Gómez,
G. Urriolagoitia-Calderón, A. González-Rebatú,
G. Urriolagoitia-Sosa, M. M. Galán Vera and E. Escalante-Rodríguez

Stress–Strain Analysis of a Longitudinal Heterogeneous

Arterial Wall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Asawinee Danpinid, Pradit Terdtoon, Phrut Sakulchangsatjatai,
Jonathan Vappou and Elisa E. Konofagou

Stresses and Strains Analysis in the Left Ventricular Wall

with Finite Deformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Tipapon Khamdaeng, Phrut Sakulchangsatjatai, Niti Kammuang-Lue,
Asawinee Danpinid and Pradit Terdtoon

Stress Distribution in Dental Implant with Elastomeric

Stress Barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Ali Merdji, Belabess Bachir Bouiadjra, Rajshree Mootanah,
Boualem Serier, Tarik Achour and Noreddine Djebbar

Biomechanical Behavior Analysis of the Sap Ascent

in Vascular Plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Jose-Luis Perez-Diaz, Juan-Carlos Garcia-Prada,
Fernando Romera-Juarez and Efren Diez-Jimenez

vii
viii Contents

Experimental Investigation of the Surface Tension

of Lipid Membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Ying Zhang, Hui Fan, Weimin Huang and Yan Chen

Finite Element Quantification of the Compressive Forces

Induced by Keratinocyte on a Liquid Crystal Substrate . . . . . . . . . . . 79
C. F. Soon, M. Youseffi, P. Twigg, N. Blagden and M. C. T. Denyer

Part II Properties

Preparation of Hydroxyapatite-Forsterite-Bioactive Glass

Composite Nanopowder for Biomedical Applications . . . . . . . . . . . . . 103
Mohammadhossein Fathi, Vajihesadat Mortazavi
and Maryam Mazrooei Sebdani

Biomechanical Properties of Coronary Arteries Neonates:

Preliminary Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Normunds Sikora, Aris Lacis, Elina Ligere, Valts Ozolins,
Lauris Smits, Inta Bergmane and Vladimir Kasyanov

Biomechanical Properties of the Aorta in Neonates and Infants . . . . . 125

Elina Ligere, Normunds Sikora, Valts Ozolins, Lauris Smits,
Inta Bergmane, Aris Lacis and Vladimir Kasyanov

Part III Fluid and Gas

Numerical Study of Blood Flow Pressure Drop in Aorta

Coronary Sinus Conduit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Siti Aslina Hussain, Tan Hong Tat, Mohd Ismail Abdul Hamid,
Norhafizah Abdullah and Azni Idris

Biomechanical Modeling of Aneurysm Growth and Rupture

Using Fluid Structure Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
M. Mazwan Mahat, A. Juliawati and Ishkrizat Taib

Study and Mathematical Modeling of Transient Gas

Compositions for Modified Atmosphere Packaging . . . . . . . . . . . . . . . 163
A. Heydari, I. Alemzadeh and M. Vossoughi

Kinetic Modeling of Biogas Generation from

Banana Stem Waste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
N. Zainol, J. Salihon and R. Abdul-Rahman
Contents ix

Part IV Related Problems

Simulation of the Errors of Refraction in the Human Eye . . . . . . . . . 187

H. D. Silva, L. C. P. Dória, C. P. Dória, C. P. Dória,
M. C. T. D. Silva, H. D. Silva Jr. and M. T. D. Silva

Human Gait: Kinematics Analysis and Mechatronic

Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Alvaro Joffre Uribe, João Maurício Rosário
and José Tenreiro Machado
 Part I
Stress and Strain Analysis
Assessment of the Structural Integrity
of C3–C5 Cervical Porcine Vertebrae
Model Based on 2D Classic CAD, 3D
Scanner and 3D Computed Tomography

J. A. Beltrán-Fernández, L. H. Hernández-Gómez,
G. Urriolagoitia-Calderón, A. González-Rebatú,
G. Urriolagoitia-Sosa, M. M. Galán Vera
and E. Escalante-Rodríguez

Abstract In this chapter, the biomechanical behavior of C3–C5 porcine cervical

vertebrae is analyzed. The objective of this evaluation is to establish the advan-
tages and limitations of three numerical procedures when a compressive load is
applied. In a first stage, a damaged C4 vertebral body is instrumented with a bone
graft and a titanium alloy (Ti-6A1-4V) cervical plate fixed with titanium alloy
screws. In the second stage, the biomechanical integrity of a healthy C3–C5 unit is

J. A. Beltrán-Fernández (&)
L. H. Hernández-Gómez
G. Urriolagoitia-Calderón

G. Urriolagoitia-Sosa
M. M. Galán Vera
Instituto Politécnico Nacional. Escuela Superior de Ingeniería
Mecánica y Eléctrica (ESIME), Sección de Estudios de Posgrado e Investigación,
Unidad Profesional Adolfo López Mateos (UPALM), Edificio 5, 3er Piso,
07738 Mexico, D.F., Mexico
e-mail: [email protected]
L. H. Hernández-Gómez
e-mail: [email protected]
G. Urriolagoitia-Calderón
e-mail: [email protected]
G. Urriolagoitia-Sosa
e-mail: [email protected]
M. M. Galán Vera
e-mail: [email protected]
A. González-Rebatú
E. Escalante-Rodríguez
Hospital Regional ISSSTE 18 de Octubre, Av. Instituto Politécnico Nacional,
Núm. 1669. Col. Magdalena de las Salinas, 07760 Mexico, Mexico
e-mail: [email protected]
E. Escalante-Rodríguez
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 3
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_1,
Ó Springer-Verlag Berlin Heidelberg 2012
4 J. A. Beltrán-Fernández et al.

studied. The required numerical models were created with three different tech-
niques; these are 2D Computer Tomography (CT), 3D ZScan and CT scanning
with a Siemens Emotion system. This was done in conjunction with Pro-E Wildfire
4.0, Scan IP 3.1, UGS NX-4 and Geomagics R 10 codes. Lateral displacements
among the upper and lower surfaces of the vertebral bodies and the bone graft, as
well as the von Misses stresses, were calculated. Numerical differences from the
biomechanical models are discussed. In order to establish a performance criterion,
the results obtained were compared against those obtained for the case of the
instrumented C3–C5 unit. In order to establish helpful criteria to optimize the
therapeutic procedures before a surgery is performed, the analysis of the results
was focused to demonstrate that DICOM methodology can be applied when a
biomechanical simulation for a patient is required. It is possible, to apply this
technique safely as it is not invasive and geometrical parameters are obtained
directly from a tomography taken at a hospital. On the other hand, classical CAD
models and Z scan methodology has shown to be useful when specimens are
numerically analyzed.

Keywords Biomechanics 3D scanning
Computed tomography
Compressive


loads Porcine cervical vertebrae

1 Introduction

In recent times, new technologies to regenerate organic structures to be applied in

the medical area have been optimized. Their purpose is to create alternative
solutions on invasive and non invasive surgeries before recommending a specific
prosthesis. While the classical Computed Aided Drawing (CAD) technique has
reported very interesting results using basic software and 2D printed X-ray plates,
the newest methodologies have revealed several advantages. They have been
influenced by Digital Imaging and Communications Format (DICOM), in con-
junction with 3D helical scanning systems. Up to date, it is well known the
complexity of the models and the long periods of time that are required for the
generation of numeric bioorganic models. In particular, bone structures require
specific considerations regarding the kind of tissue and geometrical complexity
[1]. In this chapter, the case of a cervical vertebra was the base for this study,
taking into account the replacement of the damaged body.
In the past, the classical CAD method was used and it is based on the principle
of superposition of 2D printed and scanned X-ray plates, in order to generate a
complete 3D model. Nowadays, a precise reconstruction technique has been
developed. It is an easier and precise method to process each one of the computed
tomography sections. SCAN IP Software, Geomagics and Pro-Engineer codes are
required to process, not only the complexity of the structure, but to obtain affinity
to the real case, which is about 99.9%.
Assessment of the Structural Integrity 5

Fig. 1 Classical CAD C3–

C5 cervical model [2]

The motivation of this work is based on the reevaluation of a well known study
case [2], using new advances in technology and software. This option has been
designed in order to solve the consequences of the trauma in patients who have
suffered an unexpected impact causing an explosion of the vertebral cervical body.

2 Materials and Methods

2.1 Classical CAD Method

This work considers three different reconstruction techniques. For the first one,
reported results of a previous research work are considered [2]. A numerical CAD
C3–C5 porcine cervical model was developed, it was instrumented with a bone
graft and a titanium cervical plate [3], which is fixed to the vertebral body with
titanium screws. It is under the effect of a compressive axial load. Figure 1 shows
the model.
As the classical CAD modeling technique was the first applied method to
regenerate bioorganic structures, it is important to mention that the superposition
principle is one of the most common options to be used for this and several slices,
which were scanned on a simple 2D flat scanner, were processed as a digital image
format (e.g. bmp, jpeg, gif, tiff). Figure 2 illustrates an example of images obtained
by superposition [4].
Sagital and lateral views of 2D X-ray plate and its extrusion can be seen on
Fig. 2. The z-axis on a three-dimensional reference system is the most common
variable when a slice is inserted. The optimal width for each slice has to be based
on the printed sagital X-ray plate from the axial tomography, and this is controlled
by the accuracy of each equipment (Fig. 3). 0.1, 0.5, 0.8, 1, 1.5 or 2 mm are
typical parameters and depend on the kind of trauma experienced by the patient.
These typical parameters are the most frequent axial distances considered when
the case of a patient is studied by medical doctors in the process to locate traumatic
damages.
6 J. A. Beltrán-Fernández et al.

Fig. 2 Axial 2D tomography (a) and superposition method (b) to create a preliminary 3D model
(c) [4]

Fig. 3 Axial distance data in a cervical tomographic slice. a Printed information; b 2D axial
tomography of C3 [4]

Under the CAD software environment, each pre-scanned image had to be

inserted considering the previous axial distance, once the full slices have been
inserted, specific points are defined in order to draw each profile and a solid and
consistent set of boundaries for each axial tomography. From these, a closed
polyline is generated in order to get the particular profile for each slice, as shown
in Fig. 4.
Once the correct profile is created, the process is repeated for each slice. This is
in accordance to the number of slices taken for each cervical segment of the
affected zone. Specific tools to conform the whole vertebrae body in order to
regenerate the numerical model were required. In AutoCAD, the meshing and
surfacing tools are the simplest form for its application. However, in Pro-Engineer,
the blend of profiles contributes to a better simplification to this integration.
Assessment of the Structural Integrity 7

Fig. 4 a Point definition; b profile creation in an axial tomography

Fig. 5 Integration of the classic CAD model for cervical C3–C5 instrumented model [11]

In Fig. 5, the integration of the components for the numeric cervical model is
shown. Each one of the parts was modeled, based on the already described
methodology.

2.2 DICOM Methodology and Scan IP Tomography

The considerations for the regeneration of a porcine cervical model applying

DICOM methodology were based on the use of a CT scanning Siemens Somaton
Emotion System (Fig. 6) in conjunction with Pro-E Wildfire 4.0, Scan IP 3.1, UGS
NX-4 and Geomagics R 10 codes.
8 J. A. Beltrán-Fernández et al.

Fig. 6 Siemens Somaton Emotion system

Fig. 7 Replacing a vertebral body and instrumentation technique of a cervical plate (b) and bone
graft (a) on a C3–C5 porcine cervical specimen

This methodology implies some important aspects [5, 6]; a C3–C5 instrumented
porcine cervical segment specimen was digitalized in a Siemens Somaton Emotion
System. This was done in ‘‘18 de Octubre—ISSSTE’’ Hospital. Axial computed
tomography (CAT) was developed with SCAN IP (DELL CAM Co.). Sagital,
lateral and axial anatomic planes in three dimensional perspectives were required
for a correct bioregeneration. Soft and hard tissues, as well as, the ceramic and
metallic materials were considered each by themselves (Fig. 7).
Assessment of the Structural Integrity 9

Fig. 8 a Axial, b sagital and c lateral anatomic planes from the scanning process of the C3–C5
porcine cervical specimen

Fig. 9 a Bone profile boundary; b total number of CAT and c filling of solid bone structure,
metallic plate and bone graft

Density in each tomographic slice is an important factor, as the bone structure

should be recognized. For example, the cortical bone denoted by a consistent white
color, in contrast to cancellous bone, in which the color intensity is diluted with a
lower intensity as shown in Fig. 8.
148 computed tomographic slices were processed for the regeneration of the
instrumented cervical model, using 16 slices in a range of 2.5 mm in the axial
direction. As a result, each plane is separated 0.156 mm. A delimitation of each
structure (tissue, ceramic or metallic materials) with the paint tool under the sagital
view was created, as shown in Fig. 9. The purpose is to make a clear identification
of the materials to be considered for the Finite Element Analysis.
The integration of each tomographic section in a Stereolitography format (STL)
file made it possible to obtain a preliminary numerical model in order to be studied
under compressive loading conditions. Figure 10 shows the STL cervical file
in shrink-wrap format. It was necessary as well to make a conversion; the
Pro-Engineer W.F. 4.0 code was used.
10 J. A. Beltrán-Fernández et al.

Fig. 10 STL and shrink-

wrapped numerical cervical
model

Table 1 Mechanical properties of experimental cervical porcine model

Mechanical Value Units Reference
property
Cortical bone Elastic modulus 12 GPa [13]
Poisson’s ratio 0.2
Cancellous bone Elastic modulus 466 MPa [15]
Poisson ratio 0.3
Bone graft Maximum 15 MPa [16]
loading
strength
Cervical plate and fixation screws 3317/04 OS Elastic modulus 102 GPa [10]
(ortosintese) titanium alloy Ti6A14V Poisson’s ratio 0.3
Yielding 827 MPa
strength

Figure 10 shows the resulting smooth model. The ScanIP software allows the
working of a complex structure integrating different materials and an STL file
format is the resultant model. In a next stage, specific and advanced codes, such as
Geomagics, Pro-Engineer or Unigraphics NX are necessary in order to get a useful
numerical model to be processed with a FEM code. For this case, ANSYS code
was used under the specific loading and boundary conditions mentioned bellow,
and the material properties reported on Table 1.

2.3 Z-Scan Methodology

A 3D recording system Z-Scan 700 is also applied for the regeneration of the
cervical specimen system. Beltran et al. [7, 8] discussed the main characteristics of
this system and its specific requirements on the external surfacing of each element
Assessment of the Structural Integrity 11

Fig. 11 Porcine spine model with matte finishing. a Isometric view, b frontal view

to be scanned. This system allows digitalizing irregular geometries, such as bone

surfaces, and there is the possibility to process them as quick printing prototypes.
In this case, the cervical plate was painted with a non brilliant color in order to
avoid the reflectance of laser beams. The scan process of the specimen requires a
3D reference system. This is obtained by using reflective targets over the external
surface of the cervical specimen. Figure 11 illustrates the instrumented specimen
before it was scanned.
Some advantages of this process over the classic 2D CAD are the following: (1)
the operation of the hardware and software is simpler; (2) high resolution is
available in order to get porosity and details for the organic specimens. This is the
case of soft tissues and the structure of bone; and (3) the possibility to process
Stereolithography (STL) files as a universal format, which can be handled as a
faceted solid or shrink-wrapped. The instructions suggested by the ZCorp meth-
odology for the preliminary preparation of the specimen to be scanned are:
(a) Color calibration of the specimen. It is preferred to use matte uniform colors
(Fig. 11) [9].
(b) Determination of an ideal distance for data acquisition.
(c) Noise level reduction in the data (e.g. presence of dust).
Two different digitalization methods are applied. The first one requires a ref-
erence system in which the specimen is placed (Fig. 12a). The 3D reference
system allows scanning main surfaces after calibration of the principal colors. In
the second method, the use of reflective targets is allowed, as a reflective effect is
required on the scanning process. Therefore, it is necessary to avoid the interaction
of laser ray of the scanner with metallic surfaces of the titanium cervical plate with
no painting cover (Fig. 12b). The influence of the deposition of dust on the
geometry is barely perceptible (0.017 mm/0.0007 in.), [9].
A constant scanning over the surface of the specimen is necessary in order to
keep the scanner laser into the reference system. It avoids the inclusion of objects
with similar colors in the scanning session. Another alternative to scan different
12 J. A. Beltrán-Fernández et al.

Fig. 12 Three dimensional reference system to scan vertebral model. a Reference system,
b scanning of the specimen

Fig. 13 Partial view of the specimen with ZScan 700. a STL surface, b faceted model, c cervical
plate

kind of tissues is by covering the components with diverse colors, as is shown in

Fig. 13. The porcine spine specimen was also digitalized in this way and, later,
some of the surfaces had to be touched up in order to obtain a finished solid model.

3 Loading and Boundary Conditions

As in Beltran-Fernandez et al. [7], the loading and boundary conditions were based
on preliminary experimental work reported. This information is important because
as medical and engineering specialists decided to evaluate this instrumented cer-
vical segment, in order to determine its mechanical behavior and measure the
lateral displacements among the bone graft and the lower and upper surfaces of C3
and C5 respectively. A schematic diagram is shown in Fig. 14 [10].
Assessment of the Structural Integrity 13

Loading
compression
Titanium
Cervical plate
fixed with
Bone Graft screws

Damaged C4
vertebra

Constraints
on the lower
face of C3

Fig. 14 Boundary and loading conditions. a Damaged specimen, b fixation of the cervical plate,
c constraints

The mechanical properties of in vivo cervical specimens were considered for

the numerical analysis and Table 1 reports these parameters.

4 Results

The main numerical results for each specimen are reported in Table 2. The
obtained data [2, 4, 11] validated the surgery solution applied on the cervical range
C3–C5 (Clinically called Corporectomy), stabilization of the instrumented zone is
allowed after the fusion among the bone graft and the lower and upper surfaces of
C3 and C5, respectively. Under these considerations, it is meant that the bone graft
did not have significant displacements between surfaces of the vertebrae. This is in
accordance with the Müller criterion [12], which establishes that any lateral
movement of a cervical prosthesis should be\3 mm. In this way, a human spine is
stable after a surgery. Boundary conditions, mechanical properties and geometrical
parameters for the testing were described as well in past publications. [7–9, 13].
Table 2 reports the results for three loading and boundary conditions applied on
each one of the numerical models. Case 1 represents the weight of the head (38 N)
applied over the superior surface of C3, while the case 2 considers the average
weight of a patient (637.5 N) as a compressive load over the superior surface of
C3. For the case 3, the compression load failure of the cervical C5 vertebra
(6376.5 N) was considered. It was reported on preliminary experimental testings
[10].
 14

Table 2 Numerical results comparison between CAD, ZCorp Scan and CT image models
Case of study Location Classic CAD- ZCorp—3D CT Notes
specimen scan modela image 3D
mode
a
Case 1 Maximum von Misses— Centre of the cervical plate 70.3 Pa 68.9 Pa 66.5 Pa For these specimens, only
Stress Around the hole of the 316.2 Pa 303.7 Pa 305.3 Pa the surface was created
screws on the Scan session. The
Maximum displacement Interior surface of C3, 4.75 9 10-4 mm 4.39 9 10-4 3.48 9 10-4 soft tissue elements
between the bone graft superior surface of C5 mm mm were created on Pro-E
and upper and lower C3 adjacent to the surfaces 4.0 W.F
and C5 vertebrae of the bone graft
Case 2 Maximum von Misses— Centre of the cervical plate 662.8 Pa 648.7 Pa 635.7 Pa [2, 4, 11]
Stress
Maximum displacement Around the hole of the 3351.7 Pa 3329.6 Pa 3312.2 Pa
between the bone graft screws
and upper and lower C3 Interior surface of C3, 0.0054 mm 0.0042 mm 0.0039 mm
and C5 vertebrae superior surface of C5
adjacent to the surfaces
of the bone graft
Case 3 Maximum von Misses— Centre of the cervical plate 7.025 MPa 6.96 MPa 6.92 MPa [2, 4, 11]
Stress Around the hole of the 328.2 MPa 317.8 MPa 302.6 MPa
screws
Maximum displacement Interior surface of C3, 0.0546 mm 0.0483 mm 0.0411 mm
between the bone graft and superior surface of C5
upper and lower C3 and adjacent to the surfaces
C5 vertebrae of the bone graft
J. A. Beltrán-Fernández et al.
Assessment of the Structural Integrity 15

5 Conclusions

Actual engineering evaluation of medical solutions in the orthopaedic area requires

the knowledge of the biomechanical behavior of the involved structures [12].
For this case, the bone graft contributes to the spine stability, once the fusion
process with of C3 and C5 has taken place. This is one of the most recommendable
methodologies that medics and hospitals have approved.
As regards the results of the numerical simulation, it has been found that they
agree quite well with experimental testing. Table 2 shows convergence among
important parameters using diverse modeling techniques, for each case, the von
Misses stresses and the maximum displacements are lower for the CT image 3D
model than the CAD and ZCorp—3D scanner [14]. The accuracy of each method
depends on how the in vivo specimen is reproduced. In this way, the inclusion of
the bone density for the computed tomography probed to be very helpful and the
results were validated with experimental testing.
Classical CAD models and Z scan methodology are useful one, when speci-
mens are analyzed. However, DICOM methodology is a better alternative, which
can be applied, when a biomechanical simulation for a patient is required. Its main
advantages are the following: it is a non invasive technique and geometrical
parameters are obtained directly.
As final remark, it can be said that a good biomechanical analysis requires
both experimental and numerical approaches if it is expected to obtain the best
evaluation of the structural integrity. However, it has been found that the
development of numerical simulations is not an easy task, because a great level
of accuracy can not be easily achieved due to the complexity of the bone
geometry and the material anisotropy. This situation is further complicated when
distracters or prosthesis have to be included. The use of porcine specimens in the
experimental work takes into count similarities between human vertebrae, spe-
cifically mechanical properties. For this work, the numerical model was validated
by experimental techniques, it also showed the advantages and limitations of the
three procedures discussed above. It is important to observe as well, that these
simulations could be easily extrapolated to the human case, as geometrical data
can be obtained directly from the patient following a non intrusive procedure.
Therefore, it should be quite possible to stablish a criterion to improve the
surgery technique from these data. So it should be possible that specialists on
spine surgery could draw conclusions for a corporectomy to be performed prior
to the surgery of a patient.

Acknowledgments The authors kindly acknowledge the grants awarded by the National
Council for Science and Technology, CONACYT, Instituto de Ciencia y Tecnología del Distrito
Federal (ICyTDF), the National Polytechnic Institute (SIP20091599, SIP20100496 and
SIP20113474) and the support given to this project by the Hospital 18 de Octubre—ISSSTE.
16 J. A. Beltrán-Fernández et al.

References

1. D’Amico, D.S., Cisilino, A.P., Sammartino, M.R., Capiel, C.: Modelado computacional de
estructuras óseas utilizando el Método de los Elementos Finitos y tomografías computarizadas,
Análisis de la estabilidad de un implante Gleno-Humeral. Mecánica Computacional V. XXIV.
A. Larreteguy (Editor). Argentina (2005)
2. Beltrán-Fernández, J.A., Hernández-Gómez, L.H., Urriolagoitia-Calderón, G., Urriolagoitia-
Sosa, G., González-Revattú, A., Dufoo-Olvera, M.: Mechanical behavior of a calcium
phosphate ceramic bone graft used in the rehabilitation of a C4 human vertebra. Appl. Mech.
Mater. 7–8, 101–106 (2007)
3. Ortosintese: Catalog of products 2006, Rua Friedrich Von Voith, 896-Jaragua, Brasil (2006)
4. Beltrán Fernández, J.A., Hernández Gómez, L.H., Urriolagoitia Calderón, G., Dufoo Olvera, M.,
González Rebatú, A.: Distribución de esfuerzos por la acción de cargas de compresión en la
vértebra cervical C5, empleando el Método del Elemento Finito, Científica, 9(3),135–142 (2005),
Julio-Septiembre del 2005, ISSN 1665–0654
5. Ruíz-Muñoz, E.R., Beltrán-Fernández, J.A., Hernández-Gómez, L.H., Urriolagoitia-Sosa, G.,
Urriolagoitia-Calderón, G.: Técnicas de modelado en 3D aplicado a casos de vértebras
porcinas por medio de un escaner 3D y tomografías. Conferencia XXV Congreso Nacional de
Investigación Biomédica, UANL-Facultad de Medicina. Monterrey, México (2009)
6. Ruiz Muñoz, E.R.: Análisis de la estabilización segmentaria en especímenes de columna
lumbar porcina (L2–L5) con cerclaje interespinoso empleando cinturones de poliamida 6/6.
Tesis de maestría, SEPI—ESIME Zacatenco, México (2010)
7. Beltrán-Fernández, J.A., Hernández-Gómez, L.H., Urriolagoitia-Sosa, G., González-Rebatu, A.,
Urriolagoitia-Calderón, G.: Evaluation of a C3–C5 human cervical model created by computer
tomography CT and 3D scan under compression loading, 3rd international conference on
advanced computational engineering and experimenting, ACE-X 2009, book ok abstracts,
pp. 108–109 (2009)
8. Beltran-Fernández, J.A., Hernández-Gómez, L.H., Urriolagoitia-Calderón, G., González-
Rebatú, A., Urriolagoitia-Sosa, G.: Biomechanics and numerical evaluation of cervical
porcine models considering compressive loads using 2D classic computer tomography CT,
3D scanner and 3D computed tomography, advances in experimental mechanics VII. In:
Selected peer reviewed papers from 7th BSSM International Conference on Advances in
Experimental Mechanics, Transtech Publications, Liverpool, vol. 24–25, pp. 287–295
(2010)
9. Ruíz-Muñoz, E.R., Beltrán-Fernández, J.A., Hernández-Gómez, L.H., González- Rebatú, A.:
Análisis de la estabilización segmentaria en especímenes de columna lumbar porcina L3–L5,
con cerclaje interespinoso empleando collarines de poliamida 6/6’’. IX Congreso
Iberoamericano De Ingenieria Mecanica, Cibim 09. FEIBIM—Federación Iberoamericana
de Ingeniería Mecánica, Escuela Técnica Superior de Ingenieros Industriales de la
Universidad de las Palmas de Gran Canaria, España. 17–20 Noviembre (2009)
10. Beltrán-Fernández, J.A.: Análisis numérico de las cervicales C3–C7 asociado al problema del
latigazo cervical, Tesis de doctorado, SEPI—ESIME Zacatenco, México (2007)
11. Beltrán-Fernández, J.A., Hernández-Gómez, L.H., Merchan-Cruz, E.A., Urriolagoitia-
Calderón, G., González-Rebatú, A., Dufoo-Olvera, M., Urriolagoitia-Sosa, G.: Modelling
of a cervical plate and human cervical section C3 C5 under compression loading conditions
using the finite element method. Appl. Mech. Mater. 13–14, 49–56 (2008)
12. Müller, W.: Manual of Internal Fixation, 3rd ed. Científico-Médica, Barcelona (1992)
13. Hernández-Gómez, L.H., Beltran-Fernández, J.A., Urriolagoitia-Calderón, G., González-
Rebatú, A., Galan-Vera, M.M., Urriolagoitia-Sosa, G.: Biomechanical characterization
of analysis of adjacent vertebra behavior using porcine specimens. Key Eng. Mater. 478,
103–111 (2011)
14. Zcorp: Self-Positioning Handheld 3D Scanner, Method Sheet: Scanning Multi-colored or
Multi-shaded Part, Document: MS020016 (2007)
Assessment of the Structural Integrity 17

15. Jeremy, C., et al.: Relationship between CT intensity, micro-architecture and mechanical
properties of porcine vertebral cancellous bone. Clin. Biomech. 21, 235–244 (2006)
16. Moore, Ch., et al.: The evaluation of a biphasic calcium phosphate ceramic for use in grafting
long-bone diaphyseal defects. J. Orthop. Res. 5, 356–365 (1987)
Stress–Strain Analysis of a Longitudinal
Heterogeneous Arterial Wall

Asawinee Danpinid, Pradit Terdtoon, Phrut Sakulchangsatjatai,

Jonathan Vappou and Elisa E. Konofagou

Abstract Stress distribution in the arterial wall has been shown to be associated
with several vascular disease, e.g., atherosclerosis and abdominal aortic aneurysm
(AAA). The variation of material properties of the vascular wall is related to the
local onset of disease, which leads to the alterations of stresses and strains that
provide essential clinical information. We therefore investigate the distributions of
stresses and strains in a longitudinal, heterogeneous arterial wall under physiologic
loadings. Four Models were analyzed: Model 1 was the idealized Young’s
modulus for normal aorta, Model 2 was the idealized Young’s modulus for
pathological aorta, Model 3 was the actual Young’s modulus for the normal aorta,
and Model 4 was the actual pathological (AngiotensinII-treated) aorta. The arterial
Models were assumed to have idealized geometry, i.e., axisymmetric, uniform,
cylindrical, purely elastic. In Model 1, we assumed the wall material to be
homogeneous, i.e., described by its Young’s modulus equal to 150 kPa. One stiff
inclusion was defined at the middle part of Model 2, by changing its Young’s

A. Danpinid (&)  P. Terdtoon  P. Sakulchangsatjatai

Department of Mechanical Engineering, Chiang Mai University,
Chiang Mai, Thailand
e-mail: [email protected]; [email protected]
P. Terdtoon
e-mail: [email protected]
P. Sakulchangsatjatai
e-mail: [email protected]
J. Vappou  E. E. Konofagou
Department of Biomedical Engineering, Columbia University,
New York, NY, USA
e-mail: [email protected]
E. E. Konofagou
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 19
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_2,
Ó Springer-Verlag Berlin Heidelberg 2012
20 A. Danpinid et al.

modulus 150–550–150 kPa. The local Young’s moduli, which are experimentally
estimated from our previous study, in Model 3 and 4 were equal to 140.8 ± 10.2
and 1676.8 ± 926.0 kPa, respectively. Stress and strain calculations in all cases
were performed using a finite-element method. The transmural stress and strain
decreased toward the outer wall uniformly in Model 1, i.e., stress ratio was equal
to 133.3%. But the stress dramatically increased at the boundary of stiff-soft region
in Model 2, i.e., stress ratio was equal to 187.0%. While the trends of stresses and
strains in Model 1 were uniform longitudinally, sharp-increases of stresses in
Model 2 were found at the boundaries between high and low Young’s moduli,
i.e., 25% higher of the adjacent sections. Model 3 and 4 have depicted similar
stress and strain trends as Models 1 and 2. However, Model 3 is not perfectly
homogeneous, and thus slight increases stress in axial direction, and an even
higher number of such increases were found in Model 4. This is because of the
longitudinal change in material properties that cause changes in the stress and
strain distribution. These stress discontinuities may indicate a higher risk for
rupture in the affected area. A feasibility study was thus shown for monitoring
suspicious wall regions that may be more prone to disease due to their higher
heterogeneity.

Keywords Abdominal aorta  Heterogeneous  Finite element method  Stress

strain distributions

1 Introduction

Large arteries adjust the pulsatile blood flow from the heart to be more steady in
the small arteries and capillaries. This function serves as a blood reservoir in the
diastolic phase of the heart, keeps the blood moving continuously throughout the
entire body. The aortic wall is basically composed of cells and extracellular matrix
arranged into a concentric complicated circular tube. The two main types of cells
are endothelial and vascular smooth muscle (VSM) cells working together to
control the vasomotion tone. The extracellular matrix is made of various types of
molecules, namely, elastin, collagen fibers and ground substance (gel-like mate-
rial) [1]. 50% of dry aortic wall consists of elastin and collagen, and other is VSM
cells and non-fibrous matrix. The elastin and collagen are the major contributors to
the elastic response of the tissue [2].
Most cardiovascular disease result in arterial wall damage, and are associated
with altered elastic properties. The arterial wall becomes weak, stiffer, and dys-
functional, which is directly associated to the wall stress [3–6].
Stress plays an important role in the physiology of the arterial wall. For
instance, shear stress due to blood flow is detected by endothelial cells, and the
neurological signal chemically activates the vessel vasomotion tone in order to
adjust the blood pressure. The effect of wall stress and stretch on the plaque,
Stress–Strain Analysis 21

e.g., a lipid pool underneath the arterial wall, can be mediated by the endothelial
cells, the smooth muscle cells, and the penetration of low-density lipoproteins
associated to the atherosclerosis [5]. The aorta can be ruptured when the wall stress
exceeds the wall strength in abdominal aortic aneurysm (AAA), and the damaged
area of the wall can be predicted using stress concentration [7, 8]. Previous works
have shown that the geometry plays a significant role in the stress distribution
[8, 9]. Arterial blood pressure level also influences the peak wall stress [7].
In the existing literature, the aorta was assumed to be material homogeneous
[8, 9], i.e., only one elastic modulus was employed. This is not true since the
elasticity of the wall varies from high-elastic gradually to be less-elastic, i.e.,
elastic sheet decreasing and VSM increasing, along the distance from the heart
[1, 2]. Also, the variation in material properties is directly related to the local onset
of the vascular disease.
Moreover, due to the fact that the ability to localize disease along the aorta is
needed in the clinical setting, it is very important to investigate how the variability
of material heterogeneity affects the radial stress and strain. We, therefore, focus
on distributions of stresses and strains of a longitudinal heterogeneous arterial wall
under physiologic aortic pressure using a finite-element method. In this study,
we have performed the study on normal and diseased cases, separated into
idealized and actual local E aorta. The local Young’s moduli of normal and
Angiotensin (AngII)-treated mouse aortas from our previous work were used [10]
to investigate the effect of actual heterogeneity in axial direction. AngII infusion-
based Model has been used to increase the blood pressure and inducing the
Abdominal Aortic Aneurysm (AAA) in the mouse Model [3, 11, 12].

2 Methods

Aortic Models in this study were performed using an idealized geometry,

i.e., axisymmetric circular tube. To investigate the basic on how material alterations
influence stress and strain, we simulated idealized material property cases,
i.e., idealized normal (Model 1) and idealized pathological aorta (Model 2). Actual
local E aortas were separated into normal (Model 3) and AngII-treated cases (Model
4) were subsequently simulated using local elastic moduli from our previous work
[10]. Transmural wall stress and strain distributions in four cases were calculated
using the finite-element method (FEM). The Model description of simulated and live
mouse aortas and stress–strain analysis are provided as follows.

2.1 Model Set Up

In the simulation, the aortic wall was assumed to be axisymmetric, uniform,

cylindrical and made of a purely incompressible elastic material. The wall was
inflated by static uniform physiologic aortic pressures, i.e., 75 and 120 mmHg,
22 A. Danpinid et al.

Fig. 1 a The aortic pressure over one cardiac cycle. b The corresponding aortic diameter change
over one cardiac cycle. The wall is assumed to be purely elastic material, i.e., the minimum and
maximum peaks were aligned and the deformation in systolic phase is considered. c B-mode
reference ultrasound image of a mouse aorta. Four red dots indicate the selected inner and outer
wall location to calculate stress–strain relationship. d The stress–strain relationship of the aortic
wall at the location in Fig. 1c at systolic phase

applied at the inner surface of aorta. Shear stress caused by internal blood flow was
ignored. The wall velocities at the two ends were set to be zero. The elastic moduli
were evaluated in vivo [10]. Briefly, circumferential stress and strain were eval-
uated using mouse aortic pressure from catheterization and local corresponding
aortic wall diameters from a cross-correlation technique on the radio frequency
(RF) ultrasound signals at 30 MHz and frame rate of 8 kHz [13, 14]. The stress–
strain relationship was defined to be a piece-wise linear function which consisted
of three Young’s moduli, i.e., associated to elastic lamellae (E1), elastic–collagen
fibers (E2) and collagen fibers (E3), as depicted in Fig. 1.
E 2 ¼ E 1 þ E3 ð1Þ
E2, the contribution between elastin and collagen, were chosen to be Young’s
moduli (E) for the present study, because the deformation of the wall at 75 mmHg
corresponded to the range of E2.
Stress–Strain Analysis 23

Fig. 2 Model set up of a Model 1. b Model 2. c Model 3. d Model 4

Regarding the finite element analysis, Abaqus 6.8.1, the elements were 8-node
hexahedral using curvature controlling mesh.

2.1.1 Idealized Aortas Modeling (Model 1 and Model 2)

Figure 2a shows Model 1, which was assumed to be purely homogeneous,

i.e., described by its Young’s modulus (E) was equal to 150 kPa (mean value of E2 in
normal aortas of [10]). A 0.5 mm stiffer portion, assuming to be an abnormal area,
was defined to be at the middle part of the Model 2 (E = 550 kPa, mean value of E2 in
AngII-treated aortas of [10]) as shown in Fig. 2b. The wall dimension was 3.5 mm of
length, 1 mm of inner diameter and 0.1 mm of wall thickness.

2.1.2 Actual Local E Aorta Modeling (Model 3 and Model 4)

In order to define the realistic trend of longitudinal material variability, the local
E from one normal and one AngII-treated aorta [10] of 140.8 ± 10.2 and
1,676.8 ± 926.0 kPa, were used as input for the material properties in Models 3
and 4, respectively (Fig. 2c, d). The number of selected locations, indicating the
local portions of E related to the lateral resolution of the image, of both cases was
24 A. Danpinid et al.

equal to 35. The corresponding wall dimension was 2.5 mm of length, 1 mm of

inner diameter and 0.1 mm of wall thickness. The animal preparation followed the
details in [10, 13]. The local E was consequently estimated as mentioned in
Sect. 2.1, and input into Models 3 and 4.

2.2 Stress and Strain Analysis

Von Mises stress ðrVM Þ was computed to represent the stress distribution for each
Model. It is derived from distortion energy in failure theory of a ductile material:
pffiffiffiffiffiffiffi
rVM ¼ 3J2 ; ð2Þ

where J2 is the second deviatoric stress invariant. In this chapter, stress ratioin/out
was defined to be a percentage ratio between stress at the inner (rinner) and outer
surfaces (router) to consider the change in radial direction.
rinner
Stress ratioin=out ¼  100 ð3Þ
router
To monitor the stress-discontinuity, or stress concentration, caused by the
material heterogeneity in axial direction, the percentage in different wall stress at
the adjacent section was defined as Stress Increase.
Maximumbjrith  rith1 j; jrith  rithþ1 jc
Stress Increase ð%Þ ¼  100 ð4Þ
rith
where rith is wall stress at the interested section ith. Both stress ratio and stress
increase is implied to the stress concentration that results in high risk of damage
area.
Regarding the strain, maximum principal strain in percentage was chosen to
present the strain distribution. Strain expresses the deformation of material since
the displacement compatibility is satisfied. The relatively low strain can be implied
to the low distensibility because of the high Young’s modulus.

3 Results and Discussions

Figures 3 and 4 show the von Mises stress and maximum principal strain distri-
butions in four cases in cross section view, respectively. The stress and strain at the
regions close to the two ends of the wall, were ignored, as they resulted from the
effects of constrained displacement boundaries. The Young’s moduli, transmural
stress and strain along the longitudinal direction of the aorta were compared in
Models 1 and 2, Models 3 and 4 in Figs. 5 and 6, respectively.
Stress–Strain Analysis 25

Fig. 3 Von Mises stress distributions at 75 and 120 mmHg of aortic pressures (P). a Model 1.
b Model 2. c Model 3. d Model 4
26 A. Danpinid et al.

Fig. 4 Strain distributions at 75 and 120 mmHg of aortic pressures (P) a Model 1. b Model 2.
c Model 3. d Model 4
Stress–Strain Analysis 27

Fig. 5 a Young’s modulus (E) along Model 1. b Von Mises stress at inner, middle, and outer
wall of normal Model 1 related to the E in Fig. 5a. c Strain along the axial location corresponding
to Fig. 5a, b. d Young’s modulus (E) along Model 2. e Von Mises stress at inner, middle, and
outer wall of Model 2 related to the E in Fig. 5d. f Strain along the axial location corresponding to
Fig. 5d, e

3.1 Stress and Strain Distributions in Model 1 and 2

In radial direction, the maximum stress was at the inner wall and decreased toward
the outer wall in both Models 1, 2. The differences between inner and outer wall
stresses were uniform along the axial location in the normal case (Fig. 5b), while
those in the diseased cases were extremely high at the interface between the stiff
(E = 550 kPa) and soft (E = 150 kPa) regions (Fig. 5d). Consequently, the stress
ratioin/out of the normal simulated aorta was equal to 133.0% (gray line, Fig. 7a),
and 187.0% at the boundary of different E the simulated aneurysmal aorta, while
the others region were close to that in normal case (black line, Fig. 7a). Also, the
stress in the axial direction dramatically increased at this interface area, yielding a
28 A. Danpinid et al.

Fig. 6 a Young’s modulus (E) along Model 3. b Von Mises stress at inner, middle, and outer
wall of Model 3 related to the E in Fig. 6a. c Strain along the axial location corresponding to Fig.
6a, b. d Young’s modulus (E) along Model 4. e Von Mises stress at inner, middle, and outer wall
of Model 4 related to the E in Fig. 6d. f Strain along the axial location corresponding to Fig. 6d, e

stress increase of 133.6% (black line, Fig. 7b), while that in the normal case was
close to zero (gray line, Figure 7B). The similar trends of stress were found in 75
and 120 mmHg, but latter case was higher in magnitudes (Fig. 5b, e). Therefore,
the stress ratio and stress increase of 75 and 120 mmHg were the same as shown
in Fig. 7.
Fig. 5c shows the uniform strains along the axial location in the normal simulated
case. Maximum strains occurred at the inner wall (solid lines, Figure 5c, f) and
decreased toward the outer wall. For the simulated pathological case (Fig. 5f), lower
strains (1.8 9 10-2) were found at the high E region relatively to the (7.4 9 10-2) in
low E region.
Stress–Strain Analysis 29

Fig. 7 a The stress ratioin/out of Model 1 and 2. b The stress increase of Model 1 and 2. c The
stress ratioin/out of Model 3 and 4. d The stress increase of Model 3 and 4

Since the displacement of the wall had to satisfy the compatibility conditions,
the strain increased and decreased continuously in our simple diseased case
(Fig. 5f). During the elastic deformation, the wall could not be ruptured when the
load was applied, but the material property was not homogenous in our study.
We, therefore, observed the dramatically increase in stress (Fig. 5e), i.e., stress
increase (Fig.7a, b), reflecting the results of material heterogeneity. The more
variation in longitudinal material heterogeneity is typically discovered in actual
aorta. It is expected to note these stress concentrations at the boundaries of
different E.
To our knowledge, the simulation on stress and strain distribution of a longi-
tudinal heterogeneous has not been found, but radial heterogeneous [15, 16]. It was
found that the dramatically stress increases were at the boundary of the different
Young’s modulus [15, 16]. This is correlated to our result, but with different
direction of the heterogeneity.

3.2 Stress and Strain Distributions in Model 3 and 4

In Model 3, the wall stress decreased from the inner to outer wall (Fig. 6b),
slightly non-uniformity relatively to the simulated aorta. The stress ratioin/out and
the stress increase were equal to 132.9 ± 1.8% and 2.4 ± 2.2% (mean ± std),
respectively (gray lines, Figure 7a, b). The sections for each E were very small,
30 A. Danpinid et al.

i.e., equal to the ultrasound image resolution (detailed in [10, 14]) and the
coefficient of variation (CV) was low, i.e., 0.072, therefore, the stress increase was
not clearly seen to be high. This is in agreement with the aforementioned
assumption since the normal aorta was set to be homogeneous.
In Model 4, several higher stress concentrations were noted in the AngII-treated
aorta (Fig. 6e), especially at the high different E regions. The stress ratioin/out and
the stress increase were equal to 141.3 ± 9.6% and 8.8 ± 6.4% (mean ± std),
respectively (black lines, Fig. 7c, d).
The increasing stress in this case was lower than the simulated pathological
aorta due to the gradually alteration in E along the wall. This is because the whole
vessels in the mouse have been changed after AngII treatment. The variability in
longitudinal heterogeneous was higher than that in normal one, i.e., CV were equal
to 0.158. Higher and higher stress increase were observed relatively to normal
mouse aorta, because of the AngII results in the decrease of elastic lamella and
increase the collagen formation [3, 11], consequently associated with the AAA in
mouse Model. Therefore, the region, where the high stress increase was found, can
be used to monitor the rupture risk of the damaged area [7, 9, 17].
In Models 3 and 4, similar trends of stress were also found in the case of 75 and
120 mmHg, but the latter case was only higher in magnitudes (Fig. 6b, e).
Therefore, the stress ratio and stress increase of 75 and 120 mmHg were the same,
as shown in Fig. 7.

3.3 Limitations

However, there was no evidence that indicates the location of the exact damage
area along the aorta. Our technique is also limited for a clear B-mode ultrasound
image, thus, some interesting or critical region may not be always visible. In
addition, more factors as geometry and type of material properties could contribute
the stress concentration. These factors and more animal experiments in various
pathologic conditions should be further studied.
Among several limitations, this study can provide a basic idea on how the
longitudinal material heterogeneity affects the stress concentration in radial and
axial direction of the aortic wall, which is related to the damaged area from
vascular disease.

4 Conclusion

Stress and strain concentrations have been investigated in a longitudinal, hetero-

geneous material in idealized normal and pathological cases, and also expanding to
actual local E aortas. The stress concentration in pathological aorta (Models 2 and
4), have shown to be higher than in the normal cases (Models 1 and 3), because of
Stress–Strain Analysis 31

the higher variability in material properties. This study proposes a feasible way to
indicate the high risk area of the tissue to be damaged on the longitudinal heter-
ogeneous aortic wall.

Acknowledgments This study was supported by Royal Golden Jubilee Scholarship (1.M.CM/
47/A.2) under Thailand Research Fund. We are grateful to Jianwen Luo, Phd., Kana Fujikura,
M.D. PhD., from Columbia University and Jawad Latif, M.D., from St. Luke’s-Roosevelt
Hospital Center, for conducting the experiments.

References

1. Humphrey, J.D.: Cardiovascular solid mechanics: cells, tissues, and organs. Springer,
New York (2002)
2. Nichols, W.W., O’Rourke, M.F.: McDonald’s blood flow in arteries: theoretical,
experimental, and clinical principles, 5th edn. A Hodder Arnold Publication, London (2005)
3. Tham, D.M., Martin-McNulty, B., Wang, Y.-X., Da Cunha, V., Wilson, D.W., Athanassious,
C.N., Powers, A.F., Sullivan, M.E., Rutledge, J.C.: Angiotensin II injures the arterial wall
causing increased aortic stiffening in apolipoprotein E-deficient mice. Am J Physiol Regul
Integr Comp Physiol 283, R1442–R1449 (2002)
4. Lee, R.T., Schoen, F.J., Loree, H.M., Lark, M.W., Libby, P.: Circumferential stress and
matrix metalloproteinase 1 in human coronary atherosclerosis: implications for plaque
rupture. Arteriosclerosis Thrombosis and Vascular Bio 16, 1070–1073 (1996)
5. Thubrikar, M.J., Robicsek, F.: Pressure-induced arterial wall stress and atherosclerosis. Ann.
Thorac. Surg. 59, 1594–1603 (1995)
6. Yamabe, M., Tomiyama, H., Hirayama, Y., Gulniza, Z., Takata, Y., Koji, Y., Motobe, K.,
Yamashina, A.: Arterial stiffening as a possible risk factor for both atherosclerosis and
diastolic heart failure. Hypertens. Res. 27, 625–631 (2004)
7. Raghavan, M.L., Vorp, D.A., Federle, M.P., Makaroun, M.S., Webster, M.W.: Wall stress
distribution on three-dimensionally reconstructed models of human abdominal aortic
aneurysm. J. Vascul. Surg. 31, 760–769 (2000)
8. Vorp, D.A., Raghavan, M.L., Webster, M.W.: Mechanical wall stress in abdominal aortic
aneurysm: Influence of diameter and asymmetry. J. Vasc. Surg. 27, 632–639 (1998)
9. Vorp, D.A.: Biomechanics of abdominal aortic aneurysm. J. Biomech. 40, 1887–1902 (2007)
10. Danpinid, A., Luo, J., Vappou, J., Terdtoon, P., Konofagou, E.E.: In vivo characterization of
the aortic wall stress–strain relationship. Ultrason. 50, 654–665 (2010)
11. Daugherty, A.: Mouse models of abdominal aortic aneurysms. Arterioscler. Thromb. Vasc.
Biol. 24, 429–434 (2004)
12. Manning, M.W., Cassis, L.A., Huang, J., Szilvassy, S.J., Daugherty, A.: Abdominal aortic
aneurysms: fresh insights from a novel animal model of the disease. Vascul. Med. 7, 45–54
(2002)
13. Luo, J., Fujikura, K., Tyrie, L.S., Tilson, M.D., Konofagou, E.E.: Pulse wave imaging of
normal and aneurysmal abdominal aortas in vivo. Med. Imaging. IEEE Trans. 28, 477–486
(2009)
14. Fujikura, K., Luo, J.W., Gamarnik, V., Pernot, M., Fukumoto, R., Tilson, M.D., Konofagou,
E.E.: A novel noninvasive technique for pulse-wave Imaging and characterization of
clinically-significant vascular mechanical properties in vivo. Ultrason. Imaging 29, 137–154
(2007)
15. Gao, F., Watanabe, M., Matsuzawa, T.: Stress analysis in a layered aortic arch model under
pulsatile blood flow. 5, 25 (2006)
32 A. Danpinid et al.

16. Holzapfel, G.A., Gasser, T.C., Ogden, R.W.: Comparison of a multi-layer structural model
for arterial walls with a fung-type model, and issues of material stability. J. Biomech. Eng.
126, 264–275 (2004)
17. Di Martino, E.S., Guadagni, G., Fumero, A., Ballerini, G., Spirito, R., Biglioli, P., Redaelli,
A.: Fluid–structure interaction within realistic three-dimensional models of the aneurysmatic
aorta as a guidance to assess the risk of rupture of the aneurysm. 23, 647–655 (2001)
Stresses and Strains Analysis in the Left
Ventricular Wall with Finite
Deformations

Tipapon Khamdaeng, Phrut Sakulchangsatjatai,

Niti Kammuang-Lue, Asawinee Danpinid
and Pradit Terdtoon

Abstract The dysfunction of the left ventricle causes the death of a large amount
of people each year. One of the most dangerous diseases is heart attack originated
by ischemia. In order to specify some indices for diagnosis in clinical treatment,
the material parameters of the left ventricular (LV) wall need to be determined.
These indices were determined based on the understanding of material properties
(stress–strain relationship) of normal heart to assess its fundamental mechanisms.
In this study, transmural stress and strain distributions from base to apex in the
normal canine heart during systolic and diastasis phase were determined using the
previously well-established constitutive relation for the active (Hunter et al.,
J. Biophys. Mol. Bio. 1 69,289–331, 1998) and passive (Humphrey et al., ASME
J. Biomech. Eng. 112,333–141, 1990) LV wall in terms of a pseudostrain-energy
function (W). A thick-walled truncated conical shell geometry was employed and
LV wall was assumed to be transversely isotropic, incompressible and with a
homogeneous deformation. The left ventricular anterior and posterior deforma-
tions were measured by the tetrahedron-shaped crystals implanted movement.

T. Khamdaeng  P. Sakulchangsatjatai  N. Kammuang-Lue 

A. Danpinid (&)  P. Terdtoon
Department of Mechanical Engineering,
Chiang Mai University, Chiang Mai, Thailand
e-mail: [email protected]; [email protected]
T. Khamdaeng
e-mail: [email protected]
P. Sakulchangsatjatai
e-mail: [email protected]
N. Kammuang-Lue
e-mail: [email protected]
P. Terdtoon
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 33
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_3,
Ó Springer-Verlag Berlin Heidelberg 2012
34 T. Khamdaeng et al.

Finite deformations including inflation, extension, twist and transmural shearing

were considered in the model. Deformation parameters were considered as a linear
variation from the inner to the outer LV wall. All six components of strain and
three components of stress, radial stress and two components of transmural
shearing stress were a function of radius. Moreover, stresses and strains relation is
nonlinear due to fiber orientation changing across the LV wall. The radial strains
indicate thinning and thickening of LV wall at end-diastole and end-systole,
respectively. The highest circumferential stresses occur at approximately the
middle surface, and also increase from the apex to the base at end-diastole and vice
versa at end-systole. The highest radial stresses occur at the inner surface and
decrease gradually across the LV wall.

Keywords Constitutive relation  Finite deformations  Left ventricular wall 

Stresses–strains distribution

1 Introduction

Stresses–strains distributions of the left ventricular (LV) wall were investigated [1–
3] in order to get an understanding of the LV wall behavior, mechanics and perfor-
mance which are important to the physiologist for more sophisticated diagnosis in
clinical treatment. The stresses distribution can influences coronary blood flow, LV
wall oxygen consumption, overall cardiac function, rate and extent of hypertrophy in
certain pathological conditions [2] and remodeling [1]. However, the forces and
stresses developed in the intact LV wall cannot be directly measured throughout the
cardiac cycle. Constitutive relations which characterize a material and its response to
applied forces are an alternative approach to determine stresses of LV wall [4–6].
There are various techniques to obtain deformation of LV wall. Its deformation can
be in vivo measured from the displacements of invasive crystal beads implanted in
the LV wall [7] or originated by noninvasive magnetic resonance (MR) tagging or
myocardial elastography [8] technique. For calculating in vivo LV wall forces, the
various versions of Laplace’s law were applied [9, 10]. ‘‘Modified Laplace’s law’’ [9]
was considered in thick-walled ellipsoidal ventricles at only equatorial region since
the maximum stresses occurred.
The previous constitutive relations have been mostly established from in vitro
experimental data which characterize the response of LV wall as a transversely
isotropic and orthotropic material using biaxial tests [11] and shear tests [12],
respectively. The cylindrical transversely isotropic LV models based on finite
elasticity have been applied to estimate three-dimensional stresses distributions
dominated by changing of fiber orientation across the LV wall at end-diastole [2]
and at end-systole [1]. The passive and active mechanics of the LV wall
described the gross mechanical behavior based on continuum hypothesis and
pseudo-elasticity concept. However, the different geometry of the LV was affected
Stresses and Strains Analysis in the Left Ventricular Wall 35

by the stress distribution of the LV wall. The conical model leads to more realistic
estimated stresses than the spherical and cylindrical models [13]. Therefore, the
purpose of this chapter is to estimate stresses and strains distribution by a finite
elastic analysis of the conical model.
In this chapter, previously well-established constitutive relation for passive [4]
and active [6] LV wall in terms of a pseudostrain-energy function (W) were
applied to estimate transmural stresses and strains distributions during diastasis
and systolic phase at any height of truncated conical shell from the base to the
apex. The material parameters were obtained from minimization of the values of
the circumferential stress calculated from the modified Laplace’s law and that
from the constitutive relations. From the results, the characteristic of LV wall
subjected to LV pressure can be revealed, and can be used to guide the critical
stress. Also, the basis of calculation of the LV wall stresses for clinical diagnosis is
obtained.

2 Method

2.1 Geometry Setup

A thick-walled truncated conical shell geometry was employed to study the pas-
sive and active LV. The height of the truncated cone is three times the radius at the
base which is four times the radius at the apex [13] (Fig. 1).
The deformed radius r at the inner surface was given directly from circum-
ferential strain ðEHH Þ and then r at any point in transmural direction from the base
to the apex can be determined from the incompressibility relation.

2.2 Kinematics Equations

LV wall exhibits a large deformation, i.e. the magnitude of strains during end-
diastole to end-systole is on the order of 20–30% [8], so the finite deformation
elasticity theory was used. The LV wall was assumed to be transversely isotropic,
incompressible and with a homogeneous deformation. Finite deformations
including inflation, extension, twist and transmural shearing were considered [2].
The polar coordinates in the unloaded and loaded configuration are ðR; H; ZÞ and
ðr; h; zÞ in radial, circumferential and axial direction, respectively, so it could be
written in the Cartesian coordinate as

y1 ¼ r cos h; y2 ¼ r sin h; y3 ¼ z
z ð1Þ
x1 ¼ R cos H; x2 ¼ R sin H; x3 ¼
k
36 T. Khamdaeng et al.

Fig. 1 The LV wall Ro

approximated as a thick-
walled truncated conical shell Ri

Base

Ro
4 Z=3h
Midventricle 4
Ri
4
h = 4 Ri

Apex

Furthermore, the stresses would not depend on Z-axis if the inflation with
constant extension ratio was considered, namely
r ¼ rðRÞ; h ¼ H þ wZ þ xðRÞ; z ¼ kZ þ wðRÞ ð2Þ
where w is the twist per unit unloaded length, x and w are the circumferential and
axial displacement with respect to radial directions, respectively.
From (1) and (2), the components of the deformation gradient were determined
as follow
2 3 2 0 3
or=oR or=RoH or=oZ r 0 0
FmM ¼ 4 roh=oR roh=RoH roh=oZ 5 ¼ 4 x0 r=R rw 5 ð3Þ
oz=oR oz=RoH oz=oZ w0 0 k

where

r 0 ¼ dr=dR; w0 ¼ dw=dR; x0 ¼ dx=dR

The passive and active constitutive relations were a function of the first strain
invariant ðI1 Þ and fiber stretch ratio ðaÞ only. Moreover, I1 and a were obtained
from the components of the left and right Cauchy–Green deformation tensors
which depend on r, w; k; x0 and w0 (defined as the deformation parameters). Thus,
the first strain invariant could be given by

r2
I1 ¼ r 0 2 þ w2 r 2 þ r 2 x02 þ þ k2 þ w0 2 ð4Þ
R2
Stresses and Strains Analysis in the Left Ventricular Wall 37

and the square of the fiber stretch ratio was

* * *
a2 ¼ N  C  N ¼ CMN NM NN ð5Þ
*
where N is a unit vector in the direction of a muscle fiber in the unloaded con-
*
figuration that represents the transmural muscle fiber distribution in the LV wall. C
is the right Cauchy–Green deformation tensors.
r, w; k; x0 and w0 can be calculated from the components of the Green strain at the
transmural location in the LV wall. The values of w0 and x0 can also be determined
separately for each radial location in the LV wall, 0.2 and 0.1 rad cm-1 [2],
respectively. The known parameters for the estimated stress distributions at the end-
diastole were: r = 2.31 cm, k ¼ 1:08 and w ¼ 0:0016 rad cm-1 with a LV cavity
pressure of 6.565 mmHg and those for the estimated stress distributions at the end-
systole were: r = 2.05 cm, k ¼ 1:07 and w ¼ 0:0012 rad cm-1 with a LV cavity
pressure of 75.32 mmHg. Ri and Ro were equal to 2.10 and 3.45 cm, respectively, for
the unloaded configuration.
Regarding the muscle fiber orientations ðUÞ; the equation to represent a non-
linearly helices-like distribution of muscle fiber orientations from the endocardium
to the epicardium is
 
½2R  ðR0 þ Ri Þ 3
UðRÞ ¼ U0 ð6Þ
½R0  Ri 
The value of U at the epicardium was defined as U0 = -60° and varies
cubically to the endocardium [2].

2.3 Constitutive Relations

The mechanical properties of the LV wall have been determined using the different
material laws based on the different theoretical frameworks. In fact LV wall
exhibits nonlinear, anisotropic and viscoelastic behavior [7], however, precondi-
tioned LV wall tissue can be shown that it is not truly hyperelastic material which
obtained unique stress–strain responses in separately the loaded and unloaded
protocols, thus, it can be called pseudoelastic material [14]. The deformation or the
strain of the material will be done from the strain-energy or potential energy
virtually stored in the body. The strain-energy function is theoretical analysis
result of bodies subjected to finite deformation [14]. The stress–strain curve, which
is associated with the strain-energy function, of the pseudoelastic material can be
correspondingly described via a pseudostrain-energy function.
The several constitutive equations based on assumptions of hyperelastic
material [1, 2, 10–12, 15, 16] were presented. Humphrey et al. [4] proposed the
transversely isotropic pseudoelastic constitutive equation for LV wall as a thick-
walled cylinder including inflation, extension, twist and transmural shearing with
38 T. Khamdaeng et al.

incompressible and a homogeneous deformation. It was derived directly from

biaxial test data. They also presented a polynomial form of pseudostrain-energy
function containing a coupling term between I1 and a with five material parameters
described a coupling term between a matrix and fiber orientation of LV wall.
A general form of the transversely isotropic pseudoelastic constitutive equation
can be written in terms of the Cauchy stress for describing the characteristics of
passive [4] and active [6] LV wall. The passive constitutive equation is given by
 
* * * Wa * * * * T
r ¼ pI þ 2W1 B þ FNNF ð7Þ
a

and the active constitutive equation for the contractile mechanics of the cardiac
muscle, which is generated in the muscle fiber direction, is given by
   
* * * Wa * * * * T þ * *
r ¼ pI þ 2W1 B þ F  N  N  F þ T a; ½Ca2 m  m ð8Þ
a
*
where r is the Cauchy stress, pðrÞ is a Lagrange multiplier enforcing incom-
*
pressibility derived from equilibrium and boundary conditions, I is the identity
*
tensor, W1 ¼ oW=oI1 ; Wa ¼ oW=oa; B; is the left Cauchy–Green deformation
tensors, T is the active stress which is a function of the fiber stretch
 and calcium
. 
* * * *
concentration, and m denotes the current muscle fiber direction m ¼ F  N a :
W is the polynomial form of the pseudostrain-energy function:

W ¼ c1 ða  1Þ2 þ c2 ða  1Þ3 þ c3 ðI1  3Þ þ c4 ðI1  3Þða  1Þ þ c5 ðI1  3Þ2

ð9Þ
where c1, c2, c3, c4 and c5 are the material parameters.
The equilibrium and boundary conditions satisfying the stress distribution
derived from (7) and (8) are linked to the in vivo strain calculated from the finite
strain analysis for formulating compatibility equations. All six stress distributions
at any instant automatically relate to a compatible strain state with a function of r,
w; k; x0 and w0 as well. c1, c2, c3, c4 and c5 for the constitutive equation were
solved by a parameter estimation technique.
The predicted circumferential stresses were taken to compare, by root mean
square error method, with the circumferential stresses calculated from the Modi-
fied Laplace’s law using the in vivo canine LV anterior and posterior deformations
measured by the tetrahedron-shaped crystals implanted movement (experimental
data from Wei-ning Lee, Ph.D., Columbia University, surgery by M.D. Jie Wang).
The strains were calculated from r, w; k; x0 and w0 derived from the finite strain
analysis which classified radial, circumferential and axial strains by principal
vectors corresponding to principal strains of radial, circumferential and axial
direction, respectively. Thus, the transmural distributions of stresses and strains of
the canine passive and active LV wall during the diastasis and systolic phases,
Stresses and Strains Analysis in the Left Ventricular Wall 39

respectively, at any height of the truncated conical shell from the base to the apex
were estimated from constitutive equations (7) and (8).

3 Results

The transmural distributions of stresses of the canine passive and active LV wall
during the diastasis and systolic phases at any height of the truncated conical shell
from the base to the apex are shown in Fig. 2. The first and second column is
stresses occurred at the end-diastole (ED) and end-systole (ES), respectively.
All six components of the Green strains varied as a function of radius and
depended on the deformation parameters. At end-diastole and end-systole, the
radial strains indicated thinning and thickening and the axial strains also indicated
lengthening and shortening of LV wall, respectively. The circumferential strains
show the muscle fibers tension and compression at end-diastole and end-systole,
respectively. Only three of the components of stress varied monotonically as a
function of radius. The radial stresses ðrrr Þ were changed monotonically between
pressure acting on inner (endocardium) and outer LV wall (epicardium). On the
other hand, the circumferential ðrhh Þ; axial ðrzz Þ and in-plane shearing ðrhz Þ
stresses show the complicated transmural distributions. The highest rrr occurred at
the inner surface and decreased gradually across the LV wall. We observed that the
highest rhh occurred at approximately the middle surface where the maximum
circumferential strain occurred, and also increased from the apex to the base at
end-diastole and vice versa at end-systole. The peak of rhh in the LV wall comes
from the anisotropy of the direction of the muscle fibers [2]. rhz performed the
twisting moments acting through the wall, which the direction of torsion were
opposite at the base and the apex.
The tension and compression, corresponding to the strains, of the stresses
distribution that occurred during the diastasis and systolic phases from the inner to
the outer LV wall and from the base to the apex are shown in Table 1.
The transmural distributions of rrr have the same trend as shown in previous
analyses [2, 13], i.e. the highest and lowest value occur at the endocardium and
epicardium, respectively, and rhh and rzz show a complex distribution across the
LV wall as well. The distribution of rhh at the end-diastole from the base to the
apex is also corresponding with the distribution shown in [13].
In this study, the mechanics of the canine LV wall, therefore, can be estimated
via the stress and strain distributions using the constitutive equation for the passive
LV wall with material the parameters: c1 = -0.11, c2 = -0.13, c3 = 25.7,
c4 = 0.28 and c5 = -13.1 and that for the active LV wall with the material
parameters: c1 = -0.11, c2 = 13.6, c3 = -8.44, c4 = -32.1 and c5 = 0.56, with
the active stress function of the calcium concentration ½Ca2þ  ¼ 0:44; value of
½Ca2þ  at To equals one-half of Tmax ; C50 ¼ 0:78; the sarcomere length
L ¼ 2:31 lm the maximum stress generated in the unloaded length Tmax ¼ 9:5 kPa
and b ¼ 4:41:
40 T. Khamdaeng et al.

Fig. 2 Distributions of a radial stress c circumferential stress and e axial stress during the
diastolic phase and b radial stress d circumferential stress and f axial stress during the systolic
phase of canine LV wall at any height of the truncated conical shell from base to apex (height of
the truncated cone equal to 0 and 6.29 cm refers to apex and base, respectively, in the unloaded
configuration)
Stresses and Strains Analysis in the Left Ventricular Wall 41

Table 1 Stresses distribution during the diastasis and systolic phases from the inner (endocar-
dium) to the outer wall (epicardium) and from the base to the apex

Meaning of arrow directions; was high stress and was low stress

4 Conclusion

Although the LV wall can be better described as a non-homogeneous, anisotropic

and viscoelastic material including residual stresses and strains and a nonzero
external pressure, in this study, we could not measure these data. Thus, the LV
wall was assumed to be homogeneous, elastic and transversely isotropic material
to make the mathematics tractable for the analytical solution. The anisotropy of the
LV wall is observed in the distribution of stresses and strains due to the muscle
fiber directions that depend on the varying orientation of the muscle fiber direc-
tions through the LV wall. In addition, the conical shape is not realistically rep-
resentative of the LV because its real shape is irregular. However, the effects of
geometry of the truncated conical LV wall on the distribution were shown stresses
and strains from the base to the apex properly corresponding with a physiological
LV mechanism.

Acknowledgments This research was supported by The Royal Golden Jubilee Ph.D. Program
(RGJ) under The Thailand Research Fund (TRF), contract number PHD/0243/2548. We are
grateful to Wei-ning Lee, Ph.D., from Columbia University, for all experimental data.

References

1. Guccione, J.M., Costa, K.D., McCulloch, A.D.: Finite element stress analysis of left
ventricular mechanics in the beating dog heart. J. Biomech. 28, 1167–1177 (1995)
2. Humphrey, J.D., Yin, F.C.P.: On constitutive relations and finite deformations of passive
cardiac tissue: II. Stress analysis in the left ventricle. Cir. Res. 65, 805–817 (1989)
3. Streeter, D.D., Ramesh, J.R., Vaishnav, N., et al.: Stress distribution in the canine left
ventricle during diastole and systole. J. Biophys. 10, 345–363 (1970)
42 T. Khamdaeng et al.

4. Humphrey, J.D., Strumpf, R.K., Yin, F.C.P.: Determination of constitutive relation for
passive myocardium: I. A new functional form. ASME J. Biomech. Eng. 112, 333–339
(1990)
5. Humphrey, J.D., Yin, F.C.P.: On constitutive relations and finite deformations of passive
cardiac tissue: I. A pseudostrain-energy function. J. Biomech. Eng. 109, 298–304 (1987)
6. Hunter, P.J., McCulloch, A.D., ter Keurs, H.E.D.J.: Modelling the mechanical properties of
cardiac muscle. J. Biophys. Mol. Biol. 69, 289–331 (1998)
7. Waldman, L.K., Fung, Y.C., Covell, J.W.: Transmural myocardial deformation in the canine
left ventricle: normal in vivo three dimensional finite strains. Cir. Res. 57, 152–163 (1985)
8. Zervatonakis, I.K., Fung-Kee-Fung, S.D., Lee, W.N., et al.: A novel view-independent
method for strain mapping in myocardial elastography: eliminating angle and centroid
dependence. J. Phys. Med. Biol. 52, 4063–4080 (2007)
9. Mirsky, I.: Ventricular and arterial wall stresses based on large deformation analysis.
J. Biophys. 13, 1141–1159 (1970)
10. Yin, F.C.P.: Ventricular wall stress. Cir. Res. 49, 829–842 (1981)
11. Demer, L.L., Yin, F.C.P.: Passive biaxial mechanical properties of isolated canine
myocardium. J. Physiol. 339, 615–630 (1983)
12. Dokos, S., Smaill, B.H., Young, A., et al.: Shear properties of passive ventricular
myocardium. Am. J. Physiol. Heart Circ. Physiol. 283, 2650–2659 (2002)
13. Chaudhry, H.R., Bukiet, B., Davis, A.M.: Stresses and strains in the left ventricular wall
approximated as a thick conical shell using large deformation theory. J. Biol. Syst. 4, 353–
372 (1996)
14. Fung, Y.C.: Biomechanics: Mechanical Properties of Living Tissues. Springer, New York
(1993)
15. Guccione, J.M., McCulloch, A.D., Waldmann, L.K.: Passive material properties of intact
ventricular myocardium determined from a cylindrical model. J. Biomech. 113, 42–55 (1991)
16. Nash, W.M., Hunter, P.J.: Computational mechanics of the heart. J. Elast. 61, 113–141 (2000)
Stress Distribution in Dental Implant
with Elastomeric Stress Barrier

Ali Merdji, Belabess Bachir Bouiadjra, Rajshree Mootanah,

Boualem Serier, Tarik Achour and Noreddine Djebbar

Abstract It is well known that the success of dental implants is heavily dependent
on the initial stability and long-term osseointegration due to optimal stress
distribution in the surrounding bones. This research describes a numerical study
performed with the finite element method due to the commercial code, ABAQUS,
of new dental implant system in order to know the effect of the elastomer material
under an occlusal load on the equivalent von Mises interface stresses induced.

A. Merdji (&)
Department of Mechanical Engineering, University of Mascara,
29000 Mascara, Algeria
e-mail: [email protected]
B. B. Bouiadjra
B. Serier
T. Achour
N. Djebbar
LMPM, Department of Mechanical Engineering,
University of Sidi Bel Abbes, BP 89 Cité Ben M’hidi,
22000 Sidi Bel-Abbes, Algeria
e-mail: [email protected]
B. Serier
e-mail: [email protected]
T. Achour
e-mail: [email protected]
N. Djebbar
e-mail: [email protected]
B. B. Bouiadjra
Department of Mechanical Engineering, College of Engineering,
King Saud University, Riyadh, Saudi Arabia
R. Mootanah
Medical Engineering Research Group at Anglia Ruskin University,
Chelmsford, UK
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 43
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_4,
 Springer-Verlag Berlin Heidelberg 2012
44 A. Merdji et al.

These stresses were compared with the ones provoked by the standardized implant.
The von Mises stress distribution indicated that the stress was maximal around the
top of the implant with varying intensities in the different loading cases. The stress
was highest in the cortical bone at the neck of the implant and lowest in the
cancellous bone. Overall, the novel implant provoked lower interface stresses only
in the cortical bone due to the stress shielding effect of the elastomeric
stress barrier.

Keywords Dental implant
Stress barrier
Elastomer
Stress
Finite element

method

1 Introduction

Stress and strain fields around osseointegrated dental implants are affected by a
number of biomechanical factors, including the type of loading, material
properties of the implant and the prosthesis, implant geometry, surface
structure, quality and quantity of the surrounding bone, and the nature of the
bone–implant interface [1]. As far as implant shape is concerned, design
parameters that primarily affect load transfer characteristics (the stress/strain
distributions in the bone) include implant diameter and the length of the bone–
implant interface, as well as, in the case of threaded implants, thread pitch,
shape, and depth. To increase the surface area for osseous integration, threaded
implants are generally preferred to smooth cylindrical ones [2]. Depending on
bone quality, surface treatments and a thread geometry can significantly
influence implant effectiveness, in terms of both initial stability and the bio-
mechanical nature of the bone–implant interface after the healing process [3].
For this reason biomechanical optimization is an important objective in the
design of dental implants several concepts have been developed, and many
implant types are commercially available in different sizes, shapes, materials,
and surfaces. To analyze the effectiveness and reliability of endosseous
implants, revealing possible risks of implant failure, stress analysis of bone–
implant mechanical interactions is important [4]. In recent years, the finite
element method has been used to investigate the stress distribution within
implant dentistry [3]. This method can be used as an ideal tool to investigate
the functional responses of dental implants in different conditions. It allows the
investigation of the relative merits of different parameters, shapes or designs as
well as offering insight into the internal state of stress in components or
materials within the implant or at the implant–bone interface [5]. The aim of
this study is to make a qualitative comparison, employing finite elements and a
three dimensional model, of the distribution of stresses generated by simulated
loads on two implant system designs in the molar region.
Stress Distribution in Dental Implant with Elastomeric Stress Barrier 45

Fig. 1 Components of the

models

2 Geometrical and Material Models

The modelling consists of using a CAD software to create three-dimensional

models representing the implant systems and mandibular bone respectively
(Fig. 1).
In this study two different types of implant system were compared:
• The conventional implant system is composed primarily of four parts: (a) the
crown, (b) the framework, (d) the implant and (e) the abutment.
• The new implant system with elastomer is composed with the same parts of the
conventional implant system and (c) the elastomer was interposed between the
abutment and the framework.
The mandibular bone was modelled as a cancellous core surrounded by a
cortical layer. The width and height of cortical bone model were 15.8 and
23.5 mm, respectively. The thickness of its upper part was 2 mm.
The materials of the models were considered to be isotropic and linear elastic
(Table 1). The elastomer represented a considerably small component as 0.5 mm
thickness of the implant system and was assimilated into the volume of the
framework.

3 Boundary Conditions

In order to define the boundary conditions, a 3D coordinate system was defined by

three loads in the coronal–apical direction, lingual–buccal direction and mesial–
distal direction. For the boundary conditions, three zones were considered:
• The inferior plane of the mandibular bone was defined as having zero
displacement.
46 A. Merdji et al.

Table 1 Material properties used in analysis

Parts Materials Elastic modulus, E (GPa) Poisson’s ratio Ref.
Crown Feldspathic porcelain 82.8 0.33 [6]
Framework Co–Cr alloy 218 0.33 [7]
Elastomer Silicone 0.006 0.49 [8]
Abutment Titanium 110 0.3 [9]
Implant Titanium 110 0.3 [9]
Mandibular bone Cortical bone 14.5 0.323 [10]
Cancellous bone 1.37 0.3 [5]

Fig. 2 Boundary conditions

• The most coronal plane of the crown was subjected to a load of 3 MPa in either
the lingual–buccal or mesial–distal directions or a load of 10 MPa in coronal–
apical direction.
• The other surfaces were treated as free surfaces, i.e. zero loads.
The vertical load of 10 MPa and horizontal load of 3 MPa were applied at the
central surface in the occlusale face of the crown (Fig. 2). The solid model
resulting from the intersection of implant and mandibular bone represents the
assumption of complete osseointegration, restricting any relative displacement
between implant and bone.

4 Finite Element Model

The mesh of the components is simplistic and consisted of linear tetrahedron

elements with four nodes (Fig. 3). Since the interface of bone–implant experiences
the largest deformations under load, it is necessary to mesh this boundary into
small elements. The implant system and the bone were meshed with increasingly
larger elements as the distance from the interface increases, with the size of
elements in contact with the interface being defined by the elements of the
boundary mesh.
Stress Distribution in Dental Implant with Elastomeric Stress Barrier 47

Fig. 3 Mesh model of: a the

parts of the implant system,
b the global assembly

Fig. 4 Different paths of the

bone-implant interface used
for stress distribution

5 Results

In this section, the von Mises stresses were obtained from the analysis, allowing
the consideration of maximum compressive and tensile stresses, as bone behaviour
under tension and compression is essentially different. Figure 4 shows points
distributed along the implant–bone interface at a cervical, bucco-lingual and a
mesio-distal section used to plot the von Mises stresses variation. Along the paths
shown in the same figure, graphics were generated to make comparisons between
both implant system geometries, displaying maximum and minimum von Mises
stresses for both models under vertical and horizontal loads.
Figures 5, 6 and 7 present a comparison of von Mises stress distribution along
the cross-section of both models for the three different types of loadings. Several
different cases were evaluated for each loading condition.
The largest tensile stresses occurred in the cortical bone in one side loaded
under the larger curvature region of the crown surface in the cervical area
while the highest stresses occurred on the cervical line in mesio-buccal side for
coronal–apical load, in bicco-distal side for lingual–buccal load and in distal side
for distal–mesial load (Figs. 5a, 6a and 7a).
48 A. Merdji et al.

Fig. 5 Comparison of the stress distribution (coronal–apical load)

Fig. 6 Comparison of the stress distribution (buccal–lingual load)

Fig. 7 Comparison of the stress distribution (mesial–distal load)

The conventional implant under coronal–apical load was presented a high com-
pressive peak stress concentration in one side of the cortical bone around the implant
and a smooth distribution along the body of cancellous bone (Figs. 5b and c).
These stresses, decreased in the coronal–apical direction, and sudden a slight ascends
on the curves shows the increase in stresses at the interfaces of cortical–cancellous
bone and at the base of the implant on the same side as those in cortical bone. For the
new implant with elastomer under vertical load (Figs. 5b and c), the stress distri-
bution was quite similar with the conventional implant, however, there is a small
difference in the cortical bone. A similar pattern occurred for horizontal loading
(Figs. 6 and 7), although reaching different values.
Stress Distribution in Dental Implant with Elastomeric Stress Barrier 49

In general, the curves show that the stress distribution at the interface in the
cortical bone of the model with an elastomer was lower than for the conventional
model, with the exception of the stresses in the cancellous bone where the stress
intensity is relatively similar.

6 Discussion

The aim of this study was to provide an analysis between two different geometric
configurations of implant systems, to find the pure effect upon the bone stresses of
prosthesis materials, to know the influence of the elastomer material on the load
transfer to the implant and bone and to compare their biomechanical behavior. For
this reason, it was assumed that all the parameters of both models were identical
except the prosthetic design.
In both models, the extreme stresses in the mandibular bone occur in the layer
of cortical bone adjacent to the neck of the implants. These were due to:
• The evidence of the surface area between the implant and the cortical bone is
much smaller than the surface area between the implant and the cancellous
bone. In addition, the cortical bone is more than ten times stiffer than the
cancellous bone. These are the reasons due to the high stress increments were
found in the cortical bone.
• The intimate contact at the cortical bone and implant interface; the loading
applied to the implant is directly transmitted to the cortical bone.
This suggests that great importance is to be attached to the contact of the
implant with the cortical layer of bone.
In a number of radiologic long-term studies, loaded implants showed typical
bone loss around the implant neck. This agrees well with the results of the present
finite element study, in which the highest stress levels occurred in this very area.
The cervical bone resorption always occurs to accommodate the reformation of a
‘biological width’. Preservation of peri-implant bone height depends on the
magnitude and concentration of stress transmitted to the bone by the implant.
There appears to be an optimal level of stress at which bone resorption is balanced
by apposition. The minimum required load for avoidance of cortical bone loss
appears to have been defined, but the upper limit of the physiological stress range
has not yet been fully investigated.
In the present study, stresses in the new implant were in general lower than in
conventional geometry, indicating that stress-induced bone resorption should not
be more critical in this geometry than in more usual conventional implants. This
fact was expected, as the indication of use of such implant geometry comes not
from the need to reduce stresses, but from occasional anatomic difficulties in the
use of more traditional solutions.
In our model, the larger differences in peak stresses were for horizontal loading;
this increase was also larger for the conventional implant. Interpretation of the
50 A. Merdji et al.

numerical results should take into account that in this study, during mastication,
the horizontal components of the loading are higher than the vertical components
and in parafunction, vertical loads can be dominant, representing a specially
critical situation.

7 Conclusions

Stress analysis of two different geometries, a conventional model and a new model
with elastomer, was performed using the finite element method, leading to the
following conclusions:
• Both studied geometries presented quite similar qualitative stress distributions.
• Stresses in the new implant system with elastomer were in general lower than in
the conventional implant.
• In both geometries stress concentration occurred at one side of the neck.
• High magnitudes stresses in mandibular bone were observed in the cortical area.
• The cancellous bone presented low stress concentration for both geometries.
• The use of prosthetic materials with lower stiffness was capable to diminish or to
delay the loads transmitted to implants and to the bone.

References

1. Koca, O.L., Eskitascioglu, G., Usumez, A.: Three dimensional finite element analysis of
functional stresses in different bone locations produced by implants placed in the maxillary
posterior region of the sinus floor. J. Prosth. Dent. 93, 38 (2005)
2. Papavasiliou, G., Kamposiora, P., Bayne, S., Felton, D.: Three dimensional finite element
analysis of stress distribution around single tooth implants as a function of bony support,
prosthesis type and loading during function. J. Prosth. Dent. 76, 633 (1996)
3. Cochran, D.L.: A comparison of endosseous dental implant surfaces. J. Periodontol. 70, 1523
(1999)
4. Geramy, A., Morgano, S.M.: Finite element analysis of three designs of an implant-supported
molar crown. J. Prosth. Dent. 92, 434 (2004)
5. Ishigaki, S., Nakano, T., Yamada, S., Nakamura, T., Takashima, F.: Biomechanical stress in
bone surrounding an implant under simulated chewing. Clin. Oral. Implant. Res. 14, 97
(2003)
6. Peyton, F.A., Craig, R.G.: Current evaluation of plastics in crown and bridge prosthesis.
J. Prosth. Dent. 13, 743 (1963)
7. Mellal, A., Wiskott, H.W., Botsis, J., Scherrer, S.S., Belser, U.C.: Stimulating effect of
implant loading on surrounding bone: comparison of three numerical models and validation
by in vivo data. Clin. Oral. Implant. Res. 15, 239 (2004)
8. Lídia, C., Ramos, A., Simões, A.: Finite element analysis of a dental implant system with an
elastomeric stress barrier, summer bioengineering conference, Sonesta Beach Resort in Key
Biscayne, Florida, 25–29 June 2003
Stress Distribution in Dental Implant with Elastomeric Stress Barrier 51

9. Van Oosterwyck, H., Duyck, J., Vander Sloten, J., Van Der Perre, G., Naert, I.: Peri-implant
bone tissue strains in cases of dehiscence: a finite element study. Clin. Oral. Implant. Res. 13,
327 (2002)
10. Ashman, R.B., Van Buskirk, W.C.: The elastic properties of a human mandible. Adv. Dent.
Res. 1, 64 (1987)
Biomechanical Behavior Analysis
of the Sap Ascent in Vascular Plants

Jose-Luis Perez-Diaz, Juan-Carlos Garcia-Prada,

Fernando Romera-Juarez and Efren Diez-Jimenez

Abstract A pure mechanical anisotropic model of a tree trunk has been developed
based on the 3D finite element method. It simulates the microscopic structure of
vessels in the trunk of a European beech (Fagus sylvatica) in order to study and
analyse its mechanical behaviour with different configurations of pressures in the
conduits of xylem and phloem. The dependence of the strains at the inner bark was
studied when the sap pressure changed. The comparison with previously published
experimental data leads to the conclusion that a great tensile stress—or ‘‘negative
pressure’’—must exist in the water column in order to achieve the measured strains
if only the mechanical point of view is taken into account. Moreover, the model
can help to design experiments where knowing qualitatively the strains and the
purely mechanical behaviour of the tree is required.

 
Keywords Mechanics of a trunk Finite element analysis Biomechanics Sap 
  
pressure Sap ascent Strains in a trunk Negative pressure Cavitation Water 
column stress

J.-L. Perez-Diaz  J.-C. Garcia-Prada  F. Romera-Juarez  E. Diez-Jimenez (&)

Departamento de Ingeniería Mecánica, Universidad Carlos III de Madrid,
Butarque, 15, 28911Leganés Spain
e-mail: [email protected]
J.-L. Perez-Diaz
e-mail: [email protected]
J.-C. Garcia-Prada
e-mail: [email protected]
F. Romera-Juarez
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 53
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_5,
Ó Springer-Verlag Berlin Heidelberg 2012
54 J.-L. Perez-Diaz et al.

1 Introduction

There is no unresolved topic in plant physiology that has generated more literature
than the antigravitational sap ascent in tall plants. This phenomenon has been
studied from the times of Hales [1]. In the last four decades, most of plant biology
books assume the Tension-Cohesion theory. The Tension-Cohesion theory [2–5]
proposes that the sap is pulled up through the xylem by water tension gradient
generated by water transpiration at leaves. This theory is extensively accepted
because there are evidences of it [6]. Nevertheless, many other physiologists reject
it [7, 8]. It is hard to envisage how the xylem conduits can be treated simply as
isolated pipelines. Moreover, the main problem of this theory is that very large
negative tension must exists in the water column, which implies to bring the water
up to very metastable states [9]. An alternative theory was proposed by Canny to
explain this phenomenon, the so called ‘‘Compensating-Pressure Theory’’ [10].
In some cases, it is in a great disaccord with the Tension-Cohesion theory. Recent
experimental techniques [11] show that considering plants as simple mechanical
systems is a wrong approach and other points of view (osmotic, hydraulic, elec-
trical, etc.) have to be included in order to describe completely their behavior. This
is why there is not any firm conclusion and it is still an open research topic.
In order to design experiments to clarify, at least partially, this controversy, we
have built a pure mechanical simplified FEA (finite element analysis) model of the
xylem tissue of a tree. This FEA model is based on microscopic and macroscopic
properties. Assuming the Tension-Cohesion theory, the proposed model here
relates the measurable strains of the tree with the mechanical interaction between
the sap and the vessels.
The stem of a living tree is known to be continually suffering elastic changes on its
diameter during the day mainly because of the sap ascent [12, 13]. It is commonly
assumed that, according with the Tension-Cohesion theory, periodic strains are
induced in the trunk of the tree by the stress in the water column. In this frame, it is
assumed that the water column stress is directly transmitted to the vessel walls and it
forces the whole trunk to contract. Other mechanisms like temperature, osmotic
potential or secondary growth may also contribute to these diurnal strains [14].
Nevertheless, their contribution is assumed to be much smaller than the first ones.
The phenomenon known as daily diameter changes or diurnal strains has long
been studied [15–17] relating different aspects of the diurnal trees behavior like the
water transport, sugar transport, wood morphogenesis or irrigation scheduling.
In this work we focus on understanding the relationship between the stress in
the water column inside the vessels and the diurnal strains. Several papers have
showed that there are some elastic parameters that relate these strains with the
water potential status [6, 18, 19]. Usually, these works propose some empirical
methods for the determination of these parameters whose main disadvantages are
that they cannot be applicable to different trees and that they are not constant
during the tree growth. Another kind of related works, try to establish general
analytical models to predict the mechanical trees behavior from known water
Biomechanical Behavior Analysis of the Sap Ascent 55

pressure status [20, 21]. Inversely, these last kinds of models would help to
understand the water column pressure directly from strain measurements. It is
accepted by the Tension-Cohesion theory that sap ascent is due exclusively to
transpiration-induced negative pressure gradients up to several megapascals
through continuous water columns from the roots to the foliage. Water under such
negative pressures is extremely unstable, particularly given the hydrophobicity of
the inner xylem walls and sap composition. That is why determining estimated
values for these pressures are so necessary.
We propose a pure mechanical 3D anisotropic model of a tree, based on the
finite element analysis (FEA). Its mechanical behavior will be studied admitting
that there should be a tension or negative pressure on the interfaces xylem-water
column. This is a consequence of the hypothetic negative pressure values of the
water column, according to the Tension-Cohesion theory. These interface tensions
are converted in measurable strains along the xylem and inner bark surfaces. The
model makes it possible to calculate the tangential strains, at the inner bark, or at
any point on the stem, for different values of the water column stresses transmitted
along the sap/vessel interface to the walls of the conduits. Therefore, by measuring
(for instance by extensometry) the strains at the xylem, phloem or bark, the tensile
stress between the sap and the wall of the conduits can be derived. Moreover, it
can be determined whether it follows the same pattern of daily changes as the
internal pressure of the sap, or whether it presents a limit that cannot be surpassed.
In order to answer this question only extreme values, not exact values, need to be
estimated. The aim of this chapter is not to provide an exact comparison of strains
of any particular tree species, but rather an estimation of their magnitudes and
extreme values in order to clarify the mechanical behaviour.
For the verification and comparison of the model results some empirical data
have been used. Particularly, this chapter uses the experimental data found in the
literature [22].

2 Model and Method

We have modelled a European beech (Fagus sylvatica L.). It is an angiosperm with

vessels, fibres and apotraqueal xylem parenchyma, with uneven vessel distribution.
However, in this model, vessels have been distributed evenly, with a diameter of
30 lm, with equal distances between each other. This assumption does not fun-
damentally alter the physical characteristics of the system, rather simplifies the
calculation and is more likely to lead to a convergent result.
An elastic linear anisotropic behaviour of the European beech wood was
assumed, and the values of the elastic constants given by the database of the Swiss
Federal Research Institute of Zurich have been used [23]. ‘‘L’’ denotes the axial
direction (parallel to the vessels), ‘‘R’’ denotes radial direction (perpendicular to
the rings) and ‘‘T’’ means tangent to the rings (Table 1).
56 J.-L. Perez-Diaz et al.

Table 1 Elastic anisotropic properties of the European Beech

Young’s modulus (MPa) Poisson’s ratio Shear elastic modulus (MPa)
EL = 14,000 mLR = 0.450 GLR = 1,640
ER = 2,280 mLT = 0.518 GLT = 1,080
ET = 1,160 mRT = 0.708 GTR = 465

The use of the macroscopic elastic properties for describing the microscopic
properties is actually of limited benefit. The macroscopically uniform wood
material is, in fact, composed of microscopic vessels and a complex material made
by lignin, cellulose and other matters. We have done an extrapolation of the
micrometric properties from the known microgeometry and the macrometric
properties. In spite of the complexity of this material, its mechanical behaviour
was approached as if it were constant and anisotropically known. In this sense, as
the empty vessels do not contribute to the mechanical resistance, the macroscopic
elastic constants will necessarily be smaller than the microscopic ones. By using
the macroscopic constants as an approach for the microscopic ones, a lower limit
for the elastic constants is used. This means that the real stems will be stiffer than
those in this simulation, and therefore real stresses may be larger than those
calculated in this study.
We have used the AlgorÒ software to provide a finite element analyse (FEA).
It is a well known method for solving partial differential equations, particularly
elasto-mechanical equations [24, 25], which has a great number of purposes and is
extensively used in the fields of physics and engineering. The method essentially
reduces the continuous differential equations of an elastic problem to a (large) set
of algebraic equations at certain nodes. These nodes are the vertices of the so-
called finite elements. The elements are only the mathematical representation of
the resistance tensor that transmits forces and displacement between nodes. The
shape and size of the elements must usually be chosen carefully in order to get
accurate results within a reasonable calculation time. In the case of our mechanical
analysis, forces and pressures are set as external loads and displacements are what
we expect to obtain. For that, we can divide the whole tree in elements, but FEA
requires a previous study in order to avoid unnecessary calculations. We have to
take advantages of all problem symmetries, i.e. identical direction of displace-
ments or inverse loads sets, as much as possible. Both hexaedric and tetraedric
solid elements were used, with a total of 5,000 nodes. By combining hexaedric and
thetraedric elements, the precision of the results was improved with a moderate
number of nodes [24].
The trunk was assumed to be a perfect straight cylinder, axially symmetric,
corresponding to a 1 m high stem with a trunk diameter of 10 mm. A transversal
section was taken first, followed by a circular sector. Then, just a circular sector
was modelled on the FEA program, corresponding to an angle of 1.28, which can
be applied to the whole trunk as shown in Fig 1. This simplification is coherent
with the symmetric properties of the problem which are implemented within the
Biomechanical Behavior Analysis of the Sap Ascent 57

Fig. 1 Circular sector corresponding to an angle of 1.28 of a European beech. The whole stem is
completed by symmetry

boundary conditions. The species and the size have been selected in order to fit the
model to the previously mentioned experimental data.
It was assumed that the core of the trunk (heartwood), with a diameter of
2.8 mm, has a much greater stiffness than that of the sapwood. Vessels were
distributed, keeping the density of them between 100 and 200 units per mm2—as
observed in real stems. Although the average diameter of the vessels in an adult
beech ranges from 50 to 90 lm, for a young stem an average diameter of 30 lm
has been estimated (Fig 2).
As we can observe in Fig 3 the circular sector has been pierced by holes
simulating the vessels. The sector was divided in two zones, the xylem vessels part
and the phloem one. The choice of appropriate boundary conditions is the key for
having coherent solutions to the problem. As mentioned, the core of the trunk was
considered much stiffer than the xylem and the phloem and this condition was
implemented with a restriction of all degrees of freedom, except for the axial
displacements (L direction). For the two rectangular surfaces delimiting the
angular sector, T direction displacements were restricted. If we analyse the dis-
placements of two hypothetical adjacent sectors, we realise that points of this
surfaces only could move in the R and L directions. The L displacements are
homogeneous over each the RT plane. Moreover, no torsion was assumed.1 The
outer cylindrical surface (inner bark) had an applied pressure of 0.101 MPa,
corresponding to atmospheric pressure. The upper surface, perpendicular to the L
axis, had a pressure equivalent to the weight of the upper part of the tree plus the
atmospheric pressure: 0.108 MPa. This is equivalent to the weight of half of stem

1
The stems used in the experiments and the ones for comparison were not submitted to either
torsion or strong wind loads.
58 J.-L. Perez-Diaz et al.

Fig. 2 Microscopic structure

of the European beech (Swiss
federal research institute)

Fig. 3 Boundary conditions for the FEA tree model

Biomechanical Behavior Analysis of the Sap Ascent 59

above. Inside the vessels, both xylem and phloem, a pressure (negative or positive)
was applied normal to the interface. Only these two last pressures (xylem water
pressure, PXYLEM and phloem water pressure, PPHLOEM) were varied. The rest of
the boundary conditions and geometries remained the same.
Two kinds of pressures variations were considered. The first aspect considered
was the response of the finite element model of the trunk to changes in positive
pressure, this pressure being the same at xylem and phloem, ranging from 0 to
0.60 MPa. This simulated a pressure imposed experiment like the one of Alméras
et al. [22]. Conversely, a negative pressure was assumed to be present in the xylem
in a range between 0 and -1.00 MPa; with a positive (atmospheric constant)
pressure present at the phloem. The range of values for the water column tension is
in accordance with the Tension-Cohesion theory for sap ascension [26]. In this
model it was assumed that this tensile stress is also applied on the walls of the
vessels by negative water pressures. That means that a perfect adhesion in the
interface is supposed and is not limited.

3 Results and Discussion

Although the model permits to measure all strains at each node position of the
circular sector, only shear have been recorded in order to compare with the
experimental data.
The first part of the study, Fig 4, shows the comparison of the tangent strains in
the inner bark versus positive imposed pressure values. In the first configuration,
when pressure is increased in both xylem and phloem vessels from the state of rest,
we find a linear expansion of the model, producing an increase in the tangent
strains. Despite having only a rough estimation of the elastic coefficients and the
distribution of vessels, the values obtained for the tangential deformation are close
to the experimental data obtained by Alméras cited above: for 0.6 MPa, we
obtained a result of 779 lm/m at the inner bark. The cited experimental work by
Alméras et al. gave 530 lm/m for the same position. We, to a certain degree,
overestimate the sensitivity of the trunk of the stem to the pressure of the sap.
Currently, the focus of this chapter is not on getting a very accurate description of
the microscopic elastic matrix of coefficients, or an exact number or dimension of
the vessels, but an overestimation of its sensitivity. We can then assure that the real
sensitivity of the stems in the referred experiments will be lower than that in our
model. From these results we can conclude that our model is quite realistic. In the
experiment [22] the stems were pressurized in a pressure chamber that involved
the whole stem, that means the xylem and the phloem, having both of them the
same pressurization. Therefore, in our simulations we have set as boundary con-
ditions for the PXYLEM and PPHLOEM the same values.
Once we had checked the model was checked, we have done the second part of
the study, which is applying a negative pressure into the xylem, (PXYLEM \ 0);
with PPHLOEM = PATM. So, we have obtained the new values of tangent strain, in
60 J.-L. Perez-Diaz et al.

Fig. 4 Relation between tangential strains in the inner bark and imposed positive water
pressures. Both model results and empirical data are shown

Table 2 Extrapolated data for the daily diameter changes under artificial light and dark
conditions
Amplitude of the change of tangential strains (lm/m)
Inner bark 637–740
Xylem 200–250

this case negative strains (contraction of the trunk) as expected. In this case, no
measurements of negative tension were found for the case of our species.
Nevertheless, in Almeras’ [20] paper we found daily diameter changes. In Table 2,
we show the extrapolated data taken from it.
In Fig. 5, we show the elastic response of the model to the negative pressure
into the xylem. As we can see the relation is totally linear. We have also added to
the graph the equivalent pressure that the xylem should support for reaching the
daily diameter changes amplitude of Table 2.
It is widely assumed that the sap pressure varies from +0.1 MPa to a negative
value at the xylem sap. This makes the real value of the pressure changes greater
than the estimation of 1.5 MPa. This means that the oscillation of the sap pressure
at the xylem varies from +0.1 MPa to a pressure lower than -1.4 MPa (or a tensile
stress greater than 1.4 MPa). This negative pressure of -1.4 MPa must be trans-
ferred from the sap to the walls of the vessels in order to stress the wood and to
produce the above mentioned strains. In spite of our values are not interpreted as
exact ones, they show the order of magnitude we are dealing with. This order of
magnitude cannot be explained in terms of classical thermodynamics of fluids,
which predicts cavitation to appear just at the vapour pressure of the liquid, in an
order of +0.01 MPa [11]. The apparition of cavitation could be a problem with
Biomechanical Behavior Analysis of the Sap Ascent 61

Fig. 5 Relation between tangential strains at the inner bark and negative water pressure into the
xylem. Estimation of corresponding pressure for daily diameter changes

respect to the cohesion of the water column. This therefore demonstrates the
existence of a real stress between the sap and the wood that completely overpasses
any prediction or explanation of classical mechanics and thermodynamics of fluids.
Moreover, this model can be used to design further experiments in order to
research the inner stresses and interactions among the different components of the
living plants in a non-invasive manner. This model, as it is concluded from this paper,
also considers the anisotropy of the stems as a key factor in the physics of the vascular
plants and can be a useful tool to clarify what remains unknown to the date.

References

1. Hales, S.: Vegetable Staticks 1, (1727)

2. Dixon, H.H., Joly, J.: On the ascent of sap. Ann. Bot. 8, 468–470 (1894)
3. Dixon, H.H., Joly, J.: On the ascent of sap. Philos. Trans. R. Soc. Lond. B 186, 563–576
(1895)
4. Askenasy, E.: Über das Saftsteigen. Verhandlungen des Naturhistorischen–Medizinischen
Vereins Heidelberg. NF 5, 325–345 (1895)
5. Scholander, P.F., Hammel, H.T., Bradstreet, E.D., Hemmingsen, E.A.: Sap pressure in
vascular plants. Science 148, 339–346 (1965)
6. Cochard, H., Forestier, S., Am0 eglio, T.: A new validation of the Scholander pressure
chamber technique based on stem diameter variations. J. Exp. Bot. 52, 1361–1365 (2001)
7. Zimmermann, U., Meinzer, F.C., Benkert, R.: Xylem water transport: is the available
evidence consistent with the cohesion theory. Plant Cell Environ. 17, 1169–1181 (1994)
8. Zimmermann, U., Zhu, J.J., Meinzer, F., Goldstein, G., Schneider, H.: High molecular weight
organic compounds in the xylem sap of mangroves: implications for long-distance water
transport. Botanica Acta 107, 218–229 (1994)
62 J.-L. Perez-Diaz et al.

9. Balling, A., Zimmermann, U., Büchner, K.-H., Lange, O.L.: Direct measurement of negative
pressure in artificial-biological systems. Naturwissenschaften 75, 409–411 (1988)
10. Canny, M.J.: A new theory for the ascent of sap—cohesion supported by tissue pressure.
Ann. Bot. 75, 343–357 (1995)
11. Zimmermann, U., Schneider, H., Wegner, L.H., Haase A.: Water ascent in tall tress: does
evolution of land plants rely on a highly metastable state? New Phytol. 162, 575–615 (2004)
12. Yoshida, M., Yamamoto, O., Okuyama, T.: Strain changes on the inner bark surface of an
inclined coniferous sapling producing compression wood. Holzforschung 54, 664–668 (2000)
13. Yoshida, M., Ikawa, M., Kaneda, K., Okuyama, T.: Stem tangential strain on the tension
wood side of fagus crenata saplings. J. Wood Sci. 49, 475–478 (2003)
14. Daudet, F.-A., Améglio, T., Cochard, H., Archilla, O., Lacointe, A.: Experimental analysis of
the role of water and carbon in tree stem diameter variations. J. Exp. Bot. 56, 135–144 (2005)
15. Goldhammer, D.A., Fereres, E.: Irrigation scheduling protocols using continuously recorded
trunk diameter measurements. Irrig. Sci. 20, 115–125 (2001)
16. Okuyama, T., Yoshida, M., Yamamoto, H.: An estimation of turgor pressure change as one of
the factors of growth stress generation in cell walls. Mokuzai Gakkaishi 41, 1070–1078
(1995)
17. Remorini, D., Massai, R.: Comparison of water status indicators for young peach trees.
Irrig. Sci. 22, 39–46 (2003)
18. Irvine, J., Grace, J.: Continuous measurements of water tension in xylem of trees based on the
elastic properties of wood. Planta 202, 455–461 (1997)
19. So, H.B., Reicosky, D.C., Taylor, H.M.: Utility of stem diameter changes as predictors of
plant canopy water potential. Agron J. 71, 707–713 (1979)
20. Alméras, T., Gril, J.: Mechanical analysis of the strains generated by swater tension in plant
stems. Part I: stress transmission from the water to the plant material at the cell level. Tree
Physiol. 27, 1505–1516 (2007)
21. Alméras, T., Gril, J.: Mechanical analysis of the strains generated by water tension in
plant stems. Part II: stress transmission from the water to the plant material at the cell level.
Tree Physiol. 28, 1513–1523 (2008)
22. Alméras, T., Yoshida, M., Okuyama, T.: Strains inside xylem and inner bark of a stem
submitted to a change in hydrostatic pressure. Trees 20, 460–467 (2006)
23. Schweingruber, F.H., Landolt, W.: The xylem database. Swiss Federal Research Institute
WSL Updated (2005)
24. Akin, J.E.: Finite Element Analysis with Error Estimators. Elsevier Butterworth–Heinemann,
Oxford (2005)
25. Rao, S.S.: The Finite Element Method in Engineering. Elsevier Butterworth–Heinemann,
Oxford (2005)
26. Tyree, M.T., Zimmerman, M.H.: Xylem structure and the ascent of sap. In: Timell, T. (ed.)
Springer Series in Wood Science, 2nd edn, p. 283. Springer-Verlag, Berlin (2002)
Experimental Investigation of the Surface
Tension of Lipid Membranes

Ying Zhang, Hui Fan, Weimin Huang and Yan Chen

Abstract Giant unilamellar vesicles (GUVs), known as the model of living cells,
do not always exhibit spherical shape or have constant surface tension as ordinary
liquids. In this study, the surface tension of the constituting lipid bilayer mem-
branes is experimentally investigated on the adhesion and spreading of GUVs. The
vesicles are prepared through electroformation method which is proven to be
effective in producing defects-free GUVs. The spreading experiment is conducted
on a home-made device assembled from one rectangular silicon sheet and two
transparent polymer sheets. The vertically fixed silicon sheet with a smooth
surface serves as the spreading surface for GUVs. The experimental design is
applicable based on the theoretical analysis that the spreading of micrometer-
scaled GUVs should be controlled by the surface tension of the lipid membrane.
The results obtained from the experimental work show that the pre-aspirated
vesicle would adhere onto the solid surface when the vesicle is extremely close to
the surface and the silicon surface is pre-coated with poly-L-lysine. The surface
tension of the lipid membrane varies with time, leading to the progress of

Y. Zhang  H. Fan  W. Huang  Y. Chen (&)

School of Mechanical and Aerospace Engineering, Nanyang Technological University,
50 Nanyang Avenue, Singapore 639798, Republic of Singapore
e-mail: [email protected]
Y. Zhang
e-mail: [email protected]
H. Fan
e-mail: [email protected]
W. Huang
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 63
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_6,
Ó Springer-Verlag Berlin Heidelberg 2012
64 Y. Zhang et al.

spreading. The spreading finishes in a short time and the vesicle reaches an
equilibrium state with a measurable apparent contact angle with respect to the
solid surface.

1 Introduction

As the constituent of living cells, together with proteins, lipids play an important
role in forming biological membranes. A thin layer of amphipathic lipids spon-
taneously arrange to shield the hydrophobic tails from aqueous environment,
resulting in a continuous lipid bilayer. Both the material characteristics and
mechanical properties of the lipid bilayer contribute to the functioning of
biological membranes and the realization of some cellular processes, such as
signal transduction and membrane trafficking [1–3]. It is of great significance to
study the mechanics of lipid membranes, not only due to its scientific importance
but also because of the requirement from biological applications. Cell adhesion,
which could be noticed in many cellular processes such as cell migration, can be
studied via the adhesion of lipid vesicles on substrates [4]. The mechanics of lipid
membranes would be understood and thereafter applied in biological engineering.
For example, the efficiency of drug delivery via lipid vesicles depends on the
adhesion of lipid membranes on target surfaces [5]. The adhesion between lipid
membrane and scaffolds is one of the criteria to judge whether certain biomaterials
can be used in tissue repair and regeneration [6, 7].
In recent years, giant unilamellar vesicles (GUVs) have attracted more and
more attention in that they are of similar size with living cells and can serve as cell
models in biological studies. It is one of the approaches to investigate the
mechanics of lipid membranes by studying the adhesion of GUVs on substrates
[8, 9]. The other methods include the micropipette aspiration of a single vesicle
[10–13], the spreading of supported-lipid-bilayers [14, 15], and the membrane
response under electric field [16–18].
In this work, the adhesion and spreading of GUVs on substrates is studied in a
similar way with that of liquid droplets. Different from those theoretical work
employing discrete models [19–22], this experimental study is limited within the
scope of continuum physics, in which, the total energy of a condensed matter is
divided into two parts, the energy associated with volume and the energy asso-
ciated with surfaces. The former is often represented by gravity potential energy.
Since the dimension of the GUVs used in this study is in the order of tens of
micrometers, the energy associated with surfaces would be much larger than that
associated with volume. Thus, the gravity of the GUVs can be regarded negligible
[23]. Furthermore, the spreading of GUVs is considered to be driven by conver-
sions among surface energies/surface tensions. This theoretical analysis makes it
possible to experimentally study the GUV adhesion and spreading on vertical
Experimental Investigation of the Surface Tension of Lipid Membranes 65

surface under normal microscope while it cannot provide sufficient information

when studying GUV spreading on horizontal surface.
Referring to the mechanics of biomembranes, some properties have been exten-
sively investigated, such as the bending elasticity and bending rigidity [10, 24–27].
The membrane surface tension has also been studied based on the micropipette
aspiration experiments [25, 28, 29]. However, in those studies, only the change of
surface tension with the aspiration pressure is emphasized. In this work, the surface
tension is examined in its spontaneous spreading process, and the interfacial tension
between the lipid membrane and the solid surface is also estimated.

2 Experiments

2.1 Materials

Synthetic lipid, 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), was purchased

from Avanti Polar Lipids (USA). Sucrose and glucose of the grade for microscopy,
chloroform, bovine serum albumin (BSA), and poly-L-lysine were bought from
Sigma-Aldrich (USA). Deionized water (DI water) was purified by Milli-Q water
system (Millipore, USA).

2.2 Preparation of GUVs

Giant unilamellar vesicles with diameters from 10 to 100 lm were prepared with
the electroformation method [30–32]. Briefly, a stock solution of DOPC was diluted
to 1 mg/ml in chloroform. 10 ll of this solution was deposited onto a piece of ITO
coated glass plate (ITO coating thickness: 1200–1600 Å, resistance: 5–15 X,
Delta Technologies, USA) to form a thin layer of lipid [33]. Lipid solution was
handled with a glass microsyringe (Hamilton, USA) and deposition was done under
nitrogen atmosphere to avoid lipid oxidation. The glass plate with dried lipid was
then put into a vacuum oven and maintained for at least 6 h to completely remove
the organic solvent. After that, the formation device was fabricated by two pieces of
glass plates with the ITO coated surfaces facing each other and a silicone spacer
(SylgardÒ 184 Silicone Elastomer Kit, Dow Corning, USA) separating them. It was
connected to a function generator (Thurlby Thandar Instruments, UK) using
conducting wires. 1–2 ml of sucrose aqueous solution (100 mM) was gently
introduced into the device under an AC electric field (0.2 V peak-to-peak, 10 Hz).
The voltage was immediately raised to 2.0 V after adding the buffer solution, and
maintained for 2 h. The formed vesicles were detached by decreasing the frequency
to 0.5 Hz at the final stage and transferred into a plastic tube for further use.
66 Y. Zhang et al.

2.3 Microscopic Observation

The formation of GUVs was monitored under a fixed-stage upright microscope

(BX51WI, Olympus, Japan). A monochrome CCD camera (QICAM, QImaging,
Canada) was equipped with the microscope to capture images and record videos.
Image Pro Express software (MediaCybernetics, USA) was employed to control
the camera and to process the images.

2.4 Fabrication of Spreading Device

The spreading device was a home-made one which was made of one rectangular
silicon sheet and two transparent polymer sheets. The silicon sheet was cut from
commercial silicon wafer (P type/boron doped, single side polished, Bonda
Technology Pte Ltd, Singapore) and the polymer sheets were obtained from big
cast acrylic sheet (Ying Kwang Acrylic Trading, Singapore). The three sheets were
assembled in T-shape with the silicon sheet perpendicular to the others and the
polymer sheets glued to a piece of glass slide. In this device, the vertically fixed
silicon sheet with polished smooth surface served as spreading surface for GUVs.

2.5 Adhesion and Spreading of GUVs

The solution of GUVs suspension was diluted with glucose aqueous solution
(102 mM) to increase the image contrast. The smooth silicon surface was dipped
into 0.01% poly-L-lysine aqueous solution overnight to become positively
charged. Then the silicon surface was thoroughly rinsed with DI water before use.
The high molecular weight of polylysine (150,000–300,000) is believed to
promote membrane adhesion [34]. Besides, it is known that vesicles tend to adhere
to bare glass. To prevent GUVs adhesion onto glass pipettes, 0.2% (w/v) BSA
aqueous solution was used to coat the pipettes before use [35].
30-50 ll of diluted GUVs solution was added along the pre-coated silicon sheet
under the microscope. A single GUV was selected and aspirated by a micropipette
(VacuTip, Eppendorf, Germany) whose position was controlled by a microma-
nipulator (model 210/360MT-6, The Micromanipulator Company, USA). The
micropipette was connected to a microinjector (model IM-6, Narishige, Japan)
which was used to generate an aspiration pressure.
The aspirated vesicle was transferred to the pre-coated silicon surface until it
gently touched the surface. The reflection of the vesicle on the silicon surface helps
determine the contact surface. Upon touching, the micropipette was retracted and
the GUV spontaneously adhered and spread on the pre-coated silicon surface. The
process was observed under the microscope and recorded by the CCD camera at
the frame rate of 2 fps.
Experimental Investigation of the Surface Tension of Lipid Membranes 67

Fig. 1 GUVs formed from electroformation a–c and hydration method d. a 5 min, b 0.5 h, c 2 h
after adding buffer solution and applying electric field. d 4 h after adding buffer solution

3 Results

3.1 GUVs Formed Under Electric Field

Figure 1 shows the microscopic images of GUVs formed via hydration method and
through electroformation. After 2 h under AC electric field, a lot of GUVs were
produced while there were fewer vesicles even after 4 h in the absence of electric field.
It was observed that vesicles began to form within a few minutes after adding
the buffer solution and applying the AC electric field. The prerequisite for GUVs
formation from dry lipid in aqueous solution is the bilayer separation and bending.
As discussed by Angelova and Dimitrov, external electric fields could facilitate
both [30] and result in population of giant vesicles.

3.2 Dynamic Contact Angles of Adhered Vesicle Spreading

on Pre-coated Silicon Surface

After adhering onto the pre-coated silicon surface, the dynamic spreading process
was recorded by taking images every 0.5 s. The shape of the nonattached part of
68 Y. Zhang et al.

Fig. 2 The vesicle adhered

onto the vertically fixed solid
surface

Fig. 3 Dynamic contact

angles of a vesicle spreading
on the pre-coated silicon
surface. The calculated
equilibrium contact angle was
around 117 degrees

the vesicle is assumed to be a spherical cap with contact radius and height being L
and h respectively. On the images, the dimensions of the vesicle as well as the
contact angle can be measured. Besides, the contact angle could also be calculated
from the contact radius and vesicle height. Figure 2 schematically shows the
adhered vesicle on the vertically fixed solid surface.
During the spreading process, the contact radius increased while the vesicle
height decreased. The contact angle decreased fast within the first 10 s, then
became static, and finally reached the equilibrium state after about 40–50 s. For
the vesicles of the size 15–20 lm as studied in this work, the equilibrium contact
angles fell in the range of 100 – 125 °, and most were from 110 to 120 °. Figure 3
shows the dynamic contact angles of a vesicle spreading on the pre-coated silicon
surface.
When a vesicle touched the smooth silicon surface, the reflection of the vesicle
could be seen, which facilitated the determination of contact surface and the
measurement of vesicle dimensions. Figure 4 shows the image of the adhered
vesicle and its reflection. In the meanwhile, the profile of the vesicle was high-
lighted by fitting its contour with a curve.
Experimental Investigation of the Surface Tension of Lipid Membranes 69

Fig. 4 Adhered vesicle (left

part) and its reflection (right
part)

Fig. 5 The change of vesicle

dimensions with time during
spreading process

The change of contact radius and vesicle height was plotted with time as shown
in Fig. 5. It was consistent with the evolution of contact angles. The contact radius
increased sharply while the vesicle height decreased fast during the first 10 s and
both of them almost did not change after about 40 s.

4 Discussions

4.1 Electroformation of GUVs

In the lipid-water system, there are various lipid phases, such as micelles and
lamellar phases. In the process of GUVs electroformation, dry lipid film firstly
formed on the ITO-coated glass plate. When aqueous solution was added, lipids
70 Y. Zhang et al.

were at the lamellar phase which could exist in different states [36]. Only when the
lipid bilayers were in the disordered fluid phase, vesicles might form. The critical
temperature indicating the transition from ordered gel phase to disordered fluid
phase, the so-called melting temperature, of the DOPC lipid is -20 °C which is
much lower than room temperature (about 22°C). During the experiment, little
disturbance was introduced into the formation device. Besides, the microscope
system was set up on an anti-vibration table to minimize the agitation of samples.
Thus both of the two important requirements to prepare GUVs were fulfilled: the
temperature during the vesicle preparation must be higher than the phase transition
temperature of the lipids; agitation of samples must be prevented [37].

4.2 Adhesion and Spreading of GUVs on Pre-coated

Silicon Surface

In this study, we consider the spreading of lipid vesicles in the similar way
with that of liquid droplets. Within the scope of continuum physics, the
spreading process is driven by the conversions among surface energies, and the
effect of gravity is negligible. The nonattached part of the vesicle is assumed to
be a spherical cap, and the vesicle spread at a constant volume. However, it is
different from a liquid droplet regarding the concept of surface tension. For
liquid droplets, during spreading, the molecules in the body of the droplet could
move to the surface to keep the molecular interactions unchanged, thus the
surface tension a constant at a certain temperature. As for lipid vesicles, the
number of molecules that constitute the bilayer membrane is fixed after the
formation of the vesicle. During spreading, the surface tension may change
with surface area.
Since the surface tension of a closed lipid bilayer without concentration gra-
dients across the membrane is zero, a GUV suspended in aqueous solution should
have zero surface tension by adjusting its surface area to minimize the free energy
[38–40]. To realize the adhesion, the vesicle was pre-aspirated by a micropipette
under a pressure difference imposed by a microinjector and transferred to the close
proximity of the pre-coated silicon surface. This aspiration pressure put the vesicle
membrane under tension and sucked inside the glass micropipette. At equilibrium
state, the membrane surface tension is constant over the entire surface except in the
sharp corner at the opening of the pipette. The relationship between the pressure
difference DP and the surface tension c is derived from the Laplace equation [41]
and shown as,
DPRp
c¼   ð1Þ
R
2 1  Rpv
Experimental Investigation of the Surface Tension of Lipid Membranes 71

Table 1 Liquid surface tension and surface tension components (mN/m)

cd cp c
DI water 21.8 51 72.8
glycerol 37 26.4 63.4
(cd and cp are dispersive and polar components of surface tension; c is the total surface tension)

where Rp is the radius of the pipette and Rv is the radius of the vesicle. The surface
tension changes linearly with the aspiration pressure.
When a vesicle touches the solid surface, it is still subjected to an aspiration
pressure and at its tension state. The vesicle will be pulled further inside the pipette
when increasing the aspiration pressure DP: As long as DP is smaller than DPc
which is a critical aspiration pressure, adhesion will take place, and the vesicle will
be extracted from the micropipette. The corresponding force is called the
‘‘breaking force’’ [42]. Upon adhesion, the vesicle starts to spread on the surface
and reaches its equilibrium state with a relatively high contact angle, he : This
contact angle could be related to the adhesion energy via the Young–Dupre
equation as deduced by minimizing the free energy [43, 44],
W ¼ ce ð1  cos he Þ ð2Þ
where W is the adhesion energy per unit area. As defined,
W ¼ ce þ cSV  cSL ð3Þ

where cSV and cSL are solid/vesicle and solid/liquid interfacial tensions. cSL can be
obtained from the liquid sessile droplet on the same solid surface while cSV is
aimed to be determined from the membrane surface tension. However, cSL cannot
be obtained if the surface tensions of the solid and the liquid are unknown. The
surface tension of 102 mM glucose solution was measured using a tensiometer
(model DCAT11, DataPhysics, Germany) at room temperature while the surface
tension of the pre-coated silicon can be estimated from contact angle data of
selected liquids, such as DI water and glycerol, with known surface tension
components on its surface (Table 1). Both Fowkes [45–48] and Wu [49, 50]
approaches can be applied to obtain the components of solid surface tension. The
total surface tension is the sum of its two components. Besides, the surface tension
can also be examined from the reference table based on the ‘equation of state’
approach [51, 52].
During the vesicle spreading process, the volume of the vesicle is considered as
constant while the surface area may increase slightly. The surface tension could be
related to the area change by a fundamental constitutive equation of fluid bilayers
[24, 53–55],
A  Aa kB T c Kapp
¼ ln þ ð4Þ
Aa 8pkc c0 c
72 Y. Zhang et al.

where kB is the Boltzmann constant, T is the absolute temperature, c0 is a reference

value of the surface tension for each vesicle, A is the actual area of the lipid
vesicle, kc is the bending modulus and Kapp is the apparent area compressibility
modulus. kc and Kapp could be obtained from micropipette aspiration experiment
[12, 25, 28, 32]. Thus the surface tension at every moment during spreading would
be estimated from the area dilation. Consequently, the interfacial tension between
the solid surface and the vesicle as well as the adhesion energy could be
calculated.
The study of the adhesion and spreading of lipid vesicles on solid surface would
contribute to the understanding of some biological phenomena, such as the for-
mation of tissues based on the adhesion among cell membranes. It also has
practical importance in the application of drug delivery with lipid vesicles as drug
carriers.

4.3 Other Phenomena Observed upon Adhesion of Vesicles

Besides spreading, some other phenomena were also noticed. Some vesicles gave
birth to smaller ones while others ruptured after spreading for a short while. Both
are topological changes led by vesicle adhesion.

4.3.1 Budding-Fission

Budding-fission of membranes is the expulsion of small vesicles from a larger one,

which subjects to change in some external parameter or stimulus. As shown in
Fig. 6, small vesicles were extruded from the parent vesicle, which may be due to
the presence of excess surface area. The shapes during budding-fission transitions
should be determined by the membrane configuration which minimizes the total
energy as discussed by other researchers [56–58].
As shown in Fig. 6, during the budding-fission transitions, part of the lipid
membranes burst. Some of the disassembled lipid molecules contributed to the
formation of new vesicles while others accumulated in nearby area as small
fragments. In this process, the lipids were trying to adapt the shape with optimal
curvature to minimize the free energy. At last, the two newborn vesicles separated
from the solid surface and became free vesicles. This transition is governed by the
competition between bending and adhesion energies [43].

4.3.2 Rupture

Another topological change led by adhesion is vesicle rupture. As observed in the

experiments, after adhesion, the nonattached part of the vesicle attained the shape
of a spherical cap. The adhesion energy was related to the membrane surface
Experimental Investigation of the Surface Tension of Lipid Membranes 73

Fig. 6 Microscopic images showing the budding-fission process of an adhered vesicle (left part)
and its reflection (right part). a The moment just before budding was observed, setting as
t = 0. b t = 5 s. c t = 10 s. d t = 15 s The marker in a applies to all the four images

tension via the Young–Dupre equation. With the spreading going on, the surface
area exhibited an increase while the volume was kept constant. Correspondingly,
the surface tension increased till it exceeded a certain critical value, causing the
membrane to rupture. The bursting of vesicles happened in an extremely short
time, which made the characterization of this dynamic process almost impossible
under optical microscope.
Rupture strength, quantified by the level of tension at the maximum area
stretch, is dependent on the composition of lipids, the temperature, and the tension
loading rate [59, 60].

5 Concluding Remarks

In this work, giant unilamellar vesicles prepared from DOPC lipid via electro-
formation are employed to experimentally study the surface tension of lipid
membranes. Upon touching, most of the testing vesicles adhere on the pre-coated
surfaces, followed by the spreading similar to that of liquid droplets.
74 Y. Zhang et al.

The nonattached part of the vesicle is assumed to keep as a spherical cap during
spreading. The dimensions of the adhered vesicle change in the way that the
contact radius increases and the vesicle height decreases with time. The contact
angle between the vesicle and the solid surface decreases correspondingly. The
membrane surface tension increases with the area dilation, and the relationship
between the surface tension and the adhesion energy at the equilibrium state could
be represented by the Young–Dupre equation with the contact angle as parameter.
Furthermore, the interfacial tension between the lipid membrane and the solid
surface could be estimated from the membrane surface tension.
This experimental study provides a possible way to investigate the surface
tension of lipid membranes during spreading process, but the estimation from the
surface area change is limited by the assumption that the surface tension is
isotropic.
Besides spreading, budding-fission and rupture of adhered vesicles are also
observed for the testing vesicles of the same lipid composition and similar size.
This may be attributed to the pre-aspiration history which results in different
surface area dilation, leading to different surface tension before adhesion.
Regarding the surface tension of lipid membranes, it has already been exten-
sively investigated by micropipette aspiration method. Here, we only want to look
at this problem from another angle and study it from a familiar vesicle behavior,
spontaneous adhesion and spreading. Though not all the expected results have
been obtained, the method should still have significance in explaining some
cellular behaviors and in the bioengineering applications.

Acknowledgements The first author would like to thank Dr. Guillaume Tresset for teaching her
basic experimental techniques in vesicle preparation, and Dr. Kaori Kuribayashi for sharing her
experience in electroformation of GUVs.

References

1. Nikolelis, D.P., Hianik, T., Krull, U.J.: Biosensors based on thin lipid films and liposomes.
Electroanal 11(1), 7–15 (1999)
2. Simons, K., Vaz, W.L.C.: Model Systems, lipid rafts, and cell membranes. Annu. Rev.
Biophys. Biomol. Struct. 33(1), 269–295 (2004)
3. Mukherjee, S., Maxfield, F.R.: Membrane domains. Annu. Rev. Cell. Dev. Biol. 20(1),
839–866 (2004)
4. Boulbitch, A., Guttenberg, Z., Sackmann, E.: Kinetics of membrane adhesion mediated by
ligant-receptor interaction studied with a biomimetic system. Biophys. J. 81, 2743–2751
(2001)
5. Guedeau-Boudeville, M.-A., Jullien, L., Meglio J.-M., D.: Drug delivery: piercing vesicles by
their adsorption onto a porous medium. Proc. Natl. Acad. Sci. USA 92, 9590–9592 (1995)
6. Hersel, U., Dahmen, C., Kessler, H.: RGD modified polymers: biomaterials for stimulated
cell adhesion and beyond. Biomaterials 24, 4385–4415 (2003)
7. Shoichet, M.S.: Polymer scaffolds for biomaterials applications. Macromol 43, 581–591
(2010)
Experimental Investigation of the Surface Tension of Lipid Membranes 75

8. Bernard, A.-L., Guedeau-Boudeville, M.-A., Sandre, O., et al.: Permeation through lipid
bilayers by adhesion of giant vesicles on decorated surfaces. Langmuir 16, 6801–6808 (2000)
9. Bernard, A.-L., Guedeau-Boudeville, M.-A., Jullien, L., et al.: Strong adhesion of giant
vesicles on surfaces: dynamics and permeability. Langmuir 16, 6809–6820 (2000)
10. Heinrich, V., Waugh, R.E.: A piconewton force transducer and its application to
measurement of the bending stiffness of phopholipid membranes. Ann. Biomed. Eng. 24,
595–605 (1996)
11. Olbrich, K., Rawicz, W., Needham, D., et al.: Water permeability and mechanical strength of
polyunsaturated lipid bilayers. Biophys. J. 79, 321–327 (2000)
12. Henriksen, J.R., Ipsen, J.H.: Measurement of membrane elasticity by micro-pipette
aspiration. Eur. Phys. J. E 14, 149–167 (2004)
13. Colbert, M.-J., Raegen, A.N., Fradin, C., et al.: Adhesion and membrane tension of single
vesicles and living cells using a micropipette-based technique. Eur. Phys. J. E 30, 117–121
(2009)
14. Nissen, J., Gritsch, S., Wiegand, G., et al.: Wetting of phospholipid membranes on
hydrophilic surfaces-concepts towards self-healing membranes. Eur. Phys. J. B 10, 335–344
(1999)
15. Nissen, J., Jacobs, K., Radler, J.O.: Interface dynamics of lipid membrane spreading on solid
surfaces. Phys. Rev. Lett. 86(9), 1904–1907 (2001)
16. Niggemann, G., Kummrow, M., Helfrich, W.: The bending rigidity of phosphatidylcholine
bilayers: dependences on experimental method, sample cell sealing and temperature.
J. Physique II 5, 413–425 (1995)
17. Vlahovska, P.M., Gracia, R.S., Aranda-Espinoza, S., et al.: Electrohydrodynamic model of
vesicle deformation in alternating electric fields. Biophys. J. 96, 4789–4803 (2009)
18. Isambert, H.: Understanding the elecetroporation of cells and artificial bilayer membranes.
Phys. Rev. Lett. 80(15), 3404–3407 (1998)
19. Smith, A.-S., Seifert, U.: Effective adhesion strength of specifically bound vesicles. Phys.
Rev. E 71, 061902 (2005)
20. Dimitrievski, K., Kasemo, B.: Influence of lipid vesicle composition and surface charge
density on vesicle adsorption events: a kinetic phase diagram. Langmuir 25(16), 8865–8869
(2009)
21. Dimitrievski, K.: Deformation of adsorbed lipid vesicles as a function of vesicle size.
Langmuir 26(5), 3008–3011 (2010)
22. Dimitrievski, K.: Influence of lipid-bilayer-associated molecules on lipid-vesicle adsorption.
Langmuir 26(8), 5706–5714 (2010)
23. Fan, H.: Liquid droplet spreading with line tension effect. J.Phys. Condens. Matter 18,
4481–4488 (2006)
24. Evans, E., Rawicz, W.: Entropy-driven tension and bending elasticity in condensed-fluid
membranes. Phys. Rev. Lett. 64(17), 2094–2097 (1990)
25. Rawica, W., Olbrich, K.C., Mclntosh, T., et al.: Effect of chain length and unsaturation on
elasticity of lipid bilayers. Biophys. J. 79, 328–339 (2000)
26. Niggemann, G., Kummrow, M., Helfrich, W.: The bending rigidity of phosphatidylcholine
bilayers: dependences on experimental method, sample cell sealing and temperature.
J. Physique II 5, 413–425 (1995)
27. Gracia, R.S., Bezlyepkina, N., Knorr, R.L., et al.: Effect of cholesterol on the rigidity of
saturated and unsaturated membranes: fluctuation and electrodeformation analysis of giant
vesicles. Soft Matter 6, 1472–1482 (2010)
28. Hung, V., David, E., and Marjorie, L.: Interfacial tension effect of ethanol on lipid bilayer
rigidity, stability, and area/molecule: a micropipet aspiration approach. Langmuir 18,
8988–8995 (2002)
29. Henriksen, J.R., Ipsen, J.H.: Measurement of membrane elasticity by micro-pipette
aspiration. Eur. Phys. J. E 14, 149–167 (2004)
76 Y. Zhang et al.

30. Angelova, M.I., Dimitrov, D.S.: Liposome electroformation. Faraday. Discuss. Chem. Soc.
81, 303–311 (1986)
31. Angelova, M.I., Soleau, S., Meleard, P.H., et al.: Preparation of giant vesicles by external AC
electric fields. Kinet Appl. Progress Colloid & Polym Sci 89, 127–131 (1992)
32. Fa, N., Lins, L., Courtoy, P.J., et al.: Decrease of elastic moduli of DOPC bilayers induced by
a macrolide antibiotic, azithromycin. Biochim. Biophys. Acta 1768, 1830–1838 (2007)
33. Kuribayashi, K., Tresset, G., Coquet, P.H., et al.: Electroformation of giant liposomes in
microfluidic channels. Meas. Sci. Technol. 17, 1–6 (2006)
34. Huang, W.M., Gibson, S.J., Facer, P., et al.: Improved section adhesion for
immunocytochemistry using high molecular weight polymers of L-lysine as a slide
coating. Histochemistry 77(2), 275–279 (1983)
35. Akashi, K-I., Miyata, H., Itoh, H., et al.: Formation of giant liposomes promoted by divalent
cations: critical role of electrostatic repulsion. Biophys. J. 74, 2973–2983 (1998)
36. Heimburg, T.: Thermal Biophysics of Membranes. Wiley-VCH Verlag (2007)
37. Bagatolli, L.A., Parasassi, T., Gratton, E.: Giant phospholipids vesicles: comparison among
the whole lipid sample characteristics using different preparation methods A two photon
fluorescence microscopy study. Chem. Phys. Lipids 105, 135–147 (2000)
38. Brochard, F., Gennes, P.G., Pfeuty, P.: Surface tension and deformations of membrane
structures: relation to two-dimensional phase transitions. J. Physique 37, 1099–1104 (1976)
39. Jahnig, F.: What is the surface tension of a lipid bilayer membrane? Biophys. J. 71,
1348–1349 (1996)
40. Puech, P.-H., Brochard-Wyart, F.: Membrane tensiometer for heavy giant vesicles. Eur. Phys.
J. E 15, 127–132 (2004)
41. Kwok, R., Evans, E.: Thermoelasticity of large lecithin bilayer vesicles. Biophys. J. 35,
637–652 (1981)
42. Brochard-Wyart, F., Gennes, P.-G.: Unbinding of adhesive vesicles. Comptes Rendus
Physique 4, 281–287 (2003)
43. Seifert, U., Lipowsky, R.: Adhesion of vesicles. Phys Rev A 42(8), 4768–4771 (1990)
44. Lipowsky, R., Seifert, U.: Adhesion of membranes: a theoretical perspective. Langmuir 7(9),
1867–1873 (1991)
45. Fowkes, F.M.: Attractive forces at interfaces. Ind. Eng. Chem 56(12), 40–52 (1964)
46. Fox, H.W., Zisman, W.A.: The spreading of liquids on low-energy surfaces. Hydrocarbon
surf. J. Colloid Sci III 7(4), 428–442 (1952)
47. Fowkes, F.M.: Surface effects of anisotropic london dispersion forces in n-alkanes. J. Phys.
Chem 84(5), 510–512 (1980)
48. Owens, D.K., Wendt, R.C.: Estimation of the surface free energy of polymers. J. Appl.
Polym. Sci. 13(8), 1741–1747 (1969)
49. Wu, S.: Surface and interfacial tensions of polymer melts. I. Polyethylene, polyisobutylene,
and polyvinyl acetate. J Colloid Interface Sci. 31(2), 153–161 (1969)
50. Wu, S.: Calculation of interfacial tension in polymer systems. J Polym Sci. Part C 34(1),
19–30 (1971)
51. Spelt, J.K., Neumann, A.W.: Solid surface tension: the equation of state approach and the
theory of surface tension components. Theor concept consid. Langmuir 3, 588–591 (1987)
52. Neumann, A.W., Absolom, D.R., Francis, D.W., et al.: Conversion tables of contact angles to
surface tensions. Sep Purif Methods 9(1), 69–163 (1980)
53. Helfrich, W.: Elastic properties of lipid bilayers: theory and possible experiments.
Z. Naturforsch. 28, 693–703 (1973)
54. Helfrich, W., Servuss, R.-M.: Undulations, steric interaction and cohesion of fluid
membranes. IL Nuovo Cimento 3D(1), 137–151 (1984)
55. Evans, E., Rawicz, W.: Elasticity of ‘‘fuzzy’’ biomembranes. Phys. Rev. Lett. 79(12),
2379–2382 (1997)
56. Lipowsky, R.: The conformation of membranes. Nature 349, 475–481 (1991)
57. Wiese, W., Harbich, W., Helfrich, W.: Budding of lipid bilayer vesicles and flat membranes.
J Phys Condens Matter 4, 1647–1657 (1992)
Experimental Investigation of the Surface Tension of Lipid Membranes 77

58. Miao, L., Fourcade, B., Rao, M., et al.: Equilibrium budding and vesiculation in the curvature
model of fluid lipid vesicles. Phys. Rev. A 43(12), 6843–6856 (1991)
59. Evans, E., Heinrich, V., Ludwig, F., et al.: Dynamic tension spectroscopy and strength of
biomembranes. Biophys. J. 85, 2342–2350 (2003)
60. Rawicz, W., Smith, B.A., Mclntosh, T.J., et al.: Elasticity, strength, and water permeability of
bilayers that contain raft microdomain-forming lipids. Biophys. J. 94, 4725–4736 (2008)
Finite Element Quantification
of the Compressive Forces Induced
by Keratinocyte on a Liquid Crystal
Substrate

C. F. Soon, M. Youseffi, P. Twigg, N. Blagden and M. C. T. Denyer

Abstract The traction force of keratinocytes plays a crucial role in sealing a

wound. A new method of transducing keratinocyte traction forces in the form of
compression on the surface of a cholesteryl ester liquid crystal substrate has been
developed. To quantify the compressive force induced by the keratinocyte via the
focal adhesion on the liquid crystal substrate, the finite element method was
employed. The phase displayed by the surface of the liquid crystals was studied
using cross-polarized microscopy. Physical properties of the liquid crystal,
dimensions of the focal adhesions and lateral displacement were determined using
Atomic Force Microscopy (AFM) based nano-indentation, immunofluorescence
staining and cell relaxation techniques, respectively. Traction forces formed
between a pair of focal adhesions of a cell and a liquid crystal substrate were
examined in a 3D model via the inclusion of the physical parameters of the liquid

C. F. Soon (&)  N. Blagden  M. C. T. Denyer

School of Pharmacy, University of Bradford, Bradford BD7 1DP, UK
e-mail: [email protected]
N. Blagden
e-mail: [email protected]
M. C. T. Denyer
e-mail: [email protected]
M. Youseffi  P. Twigg
School of Engineering, Design and Technology, Medical Engineering,
University of Bradford, Bradford, BD7 1DP, UK
e-mail: [email protected]
P. Twigg
e-mail: [email protected]
C. F. Soon
Faculty of Electrical and Electronic Engineering, University of Tun Hussein Onn
Malaysia, 86400 Batu Pahat, Johor, Malaysia
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 79
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_7,
Ó Springer-Verlag Berlin Heidelberg 2012
80 C. F. Soon et al.

crystal in a linear static stress analysis based on the Finite Element Method (FEM).
The Young’s modulus of the linear viscoelastic liquid crystal surface was deter-
mined at 108 ± 20 kPa and the Poisson’s ratio of the liquid crystals was assumed
to be 0.49, close to that of the rubber. Vinculin immuno-staining indicated that
focal adhesion related to the accumulations of vinculin in cells cultured on the
liquid crystals were 1.03 ± 0.4 lm in length. The relaxation of cell in releasing
the axial deformation on the surface of the liquid crystals provided a means of
determining the lateral displacement of the liquid crystal induced by the
compressive force applied via the focal adhesions. This result also confirmed the
use of a compression model for the focal adhesion-liquid crystals interface. The
model produced compressive forces in the range of 3–38 nN per focal adhesion.
This is comparable to the forces reported in previous studies.

 
Keywords Finite element method Keratinocyte Compressive force  Liquid
 
crystals Static analysis Atomic force microscopy

1 Overview on Cells Contractility and Traction Forces

Measurement Techniques

Contractility of cells regulates important functionalities of cells such as

cytokinesis, migration and wound healing. Actin filaments are one of the main
cytoskeletons components that function by contraction to maintain the structure of
a cell. Along these actin filaments, actin and myosin II proteins interact and, thus,
enable the contractility mechanism. The contractility of these filaments allows
cells to actively re-structure their cell body to accommodate to the stiffness of the
surrounding environment. When cells are grown on a glass cover slip, a broadly
spread morphology is acquired by cells to sustain their bodies on the stiff substrate.
For cells attached to a hard surface, isometric contracted and parallelly aligned
stress fibers are the dominant feature of the cytoskeleton. Some studies [1, 2] have
shown that cells are very sensitive to their interface tension and this can be seen
when tension was applied to a cell’s membrane in which the cell rapidly generate
tension stresses fibers at that localized region. A direct visualization of these
stresses generated within the cells is possible if cells are grown on a flexible
silicone substrate [3]. The effect of the stresses that were transduced in the form of
wrinkles on the silicon substrate is an expression of the contractile forces gener-
ated internally. The wrinkles disappeared rapidly when cells were detached from
the soft substrate. During contraction of a cell on a surface, the cell must anchor
themselves to anchorage sites or focal adhesions where the contraction forces can
be transmitted to the extracellular matrix (ECM). Consequently, focal adhesions
are stimulated and develop into clusters in responding to the onset of the
contractility [4]. Reversible disassembly of the focal adhesions results in a
Finite Element Quantification of the Compressive Forces 81

disruption in the contractility of cells [5]. A higher recruitment of stress fibers

increased the size of the focal adhesions and hence induced higher contractile
forces [6]. A focal adhesion is a group of macro molecules consisting of paxillin,
vinculin, talin and integrins that are linked to the actin filament at one end and the
other end to the ECM. Among these protein molecules, vinculin is the major
protein molecule at the sub-membrane plaque [7]. Both focal adhesions and stress
fibers are regulated by the monomeric GTPase Rho, which functions in controlling
the acto-myosin based contractility [8].
The motility of cell is closely associated with the contractility of a cell. During
locomotion, the focal adhesions at two ends of an actin filament are able to attach
and re-attach to a new adhesion site by contracting and relaxing the actin filament.
Locomotion is an important functionality of keratinocyte which enables the cells
to migrate over a wound bed to form a monolayer or epithelial sheet. The onset of
the migration of the keratinocyte which is relative to the increase of contraction
and traction forces is triggered by growth factors being released upon wounding
[9]. In closing a wound, epithelial cells contract to form a strongly bonded barrier
at the edges of the wound [10]. The reinforcement of the tissue integrity at the
wound edges is achieved by assembling the circumferential stress fibers bundles
that are coupled via the adherent junctions from cell to cell at the connective tissue
[10, 11].
With the aim of quantifying the traction force of a single cell, various
techniques based on soft substrates have been developed [3, 12–15]. Continuum
substrate stretching is a commonly applied technique to measure cell surface
forces and often employs materials such as silicone rubber, polyacrylamide
(PAA), polydimethylsiloxane (PDMS) and collagen gel. Buckling effects on
these substrates were used to quantify the traction force induced by a cell. Most
of these techniques developed were targeted at highly motile fibroblasts that are
able to induce high traction forces. The study of the traction force exerted by a
human keratinocyte and epithelial cell is limited [16]. Two of the very few
examples involved examining fish Keratocytes [13] and Madin-Darby canine
kidney epithelial cell [16]. Keratocytes have a unique migration pattern which is
distinctly different from the fibroblasts [13, 17]. The direction of the locomotion
is perpendicular to the lateral compressive force seen along a broadly extended
cell body on a silicon substrate. The release of the compressive forces at the
edge of the lamellipodium allows the cell to retract and propel forward. Basi-
cally, on a continuum, traction of a cell is formed by compression or ‘‘pinching’’
mechanism in a lateral direction and sheared centripetally. This mechanism is
very different from the mechanisms seen in a cell attached to patterned silicon
micro-posts, in which, traction of single focal adhesion is expressed by shearing in an
axial direction [16, 18]. The technique based on silicon or PDMS sheet needs to be
improved because: (1) The Young’s modulus of these substrates are in the range of
MPa to GPa that may be interpretated by the cells as pathological tissue [19, 20].
Conversely, the mechanical properties for soft tissues are between a few Pa to a few
hundred kPa [20, 21]. (2) The most intense indentation did not correlate with the
location of compressive sites and it was difficult to identify isolated regions of
82 C. F. Soon et al.

compression [18]. (3) Soft substrates that used fluorescent markers to trace the changes
of strain in correlation with the cells movement suffers from high signal to noise ratio
[18]. Furthermore, the determination of strain requires a reference image of surface
patterns before deformation. To discriminate smaller forces exerted by the focal
adhesion down to a few nano-Newton (nN) or even pico-Newton (pN), micro-
patterned cantilevers or silicon substrates can be used. However, micro-patterned soft
substrates used to promote contact guidance for cells strongly alter the behavior of the
cells and, thus, could impact on cell force generation [22, 23]. Hence, the responses of a
cell on patterned substrate may not be as natural as on a continuum. Furthermore, this
technique requires refinement in the issue of spatial resolution, compliance of
substrate, density of proteinscoating and biophysical compatibility to the tissue or cell.
Soft substrates in the form of a continuum is still a more natural and reliable
environment as an interface to the cells.

2 Introduction to Liquid Crystals as a Keratinocyte Force

Transducer

The benefit of developing a means to measure traction force of keratinocyte is not

just restricted to studying the locomotion pattern of a cell but also with an aim to
provide a real time monitoring tool to investigate the response of the cell to a
variety of growth factors which are important pre-cursors in wound healing. We
report a novel measurement technique to transduce the localized cell’s traction
force through an elastic deformation formed on the surface of a liquid crystal
substrate. The objective of the current study is to develop a quantification method
of traction forces for a single cell using a novel cell force transducer based on the
cholesteryl ester based liquid crystals. Transducing the traction forces of a single
non-motile keratinocyte is challenging because the forces generated may be lower
than highly motile cells and require a highly sensitive soft substrate. In addition,
the coating of synthetic ECM proteins on the transducer such as fibronectin,
laminin or collagen should be excluded to avoid the argument about the different
responses of cells to different type of adhesion proteins and different coating
densities. This is based on the ability of the keratinocytes to generate self-derived
ECM proteins in the culture [24, 25] and the affinity that cells show for liquid
crystals [26]. The advantages in using wrinkles or deformation to inform on the
traction force have been indentified [13]. Besides providing high spatial and
temporal resolution, force measurement can be done for a single image in real
time. However, some findings showed that the mechanical properties of crossed
link polymers such as silicon or PDMS is dominated by viscous behavior at low
shear rate when characterized in rheology [27, 28]. Hence, silicon rubber based
cell force sensors may not response elastically to the exerted force due to low shear
rate which is usually characterized by motility of a cell. Liquid crystals have the
potential to overcome this shortcoming because: (1) Liquid crystals molecules are
Finite Element Quantification of the Compressive Forces 83

not bound by cross-linkers like those of the polymers used previously. Hence, the
mesogens that are arranged compactly in long-range order could provide high
sensitivity to localized traction forces of a cell. (2) Cholesteryl ester based liquid
crystals are biocompatible and do not require coating of ECM proteins to facilitate
cells adhesion [26]. (3) The liquid crystals used are profiled by rheology as a
linearly elastic material at low shear rate \1 s-1 [29] and within a shear strain of
10% [30].
In previous studies [13, 14], a cell generated force was quantified by multi-
plying the length of the wrinkles to the stiffness of the silicon expressed in nN/lm,
in which, the wrinkle is a secondary effect due to the lateral compression of focal
adhesions. Unlike previous work, we consider the lateral displacement of the
liquid crystals which induced an axial deformation (wrinkle) and calculate the
applied force (F) required in inducing that displacement. To achieve this objective,
cytochalasin-B was used to relax the cell and the lateral displacement on the
surface of the liquid crystal substrate can be determined. Under linear static
analysis, a three dimensional (3D) model incorporated with the Finite Element
Method (FEM) was applied to quantify the applied load of the focal adhesion to
the liquid crystal substrate. The 3D model allowed us to study the compression
pattern on the liquid crystals induced by the focal adhesions. The dimensions of
the focal adhesion for keratinocytes attached to the liquid crystal substrate were
determined from the micrographs of the immunofluorescence staining. Prior to
determining the Young’s modulus of the liquid crystals, which is required as an
input parameter for the model of the liquid crystal substrate, the phase displayed
by the surface of the liquid crystals after immersing in the cell culture media was
investigated in a cross-polarized microscope. Young’s modulus of the surface of
the liquid crystals was determined by using an Atomic Force Microscopy (AFM)
nano-indentation system. The shear modulus was obtained from a published
rheology study of the cholesteryl ester liquid crystals [26].

3 Quantification of Compressive Force Induced

by a Keratinocyte

3.1 An investigation on the Liquid Crystal Phase

Cholesteryl ester liquid crystal coated substrates (CLC) were prepared as reported
previously [29]. The density of the liquid crystals in the cholesteric phase was
determined at 8.49 g/mm3. The liquid crystal in its isotropic phase (heated up to
melting temperature 70°C) was spread at a thickness of approximately
100–200 lm on the petri dish using a cell scraper (Corning, 6101). Upon cooling
to the room temperature at 25°C, the liquid crystal returned to its cholesteric phase.
Subsequently, 6 ml of RPMI-1640 cell culture media (Sigma Aldrich) was added
to the petri dish and the substrates were incubated at 37°C in an incubator.
84 C. F. Soon et al.

The phase of the liquid crystal surface to which the cells were attached was
investigated using a phase contrast and Zeiss Axioplan2 polarizing microscope at
25x and 40x magnification, respectively. The micrographs of the liquid crystal
phase were captured by using an Axiocam MRc5 digital camera bundled to the
AxioVision 4.6 software.

3.2 AFM Nano-Indentation on the Surface of the Liquid Crystal

Substrate

Liquid crystal coated substrates were prepared as described in Sect. 3.1 and cultured
in petri dishes containing cell culture media at 37°C in an incubator. The objective
of this experiment was to determine the stiffness (S) and Young’s modulus (E) of
the surface of the liquid crystal in the fluid using an AFM based nano-indentation
technique. After 24 h of incubation, the petri dish containing the liquid crystal
coated substrates were placed in a Hysitron Triboscope nano-mechanical testing
system equipped with Hysitron Triboscope 3.5 software to obtain the force
displacement curve measured in the culture medium. A Berkovich pyramidal
indenter with a tip of 200 nm in radii was immersed in the fluid (Fig. 1a). A loading
rate of 750 lN/s and a hold period of 3 s were applied during the measurements.
The hold time compensated for the creep of the material. During the measurement
using AFM based nano-indentation, careful consideration was made to take into
account the fluid buoyancy, hold time and re-cycling indentations in the
measurements. Averagely, four to six measurements per substrate were taken at
various sites of indentation. The data from the force–displacement for each
indentation was collected and plotted using the Statistical Package for Social
Sciences (SPSS) software. The contact stiffness, S, of the liquid crystals was
determined from the unloading curve using the power law fits [31–33],
dp  m1
S¼ ðh ¼ hmax Þ ¼ Am  hmax  hf ð1Þ
dh
where p is the loading force in lN, hmax and hf are the maximum displacement and
final unloading depth in nm, respectively. A and m are the curve fitting parameters
determined empirically by the Triboscope software.
The contact depth, hc as shown in Fig. 1b was calculated from the force-
displacement data by including the geometric constant, e = 0.75 [33] in the
following equations,
hc ¼ hmax  hs ð2Þ
Pmax
hs ¼ e ð3Þ
S
Finite Element Quantification of the Compressive Forces 85

Fig. 1 The configuration of a loading and b unloading the indenter to the surface of the liquid
crystal which is immersed in the cell culture media and the associated indenting depth
parameters: the contact depth (hc), perimeter contact depth (hs) and maximum loading depth
(hmax)

where hs is defined as the perimeter contact depth and maximum loading force,
Pmax was determined from the peak of the force-displacement curve.
Then, the contact area for an ideal Berkovich indenter was established by [33]

Area ¼ 24:5h2c ð4Þ

Once the contact area was estimated, the reduced modulus, Er can be
determined,
pffiffiffi
S p
Er ¼ pffiffiffiffiffiffiffiffiffiffi ð5Þ
2 Area
The reduced modulus, Er is a measurement with contributions from both the
specimen and the indenter, therefore, the following relationship [32] was used to
determine the Young’s modulus of the liquid crystal.

1 1  v2 1  v2i
¼ þ ð6Þ
Er E Ei
where E and v are the elastic modulus and Poisson’s ratio of the test specimen.
E and v with subscript i are for the indenter of the same parameter definitions. The
Poisson’s ratio for liquid crystals was set to 0.49, close to that of the rubber. The Ei
and vi for the diamond Berkovich pyramidal indenter are 1,141 GPa and 0.07 [31],
respectively. This experiment was repeated in triplicates and the Young’s modulae
determined for all the measurements were averaged.

3.3 Determining the Size of Vinculin Accumulations Using

an Immunofluorescence Staining Technique

Immunofluorescence staining techniques were used to profile the size of the

vinculins accumulations associated with focal adhesions. An accumulation of
86 C. F. Soon et al.

vinculin consists of 2–3 vinculins [4]. Human keratinocyte cell line (HaCaT) was
prepared based on the standard cell culture procedures as described previously
[26]. The cells were maintained in RPMI-1640 media supplemented with fetal calf
serum, L-glutamine, fungizone, penicillin and streptomycin. One petri dish con-
taining cells plated on a plain glass cover slip and the other has cells plated on
cover slip coated with liquid crystals. Both had cells at a density of 1.3 9 103 cells/
cm2. After the cells reached sub-confluency, the glass cover slips were removed
from the petri dishes and washed twice with Hanks Balanced Salt Solution (HBSS,
Sigma Aldrich). Subsequently, the cells were fixed with 1% formaldehyde (Sigma
Aldrich) in HBSS for 6 min, rinsed twice with HBSS and permeabilized with 0.1%
Triton X-100 (Sigma Aldrich) for 3 min. The primary antibodies used to stain the
focal adhesions was anti-human vinculin antibody (1:40, Sigma Aldrich) in 1%
BSA (Bovine Serum Albumin, Sigma Aldrich) in HBSS followed by another three
washes. All the immunofluorescence staining were observed using a Nikon Eclipse
80i fluorescence microscope under dark field (DF) at 40x magnification and
images were captured with ACT-2u software. The length of the vinculin accu-
mulations in the micrographs of immunofluorescence staining was determined
using ImageJ software. This experiment was performed in triplicates.

3.4 Determining the Lateral Displacement Using the Cell

Relaxation Technique

HaCaTs were prepared as previously described [26] and the coating of the liquid
crystal substrate was prepared as described in the experiment of AFM nano-
indentation. The cell suspension was added into a petri dish with a liquid crystal
coated cover slip at a cell density of 500 cells/cm2 and incubated at 37°C for 24 h.
After incubation, the cells cultured on the liquid crystals were found adhered and
contracted on the surface of the liquid crystals causing localized deformation lines
at the stress site. Each deformation line formed in an axial direction was induced
by the compressive forces originated from two groups of focal adhesion that
sheared the surface of the liquid crystal substrate over a short distance known as
lateral displacement. The lateral displacement can be determined by relaxing the
cell using an actin depolymerization drug, cytochalasin-B. Cytochalasin-B
specifically inhibits actin polymerization. In this experiment, the adherent cells
were treated with 5 ll of 30 lM cytochalasin B for one hour at 37°C. Cytocha-
lasin-B (35 mg/ml, Sigma Aldrich) was solubilized in 0.042% (v/v) ethanol (in
distilled water). For the treatments, the petri dish was placed on a hot stage
maintaining at 37°C while time lapse images were captured every five minutes for
an hour via a phase contrast microscope bundled with a digital camera. In a
micrograph showing a cell before treatment with cytochalasin-B, the contraction
profile was plotted across the width of the deformation line near the cell periphery.
Similarly, the relaxation profile was obtained from the same location in a
Finite Element Quantification of the Compressive Forces 87

micrograph showing the same cell after 30 min of cytochalasin-B treatment. Both
profiles were plotted on the same graph for determining the lateral displacement
induced by the focal adhesions. Lateral displacement can also be approximated by
taking half of the difference between the widths of deformation line at the
periphery of the cell during contraction and after 30 min of relaxation. This
experiment was repeated for ten similar cultures to obtain lateral displacement for
ten randomly selected cells. Each cell induced an average of 4–5 stress sites.

3.5 Implementation of a 3D Finite Element Model

and Quantification of Compressive Forces

In developing the finite element model for the focal adhesion-liquid crystals, we
considered that the adhesion to the liquid crystal substrate was loaded with a
certain geometry of a cell’s shape and each deformation line was grafted with two
groups of focal adhesion in the liquid crystal substrate (Fig. 2a) which has a height
(Hlc), length (Llc) and width (Wlc) of 30 9 100 9 100 lm, respectively (Fig. 2b).
These focal adhesions located at the margins of cells are interconnected to the
circumferential actin filaments as reported in [34]. The oval shaped focal adhesion
which is approximated to a rectangular was embedded in the liquid crystal sub-
strate. In order to apply a normal force (F) laterally to the interface of the focal
adhesion-liquid crystal, a hollow space was set for the focal adhesion. Therefore,
both ends of the stress site were loaded with a pair of normal lateral forces
(Fig. 2c, top). The focal adhesion has a length (Lfa) approximately 1–1.4 lm and a
thickness (Hfa) of *1 lm as approximated in the immunofluorescence staining
experiment. Therefore, the size of the contact plane of the focal adhesion-liquid
crystals is obtained by Hfa 9 Lfa. The width (Wfa) of the focal adhesion was
assumed to be the widths of two or three vinculin accumulations that are possibly
located at the adhesion junctions which is about 0.3–0.6 lm. A stress tensor
diagram (Fig 2c, bottom) was used to describe the effects of the principal stress in
the x-plane from the focal adhesions and shear stress to the liquid crystal slab in
the y and z directions (sxy and sxz). As a result, the compression applied to the slab
causes deformation of the liquid crystals in y and z directions.
The 3D model of the focal adhesion-liquid crystal interface was implemented
using the Solidworks software. Linear static analysis based on finite element
method available in the Cosmosworks/Solidworks was used to quantify the trac-
tion force of the focal adhesion. Young’s modulus of 108 ± 20 kPa determined by
AFM nano-indentation, shear modulus of 0.464 Pa [26] and Poisson’s ration of
0.49 were set for the liquid crystal model. The quantification of the compressive
force was simulated iteratively to match the lateral displacement obtained from the
contraction and relaxation profile obtained from the experiment of cell relaxation.
Restraint was set at the bottom of the liquid crystal substrate. A total of 10,842
triangular elements at a size of 5.7 lm were used in generating the mesh model.
88 C. F. Soon et al.

Fig. 2 a Schematic diagram showing attachment of a cell on the liquid crystals mediated by the
traction force of the focal adhesions originated from the polymerization of actin filaments.
b Parametrical geometry of the focal adhesion-liquid crystal interface. Lfa, Hfa, Hlc, Wfa and Wlc
are the length of focal adhesion, height and width of the focal adhesion (denoted as fa) and liquid
crystal (denoted as lc) substrate, respectively. c Schematic diagram of a stress site under uni-axial
stress. The stress in two planes are applied by compressive loading (F) from the focal adhesions.
Bottom the associated tensor diagram in three dimensional. The plane stress, rxx causes shear
stress of sxz and sxy in the y and z directions

The von Mises stress on the liquid crystal substrate was also obtained from the
finite element model which was meshed using triangular elements. The Cosmos-
works software was used to solve the linear elastic problem using Hooke’s law and
minimum total potential energy method [35] in which the force {F} is related to
the displacement {U} by
fF g ¼ ½kfU g ð6Þ
where [k] is the stiffness matrix.

3.6 Statistical Analysis

All the values of Young’s modulus, length of vinculin accumulations, compressive

force and maximum nodal stress were expressed in mean ± standard deviation
(SD). N is the size of the sample. In the Statistical Package for Social Sciences
(SPSS) software, the difference of the length of vinculin expressed by cells on the
glass cover slip and liquid crystal substrate was compared using the paired t-test
Finite Element Quantification of the Compressive Forces 89

which is significant for p \ 0.001. Normality of the compressive force distribution

was determined using one-sample Kolmogorov–smirnov test at a significance level
of p [ 0.05.

4 Results and Discussion

The development of a method to quantify keratinocyte contraction forces exerted

on a liquid crystal substrate involves four steps. Firstly, the phase displayed by the
surface of the cholesteryl ester liquid crystals in cell culture media was deter-
mined. Secondly, the project involved characterizing the Young’s modulus of the
liquid crystal and the size of focal adhesions. Thirdly, determining the displace-
ment formed on the surface of the liquid crystal substrate which is due to the
compression forces exerted via focal adhesions. Finally, this allowed the
computation of the compressive forces using the finite element method based on
the displacement information. After the cholesteryl ester liquid crystals was
cultured in the cell culture media, a translucent film was seen on the surface of the
cholesteryl ester liquid crystals in phase contrast microscopy and this liquid crystal
film is an adhesion layer to the keratinocytes (Fig. 3a). In the cross-polarized
microscopy, we found a similar translucent film formed on top of the colorful
cholesteric liquid crystals substrate (Fig. 3b). To investigate the phase of the film,
we decided to examine the liquid crystal film in isolation. The liquid crystal film
which is approximately 10–20 lm thick was carefully transferred to a glass cover
slip and examined in a Differential Interference Contrast (DIC) and cross-polarized
microscope. The film appears pinkish in DIC microscopy (Fig. 3c) but dark with a
few white bands of streaks and some dark focal conic texture (image not shown) in
cross-polarized microscopy (Fig. 3d) which indicates the formation of smectic
phase for the lipid liquid crystals. Since the amphiphilic molecules are aligned in
perpendicular to the glass surfaces, the well-aligned lipid molecules in smectic
appear dark when viewed down the cross-polarizer which is optically isotropic
[36]. This film known as the lamella mesophase consists of bilayers of lipid
molecules interlaced by layers of water molecules and they are uniform under
defect free condition [37]. For this film, regardless on which direction it flowed or
deformed would always displayed with dark deformation lines (Fig. 3b) and,
hence, it is also assumed to be physically isotropic.
Based on the previous finding, AFM nano-indentation was suggested to
determine the Young’s modulus of the surface of the cholesteryl ester liquid
crystal substrate which has shown isotropic property. An example of the force-
displacement curve acquired for the liquid crystal substrate using the AFM nano-
indenter is as shown in Fig. 4. During the unloading phase, the force measured by
the indenter was also contributed from the lateral force by the adherent liquid
crystals that were being sheared laterally while moving upward together with the
indenter (Fig. 1b). Hence, the Young’s modulus determined is also a function of
the lateral elasticity of the liquid crystals. Overall, both loading and unloading
90 C. F. Soon et al.

Fig. 3 Micrographs showing a liquid crystal film rupturing off the bulk cholesteryl ester liquid
crystals in a phase contrast microscopy (with cell attached) and b the corresponding cross-
polarized micrographs of (a). Similar thin film was isolated from the bulk cholesteryl ester liquid
crystals, relocated to a glass cover slip and examined in the c differential interference contrast
(DIC) microscopy and the corresponding cross-polarized micrograph in d showing the isotropic
phase of the liquid crystal film. Scale bar: 20 lm

curves are non linear except the initial portion of the unloading curve (Fig. 4). The
calculation of the Young’s modulus was based on the data obtained from the
unloading curve. In Fig. 4, the liquid crystal immersed in the culture media was
indented with a peak load of Pmax = 4.84 lN at a maximum depth of
hmax = 4,910 nm. The final depth after complete unloading, hf, was determined
from the unloading data as 2,713 nm. Retraction of the indenter was halted at a
depth of hs = 1,068 nm. The power law fitting parameters as determined by the
Triboscope software from the unloading data were 1.1 and 1.957 for A and m,
respectively, using the least squares fitting procedures [33]. The average stiffness
of the liquid crystals calculated from Eq. 1 was 0.0017 ± 0.00067 lN/nm
(mean ± SD) and the Young’s modulus derived from Eq. 6 was 108 ± 20 kPa
(mean ± SD). This value was considerably lower than 0.3 MPa of the thermo-
tropic cubic phase liquid crystals determined in air [38]. This difference is due to
the hydration effects induced in our experiment. Interestingly, the Young’s
modulus of the surface of the liquid crystals appeared to mimic the elastic modulus
of the epidermis layer [39, 40] and other types of soft tissue [20] which is an added
advantage in our cells force transducer development besides having a liquid crystal
system similar to the cell membrane. At this elastic modulus, the cells are best
supported to grow because the substrate closely resembles the stiffness of the in
Finite Element Quantification of the Compressive Forces 91

Fig. 4 A force-displacement
curve of the surface of the
liquid crystals obtained using
the AFM based nano-
indentation. The inset shows
the related finite loading
depth (hf), contact depth (hc)
and maximum loading depth
(hmax) at a maximum loading
force (Pmax) in a force-
displacement curve

vitro system [20, 21], hence provides a more natural environment in which to sense
cell traction forces. A material with much higher or lower elastic modulus than that
of the in vivo tissue utilized to design a cell mechano-transducer could be inter-
pretated as pathological condition by the cells [19, 20].
The length of the vinculin accumulations is a crucial parameter for the inputs of
the model because this molecule interfaces to the liquid crystals and drives the
lateral force which is steered by the actin filaments against the wall of interface.
We compared the length of the vinculin accumulations and their distribution on the
glass cover slip and liquid crystals to show that the focal adhesion expressions of
cells are different on soft and hard surfaces. On the hard cover slip, vinculin
staining is randomly distributed and broadly spread (Fig. 5a). In contrast to the
cells on the liquid crystal substrate (Fig. 5b), vinculin staining is predominant at
the periphery of the cells and at cell–cell junctions. Additionally, cells grown on
the liquid crystals are more rounded compared with those on the cover slips
(Fig. 5a, b). In corresponding to the immuno-staining, the length of vinculin
accumulations for cells grown on the glass and liquid crystals are 1.41 ± 0.55 lm
and 1.03 ± 0.4 lm, respectively. They are significantly different for p \ 0.001 at
N = 525 (Fig. 5c). As reported in [41], a typical size of the vinculin on the glass
cover slip is 2–5 lm and approximately 1 lm for small dot-like focal complex.
Based on this data, we found that focal complex and focal adhesions are both
present in the cells adhered to the liquid crystal substrate. Different structures of
adhesions plaques expressed by the cells on the liquid crystal substrate and cover
slip show the sensitivity of the cell to the underlying substrate [42, 43].
On the surface of the liquid crystal substrate, keratinocyte show a nearly
iso-diametric characteristic. In a quiescent state, the contractility ceased and the
focal adhesions remain compressing the surface of the liquid crystals and the
traction is expressed in a form of deformation (Fig. 6a). Due to the lateral force at
the stress plane (rxx), the contact area was displaced by a small distance which is
92 C. F. Soon et al.

Fig. 5 Micrographs of immunofluorescence staining for vinculins adhered to the a glass and
b liquid crystal substrate. c Comparison of the length of vinculin accumulations (mean ± SD,
N = 525) for adherent keratinocytes on the glass cover slip and liquid crystals which is
significantly different for P \ 0.001. Scale bar: 25 lm

defined as the lateral displacement. Consequently, deformations were formed in

the other two planes due to shear stress, sxz and sxy (Fig. 2c) and this was observed
as a short deformation line (Fig. 6a, 0 min). The liquid crystal substrate constitutes
of closely packed molecules that tilt easily when there is an external perturbation
within a finite region. This sensing flexibility can be seen in Fig. 6a. In order to
determine the lateral displacement due to the lateral force applied by a pair of focal
adhesions at two planes (rxx in opposite directions), cytochalasin-B was used to
relax the cells and allow the cells to release the compressive loading (Fig. 6a,
30 min). For the cell in Fig. 6a, the stress site or deformation line contains in a box
has a width of 6 lm was chosen as an example for the following discussion. In
Fig. 6b, the contraction and relaxation profile were obtained from the profile lines
at the same location as marked in the bottom image of Fig. 6a at 0 min and at
30 min, respectively. The lateral displacement obtained from the comparison of
Finite Element Quantification of the Compressive Forces 93

Fig. 6 a Top the treatment effects of 30 lM cytochalasin B in phase contrast micrographs taken
at 0 (contraction) and 30 min (relaxation) for a cell cultured on the surface of the liquid crystals
(Scale bar: 10 lm). Bottom enlarge exerts for a stress site during contraction (left) and relaxation
(right) as labeled in the black boxes of (a), respectively. The lines shown on the left and right
images are the contraction and relaxation profile lines, respectively (enlarge exert scale bar: 5
lm). b Plots of the contraction and relaxation profile for the stress site contained in the black
boxes of (a). Lateral displacement determined for each side was 0.4 lm

profiles was determined at 0.4 lm. The contraction of the cell is attributed to the
simultaneous shortening of the circumferential actin filaments leading to a
reduction of distance between the two stress planes or focal adhesions located at
94 C. F. Soon et al.

the periphery of the cell [4, 11]. Under static condition, compression is appro-
priately used to describe the mechanics of the quiescent cells on the liquid crystal
surface.
With all the physical properties of the liquid crystal and dimensions of the focal
adhesions determined, a finite element model (Fig. 7) for a cell deforming the
surface of the liquid crystal substrate (Fig. 6a) was developed. The geometry of the
focal adhesion used and the generated mesh model for the liquid crystal substrate
is as shown in Fig. 7a and b, respectively. The simulated displacement and stress
distribution on the liquid crystals are as shown in Fig. 7c and d. Two identical
focal adhesions were formed in contact with the liquid crystals substrate and
similar compressive forces were applied at each contact plane during the simu-
lation. In this example, the applied force, F = 31.5 nN per focal adhesion, induced
a maximum displacement of 0.4 lm at each side of the contact plane of focal
adhesion-liquid crystal. This yielded a maximum nodal stress of 25.29 nN/lm2 at
the contact plane (Fig. 7d). At the contact planes, the nodal stress and lateral
displacement are the highest and they gradually decreased quadratically as the
distance increases to the center of the stress site (Fig. 7c and 8a). The responses of
the stress and displacement curves at the stress site were found to be an inversion
of the contraction profile as indicated in Fig. 6b which shows that our focal
adhesion-liquid crystals model has a good correlation with the effect of
compression produced by a cell on the surface of the liquid crystals seen in the
experiment (Fig. 6a). A similar result was observed at other stress sites. This
supports the view that a compression model is a more accurate estimation for the
deformation induced by the focal adhesions on the linear elastic liquid crystals or a
continuum. For other stress sites of the same cell in Fig. 6a, we found that the
compressive forces induced by the focal adhesion (1–1.4 lm2) linearly increases
with the lateral displacement of the liquid crystals (Fig. 8b). To study the effects
of the size of focal adhesions to the compressive forces and stress on the surface of
the liquid crystals, focal adhesion from 1 to 1.4 lm2 were applied. The range of
size was as determined by the immuno-staining for cells adhered to the liquid
crystal substrate. As a result, we found that the variation of the forces
(± 0.8–2 nN) and stress (±0.9–3 nN/lm2) on the liquid crystal substrate due to
the size of focal adhesions is rather small while the stress is linearly proportional to
the applied force (Fig. 8c). This result shows that the compressive force is majorly
influenced by the lateral displacement induced to the liquid crystals and less due to
the size of focal adhesion used. Although aggregation of the focal adhesions is
associated with the responses in increasing contractility or high Rho activity
[44], the force applied is still mainly steered by the contractility of the actin
cytoskeleton.
For the fifty pairs of focal adhesion obtained from ten non-motile keratinocytes
adhered to the liquid crystal substrate, the compressive forces exerted via the focal
adhesions for non-motile keratinocytes as determined by our focal adhesion-liquid
crystals model was in the range of 3–38 nN per focal adhesion (Fig. 8d). The
compressive forces were normally distributed (p = 0.536, N = 50 pairs of focal
adhesions, 10 cells) and have a mean ± SD = 19.86 nN ± 9.57 nN. These forces
Finite Element Quantification of the Compressive Forces 95

Fig. 7 a The dimensions of a focal adhesion in the finite element model. b A mesh model of
cell-liquid crystal consists of 10,842 triangular elements at a size of 5.7 lm. c A 3-D topography
of the displacement of the liquid crystals induced by the uni-axial loading of the focal adhesions
(1.4 lm2) at one of the stress sites. d von Mises stress computed for the compressive loads
induced via the focal adhesions (1.4 lm2) of a cell at various stress sites

are comparable with those of epithelial cells in the range of 5–35 nN [16] although
the cells on the micro-fabricated cantilever tend to express isolated focal adhesion
rather than clustered focal adhesions that are usually found for cells adhering on a
continuum [4, 8]. By using an AFM cantilever to probe cells, Sagvolden and
co-worker [1] showed that cervical carcinoma cell requires 100–200 nN to dis-
sociate from the surface. Moreover, it has been shown that epithelial cells detached
from the surface when exposed to tangential force of 100 nN in a centrifugation
system [45]. Whereas, a higher range of traction force from 10 up to 1,200 nN was
reported for fibroblast grown on an elastic substrate [46]. Comparatively, the
96 C. F. Soon et al.

Fig. 8 a Stress and lateral displacement plotted across a stress site. b Compressive force versus
lateral displacement for five stress site. c Maximum nodal stress versus compressive force plotted
for the stress sites. The compressive force and nodal stress (mean ± SD) were determined using
focal adhesions with size of 1–1.4 lm2. d Normally distributed compressive forces (p = 0.536,
mean ± SD = 19.86 ± 9.57 nN, 50 pairs of focal adhesions) estimated for ten non-motile
keratinocytes

forces expressed by the keratinocytes on the liquid crystals are literally lower than
the cells adhered to a stiff surface such as glass. For non-motile cell like the
keratinocyte, the forces exerted on a substrate should be relatively lower than the
fibroblast which is highly motile because motility is proportional to the traction
forces of cells.

5 Conclusions

The surface of the cholesteryl ester liquid crystals after interfacing with the cell
culture media has transformed into an optically isotropic layer in cross-polarized
microscopy. Characterization of the physical properties for the surface of the
liquid crystals and size of the focal adhesion has been presented. The physical
properties of the liquid crystals determined were included in the focal adhesion-
liquid crystal model. Based on the deformation induced by a single cell on the
surface of the liquid crystals, linear static analysis based on the finite element
Finite Element Quantification of the Compressive Forces 97

method was applied to the model in quantifying the compressive forces of cells.
Although focal adhesions may vary in sizes on the liquid crystals, we found that
the variation of loading forces is just within a few nano newtons. On the other
hand, the magnitude of the force expressed is dependent on the lateral displace-
ment induced via the focal adhesion. The profile of stress and displacement shown
in the model correlated well with the deformation formed by the small cluster of
focal adhesions at the stress site. The range of forces expressed by keratinocytes on
the liquid crystal substrate is lower compared to other measurement techniques
could be attributed to the compliance of the liquid crystals used.

Acknowledgments The author would like to acknowledge Dr. Steve Britland and Dr. Samira
Batista Lobo for providing technical advice and many thanks to Professor Des Tobin for his
permission to use a fluorescence microscope in his laboratory. Thanks to Dr Ravindra S. Dhumal
for his useful discussion about the liquid crystal phase. The authors are grateful to the Malaysia
Ministry of Higher Education for continuous financial support.

References

1. Sagvolden, G., Giaever, I., Pettersen, E.O., et al.: Cell adhesion force microscopy. PNAS 96,
471–476 (1999)
2. Rivelinea, D., Zamir, E., Balaban, N.Q., et al.: Focal contacts as mechanosensors: externally
applied local mechanical force induces growth of focal contacts by an mDia1-dependent and
ROCK-independent mechanism. Cell Biol. 153, 1175–1185 (2001)
3. Harris, A., Wild, P., Stopak, D.: Silicone rubber substrata: a new wrinkle in the study of cell
locomotion. Science 208, 177–179 (1980)
4. Peterson, L.J., Rajfur, Z., Maddox, A.S., et al.: Simultaneous stretching and contraction of
stress fibers in vivo. Mol. Biol. Cell 15, 3497–3508 (2004)
5. Burridge, K., Chrzanowska-Wodnicka, M.: Focal adhesions, contractility and signaling.
Annu. Rev. Cell Dev. Biol. 12, 463–519 (1996)
6. Goffin, J.M., Pittet, P., Csucs, G., et al.: Focal adhesion size controls tension-dependent
recruitment of smooth muscle actin to stress fibers. Cell Biol. 172, 259–268 (2006)
7. Schwartz, M.A., DeSimone, D.W.: Cell adhesion receptors in mechanotransduction. Curr.
Opin. Cell Biol. 20, 551–556 (2008)
8. Burridge, K., Chrzanowska-Wodenicka, M., Zhong, C.: Focal adhesion assembly. Trends
Cell Biol. 7, 342–347 (1997)
9. Kirfel, G., Herzog, H.: Migration of epidermal keratinocytes: mechanisms, regulation, and
biological significance. Protoplasma 223, 67–78 (2004)
10. Pellegrin, S., Mellor, H.: Actin stress fibres. J. Cell Sci. 120, 3491–3499 (2007)
11. Owaribe, K., Kodama, R., Eguchi, G.: Demonstration of contractility of circumferential actin
bundles and its morphogenetic significance in pigmented epithelium in vitro and in vivo. Cell
Biol. 90, 507–514 (1981)
12. Merkel, R., Kirchgeßner, N., Cesa, C.M., et al.: Cell force microscopy on elastic layers of
finite thickness. Biophys. J. 93, 3314–3323 (2007)
13. Oliver, T., Dembo, M., Jacobson, K.: Traction forces in locomoting cells. Cell Motil.
Cytoskelet. 31, 225–240 (1995)
14. Burton, K., Taylor, D.L.: Traction forces of cytokinesis measured with optically modified
substrata. Nature 385, 450–454 (1997)
15. Beningo, K., Lo, C., Wang, Y.: Flexible polyacrylamide substrata for the analysis of
mechanical interactions at cell-substratum adhesions. Methods Cell Biol. 69, 325–339 (2002)
98 C. F. Soon et al.

16. Roure, O.D., Saez, A., Buguin, A., et al.: Force mapping in epithelial cell migration. Proc.
Natl. Acad. Sci. U S A 102, 2390–2395 (2005)
17. Jurado, C., Haserick, J.R., Lee, J.: Slipping or gripping? fluorescent speckle microscopy in
fish keratocytes reveals two different mechanisms for generating a retrograde flow of actin.
Mol. Biol. Cell 16, 507–518 (2005)
18. Beningo, K.A.: Flexible substrata for the detection of cellular traction forces. Trends Cell
Biol. 12, 79–84 (2002)
19. Engler, A.J., Griffin, M.A., Sen, S., et al.: Myotubes differentiate optimally on substrates with
tissue-like stiffness pathological implications for soft or stiff microenvironments. Cell Biol.
166, 877–887 (2004)
20. Levental, I., Georgesa, P.C., Janmey, P.A.: Soft biological materials and their impact on cell
function. Soft Matter 3, 299–306 (2007)
21. Solon, J., Levental, I., Sengupta, K., et al.: Fibroblast adaptation and stiffness matching to soft
elastic substrates. Biophys. J. 93, 4453–4461 (2007)
22. Pathak, A., Deshpande, V.S., McMeeking, R.M., et al.: The simulation of stress fibre and
focal adhesion development in cells on patterned substrates. J. Roy. Soc. Interface 5, 507–524
(2008)
23. Britland, S., Perridge, C., Denyer, M., et al.: Morphogenetic guidance cues can interact
synergistically and hierarchically in steering nerve cell growth. Exp. Biol. Online 1, 1–15
(1996)
24. Stanley, J., Hawley-Nelson, P., Yaah, M., et al.: Laminin and bullous pemphigoid antigen are
distinct basement membrane proteins synthesized by epidermal cells. J. Invest. Dermatol. 78,
456–459 (1982)
25. O’Toole, E.A.: Extracellular matrix and keratinocyte migration. Clin. Exp. Dermatol. 26,
525–530 (2001)
26. Soon, C.F., Youseffi, M., Blagden, N., et al.: Characterization and biocompatibility study of
nematic and cholesteryl liquid crystals. Proc. WCE 2, 1872–1875 (2009)
27. Ghannam, M.T., Esmail, M.N.: Rheological properties of poly(dimethylsiloxane). Ind. Eng.
Chem. Res. 37, 1335–1340 (1998)
28. Boutelier, D., Schrank, C., Cruden, A.: Power-law viscous materials for analogue
experiments: New data on the rheology of highly-filled silicone polymers. J. Struct. Geol.
30, 341–353 (2008)
29. Soon, C.F., Youseffi, M., Blagden, N., et al.: Interactions of cells with elastic cholesteryl
liquid crystals. Proc. IFBME 25/X, 9–12 (2009)
30. Soon, C.F., Youseffi, M., Gough, T., et al.: Rheological characterization of the time-
dependent cholesteric based liquid crystals and in situ verification. (2010, Submitted)
31. Pharr, G.M.: Measurement of mechanical properties by ultra low load indentation. Mater. Sci.
Eng. A 253, 151–159 (1998)
32. Doerner, M.F., Nix, W.D.: A method for interpreting the data from depth-sensing indentation
instruments. J. Mater. Res. 1, 601–609 (1986)
33. Oliver, W.C., Pharr, G.M.: An improved technique for determining hardness and elastic
modulus using load and displacement sensing indentation experiments. J. Mater. Res. 7,
1564–1583 (1992)
34. Mohl, C., Kirchgebner, N., Schafer, C., et al.: Becoming stable and strong: the interplay
between vinculin exchange dynamics and adhesion strength during adhesion site maturation.
Cell Motil. Cytoskelet. 66, 350–364 (2009)
35. Fagan, M.J.: Finite Element Analysis: Theory and Practice. Longman, Singapore (1992)
36. Asher, S.A., Pershan, P.S.: Alignment and defect structures in oriented phosphatidylcholine
multilayers. Biophys. J. 27, 393–421 (1979)
37. Hyde, S.: Identification of lyotropic liquid crystalline mesophase. In: Holmberg, K. (ed.)
Handbook of Applied Surface and Colloid Chemistry. Wiley, New York (2001)
38. Even, C., Imp0 eror-Cler M., Pieranski, P.: Exploring the facets of ‘‘soft crystals’’ using an
atomic force microscope. Eur. Phys. J. E 20, 89–98 (2006)
Finite Element Quantification of the Compressive Forces 99

39. Takeo, M.: Skin biomechanics from microscopic viewpoint: mechanical properties and their
measurement of horny layer, living epidermis, and dermis. Fragr. J. 35, 36–40 (2007)
40. Hendriks, F.M., Brokken, D., Oomens, C.W.J., et al.: The relative contributions of different
skin layers to the mechanical behavior of human skin in vivo using suction experiment. Med.
Eng. Phys. 28, 259–266 (2006)
41. Geiger, B., Bershadsky, A., Pankov, R., et al.: Transmembrane extracellular matrix-
cytoskeleton crosstalk. Natl. Rev. Mol. Cell Biol. 2, 793–805 (2001)
42. Yeung, T., Georges, P.C., Flanagan, L.A., et al.: Effects of substrate stiffness on cell
morphology, cytoskeletal structure, and adhesion. Cell Motil. Cytoskelet. 60, 24–34 (2005)
43. Discher, D.E., Janmey, P., Wang, Y.-l.: Tissue cells feel and respond to the stiffness of their
substrate. Science 310, 1139–1143 (2005)
44. Geiger, B., Bershadsky, A.: Assembly and mechanosensory function of focal contacts. Curr.
Opin. Cell Biol. 13, 584–592 (2001)
45. Thoumine, O., Ott, A., Louvard, D.: Critical centrifugal forces induce adhesion rupture or
structural reorganization in cultured cells. Cell Motil. Cytoskelet. 33, 276–287 (1996)
46. Burton, K., Park, J.H., Taylor, D.L.: Keratocytes generate traction forces in two phases. Mol.
Biol. Cell 10, 3745–3769 (1999)
 Part II
Properties
Preparation of Hydroxyapatite-Forsterite-
Bioactive Glass Composite Nanopowder
for Biomedical Applications

Mohammadhossein Fathi, Vajihesadat Mortazavi

and Maryam Mazrooei Sebdani

Abstract Bioceramics with reduced grin size below 100 nm have superior
mechanical properties and more bioactivity than conventional ceramics. The aim
of this work was to prepare and characterize a novel hydroxyapatite-forsterite-
bioactive glass composite nanopowder. The novel hydroxyapatite-forsterite-bio-
active glass composite nanopowder was fabricated by incorporation of forsterite
and bioactive glass nanopowder in hydroxyapatite matrix via a sol–gel process.
X-ray diffraction (XRD), scanning electron microscopy (SEM) and transition
electron microscopy (TEM) techniques were utilized in order to determine the
phase composition, and evaluate the morphology and particle size of the synthe-
sized nanopowders. The effect of the sintering temperature was also investigated.
The results show that the appropriate temperature for desired calcination was
600°C and that the particle size of the prepared composite nanopowder was about
60–70 nm. The obtained results suggest that the prepared composite nanopowder
might be a good candidate for biomedical applications.

 
Keywords Biomaterials Bioceramic nanopowder Composite nanopowder 
 
Nanomaterials Sol–gel processes Biomedical applications

M. Fathi (&)  M. M. Sebdani

Biomaterials Group, Department of Materials Engineering,
Isfahan University of Technology, 8415683111 Isfahan, Iran
e-mail: [email protected]
M. M. Sebdani
e-mail: [email protected]
V. Mortazavi
Torabinejad Dental Research Center, School of Dentistry,
Isfahan University of Medical Science, 8174673461 Isfahan, Iran
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 103
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_8,
Ó Springer-Verlag Berlin Heidelberg 2012
104 M. Fathi et al.

1 Introduction

Over the past decades, advances in many special bioactive ceramics such as
hydroxyapatite and bioactive glasses have contributed a certain role to the
development of a modern health care industry and have improved the quality of
human life [1]. These bioactive ceramics have been widely used for a lot of
applications [2]. One of the applications is to use them as an implant. The most
important and specific features of any ceramic material for using as an implant in
vivo depends on resisting to complex stresses at the site of application and their
compatibility with the biological environment [3]. Therefore, selection of a proper
bioceramic is very important.
One of the most unique bioceramics is artificial hydroxyapatite due to its
structural and compositional similarity to the mineralized matrix of natural bone
and consequently good bioactivity and osseoconductivity [4]. Hydroxyapatite
(HAP, Ca10(PO4)6(OH)2) is the most ubiquitous calcium phosphate that has been
studied extensively for medical applications in the form of powders [5],
composites or even coatings [5]. However, the major obstacle for its use is its poor
and insufficient mechanical properties and consequently, its scope of clinical
applications is limited to non-load bearing applications such as osteoconductive
coatings on metallic prosthesis and as nano-powders in spinal fusion. On the other
hand, the incorporation of a ceramic reinforcement as fibers, whiskers, platelets or
particles in a ceramic matrix improves the mechanical properties [6]. In experi-
mental studies, it has been observed that the incorporation of glasses within the
CaO-SiO2-P2O5 system is a suitable choice for improving its properties. Bioactive
glass added hydroxyapatite composite is able to bond to both soft and hard tissue
without an intervening fibrous layer and as a result, exhibits greater biological
activities than commercial hydroxyapatite [7, 8]. However, similar to hydroxy-
apatite, these glasses could not be used in load-bearing applications and
consequently, these composites have not achieved adequate mechanical properties
[9]. Therefore, it is expected that the incorporation of other ceramics with better
mechanical properties such as forsterite (Mg2(SiO4)) improve the mechanical
properties of hydroxyapatite-bioactive glass composites. In contrast to hydroxy-
apatite and bioglass, forsterite ceramics show considerable fracture toughness and
hardness. The fracture toughness of forsterite ceramics is 2.4 MPam1/2 higher to
the lower limit reported for bone implants and hydroxyapatite ceramics
(0.6–1 MPam1/2) [10]. On the other hand, results showed that ceramics with grain
size lower than 100 nm have superior mechanical and biological properties [10]. In
this case, Kharaziha and Fathi [11] also showed that the dissolution rate of the
forsterite nanopowder is higher than conventional forsterite powders and apatite is
formed. Therefore, it is expected that the composite nanopowder has better bio-
activity and mechanical properties than coarse crystals.
Several different synthetic methods have been developed to generate composite
nanopowder. Among the several alternative methods, sol–gel synthesis of com-
posite ceramics has been widely used. The specific feature of the sol–gel product is
Preparation of Hydroxyapatite-Forsterite-Bioactive Glass Composite Nanopowder 105

its nano-size dimension of the primary particles. This small domain is a very
important parameter for improvement of the contact reaction and the stability at
the artificial/natural bone interface [12–14].
The aim of this work was to prepare and characterize a novel hydroxyapatite-
forsterite-bioglass composite nanopowder.

2 Experimental Procedure

2.1 Preparation of Hydroxyapatite- Forsterite-Bioglass

Composite Nanopowder

The starting materials were phosphoric pentoxide (P2O5, Merck), nitrate calcium
tetrahydrate (Ca(NO3)2.4H2O, Merck) and ethanol (Merck, 99.9% purity) for
preparing composite powder. Forsterite nanopowder [11] and 45S bioactive glass
nanopowder (46% CaO, 45% SiO2 and 9% P2O5) [15] were also used for preparing
the new composite powder. The experimental procedure consisted in dissolving a
designed amount of phosphoric pentoxide and calcium nitrate tetrahydrate
in absolute ethanol to form a 1.67 mol/l solution. Forsterite nanopowder and
bioactive glass nanopowders with 10 wt% total content were added to the solution
after 30 min, respectively. The mixture was continuously stirred for 24 h at
ambient temperature. At this time, a transparent gel was obtained. After that, the
obtained transparent gel was dried at 80°C for 24 h in an electrical air oven and
aged for 24 h. The dried gels were individually heated at a rate of 5°C/min up to
600°C in a muffle furnace and then were placed in air for cooling to ambient
temperature. The same processes were performed for other dried gel samples at the
temperatures of 700 and 800°C.

2.2 Powder Characterization

The phase composition and crystalleinity of as prepared and calcined hydroxy-

apatite-forsterite-bioglass nanoparticles were analyzed and determined using
X-ray diffraction (XRD, Philips X’Pert-MPD) using a CuKa radiation generated at
40 kV and 30 mA, in the range of 20° \ 2h \ 80° at a scan speed of 5°/min. The
crystallite size of the composite nanopowders was calculated by using Scherer
equation (1) [16]:
0:89k
b¼ ð1Þ
t cos h
where b is the width of peak in the middle of its height, k is the wavelength
(= 0.154 nm), h is the Bragg angle and t is the apparent crystallite size.
106 M. Fathi et al.

Fig. 1 X-ray diffraction

patterns of composite powder
sintered at 600°C

The scanning electron microscopy (SEM, Phillips XL30) and energy dispersive
X-ray (EDX, Seron AIS-2100) techniques were utilized to study the morphology
and elemental analysis of the synthesized composite nanopowder. The samples
were sputter coated with a thin layer of gold (about several nanometers thick)
using a PVD define-apparatus for enhanced resolution for secondary electron
microscopy. Transmission electron microscopy (TEM; CM 2000 PHILIPS)
technique was utilized to characterize the morphology and grain size of the
synthesized composite nanopowder.

3 Results and Discussion

3.1 XRD Diffraction Analysis

The XRD technique was employed to assess the phase purity and crystallographic
changes. Figure 1 shows an XRD pattern of a sol–gel prepared composite nano-
powder that was sintered at 600°C. The characteristic peaks of hydroxyapatite [17]
and forsterite [18] can be seen.
The effect of the sintering temperature on the formation of the composite
nanopowder can be observed in Fig. 2. As can be seen that the sintering
temperature performs a considerable role on the formation of the composite
nanopowder and leads to the decomposition of hydroxyapatite into the additional
crystalline phases (b-TCP and CaO) as the sintering temperature increases to
700°C or above. The peak intensities corresponding to the additional peaks
decreased as the temperature decreased from T = 700°C (Fig. 2). Therefore, the
desired temperature was 600°C. The following reaction is suggested:

Ca10 ðPO4 Þ6 ðOHÞ2 = 3Ca3 ðPO4 Þ2 + H2 O

Several peaks belonging to the hydroxyapatite became more distinct and also
the widths of the peaks became narrower, which suggested an increase in the
degree of powder crystallineity [19, 20] as the sintering temperature increased
Preparation of Hydroxyapatite-Forsterite-Bioactive Glass Composite Nanopowder 107

Fig. 2 X-ray diffraction

patterns of composite
powders calcined at different
temperatures

Fig. 3 SEM micrograph of

composite nanopowder after
heat treatment at 600°C
(White square box was used
for EDX spectrum)

from 600°C to 800°C. The crystallineity of the composite nanopowder could be

determined by the following Eq. 2 [21]:
V112=300
XC ¼ 1  ð2Þ
I300
Where V112/300 is the intensity of the hollow between (112) and (300) diffrac-
tion peaks of hydroxyapatite and I300 is the intensity of (300) diffraction peak.
By Eq. 2, the crystallineity of the obtained powder was about 70%. The grain
size of sol–gel prepared particles sintered at 600°C determined using Eq. 1 was
about 60–70 nm.

3.2 SEM Evaluation and EDX Analyses

Figure 3 shows the SEM micrograph of the powder particles after heat treatment at
600°C. The EDX spectrum of the obtained composite nanopowder which is
defined in Fig. 3 is depicted in Fig. 4. The EDX spectrum indicates that the
particles consisted of all elements.
108 M. Fathi et al.

Fig. 4 EDX spectrum of the

composite nanopowder
sintered at 600°C

Fig. 5 TEM micrograph of

composite nanopowder
calcined at 600°C

3.3 TEM Analysis

Figure 5 illustrates the TEM micrograph of the composite nanopowder sintered at

600°C. As it can be observed, the particle size of the composite nanopowder is less
than 100 nm and the composite nanopowder shows an ellipse-like morphology.

4 Conclusion

Hydroxyapatite-forsterite-bioglass composite nanopowder was prepared via a

sol–gel method by adding forsterite and bioglass to hydroxyapatite sol. The
Optimum temperature for synthesizing of the composite was 600°C and the par-
ticle size of the composite nanopowder was about 60–70 nm at this temperature.
The prepared composite nanopowder might be a good candidate for biomedical
applications.
Preparation of Hydroxyapatite-Forsterite-Bioactive Glass Composite Nanopowder 109

Acknowledgments The authors thank for support of this research by Isfahan University of
Technology.

References

1. Czka, M., Cholewa, K., Czka-Osyczka, A.: Gel-derived powders of CaO-P2O5-SiO2 system
as a starting material to production of bioactive ceramics. J. Alloy. Compd. 248, 42–51
(1997)
2. Ohgushi, H., Goldberg, V.M., Caplan, A.I.: Hetrotopic osteogenesis in porous ceramics
induced by marrow cells. J. Orthop. Res. 7, 568–578 (1998)
3. Mazaheri, M., Haghighatzadeh, M., Zahedi, A.M., et al.: Effect of a novel sintering process
on mechanical properties of hydroxyapatite ceramics. J. Alloy. Compd. 471, 180–184 (2009)
4. Halouani, R., Bernache-Asolant, D., Champion, E., et al.: Microstructure and related
mechanical properties of hot pressed hydroxyapatite ceramics. J. Mater. Sci. Mater. Med. 5,
563–568 (1994)
5. Fathi, M.H., Mohammadi Zahrani, E.: Fabrication and characterization of fluoridated
hydroxyapatite nanopowders via mechanical alloying. J. Alloy. Compd. 475, 408–414 (2009)
6. Oktar, F.N., Guller, G.: Sintering effects on mechanical properties of glass-reinforced
hydroxyapatite composites. Ceram. Int. 28, 617–621 (2002)
7. Daniel, C., Clupper, J.J., Mecholsky, D.C., Greenspan, D.: Bioactivity of tape cast and
sintered bioactive glass-ceramic in simulated body fluid. Biomaterials 23, 2599–2606 (2002)
8. Sepulveda, P., Jones, J.R., Hench, L.L.: Bioactive sol–gel foams for tissue repair. J. Biomed.
Mater. Res. 59, 340–348 (2002)
9. Suchanek, W., Yoshimura, M.: Processing and properties of hydroxyapatite-based
biomaterials for use as hard tissue replacement. J. Mater. Res. 13, 94–117 (1998)
10. Fathi, M.H., Kharaziha, M.: The effect of fluorine ion on fabrication of nanostructure
forsterite during mechanochemical synthesis. J. Alloy. Compd. 472, 540–545 (2009)
11. Kharaziha, M., Fathi, M.H.: Synthesis and characterization of bioactive forsterite
nanopowder. Ceram. Int. 35, 2449–2454 (2009)
12. Liu, M., Troczynski, T., Tseng, W.J.: Water-based sol–gel synthesis of hydroxyapatite:
process development. Biomaterials 22, 1721–1730 (2001)
13. Jillavenkatesa, A., Condrate, R.A.: Sol-gel processing of hydroxyapatite. J. Mat. Sci. 33(16),
4111–4119 (1998)
14. Kim, S., Kumta, P.: Sol-gel synthesis and characterization of nanostructured hydroxyapatite
powder. Mater. Sci. Eng. B 111, 232–236
15. Fathi, M.H., Doostmohammadi, A.: Bioactive glass nanopowder and bioglass coating for
biocompatibility improvement of metallic implant. J. Mater. Process. Technol. 209(3), 1385–
1391 (2009)
16. Cullity, B.D.: Elements of X-Ray Diffraction. Addison-Wesley, London (1978)
17. JCPDS Card No. 9-432, 1994
18. JCPDS Card No. 34-0189, 1984
19. Muralithran, G., Ramesh, S.: Effects of sintering temperature on the properties of
hydroxyapatite. Ceram. Int. 26, 221–230 (2000)
20. Fathi, M.H., Hanifi, A.: ‘‘Evaluation and characterization of nanostructure hydroxyapatite
powder prepared by simple sol–gel method. Mater. Lett. 61, 3978–3983 (2007)
21. Pang, Y.X., Bao, X.: Influence of temperature, ripening time and calcination on the
morphology and crystallinity of hydroxyapatite nanoparticles. J. Euro. Ceram. Soc. 23(10),
1697–1704 (2003)
Biomechanical Properties of Coronary
Arteries Neonates: Preliminary Results

Normunds Sikora, Aris Lacis, Elina Ligere, Valts Ozolins,

Lauris Smits, Inta Bergmane and Vladimir Kasyanov

Abstract One of the most important issues in pediatric cardiac surgery is myo-
cardial protection when a cardioplegic solution mixed with oxygenated blood is
injected into coronary arteries with a pump. In this case, it is necessary to establish
the right pressure of the cardioplegic solution in coronary arteries taking into
account their biomechanical properties. Biomechanical properties of eight speci-
mens of coronary arteries from neonates 12.3 ± 13.7 days old and weight 4.1 ±
0.9 kg were investigated and compared with adult arteries. Specimens were
pressurized from 0 to 200 mmHg in steps of 20 mmHg while maintaining the
length of the sample in situ. We observed that the relationship between stress and
strain in neonates was non-linear. There was a rapid increase of strain until the
inner pressure reaches 80–100 mmHg and not as rapid regarding to the stress in the
arterial wall. When the internal pressure exceeds 100 mmHg the strain of the
arterial wall increases much slower but at the same time the wall stress and
modulus of elasticity begin to increase rapidly. It means that the structural ele-
ments of the arterial wall have been straightened and possible damage in the wall
of coronary arteries of neonates may appear. These results were compared with
biomechanical properties of arterial walls of adults and differences had been
found. Our first experimental results show that the pressure of the cardioplegic
solution in neonatal coronary arteries should not exceed 100 mmHg to decrease the
risk of structural damage of the vascular wall.

 
Keywords Biomechanics Biomechanical properties Neonatal coronary artery 
Pressure of cardioplegic solution

N. Sikora (&)  A. Lacis  E. Ligere  V. Ozolins  L. Smits  I. Bergmane  V. Kasyanov

Clinic for Pediatric Cardiology and Cardiac Surgery, Children’s University Hospital,
Vienibas 45, Riga, LV-1004, Latvia
e-mail: [email protected]; [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 111
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_9,
Ó Springer-Verlag Berlin Heidelberg 2012
112 N. Sikora et al.

1 Introduction

One of the most important issues in pediatric cardiac surgery is myocardial pro-
tection when a cardioplegic solution mixed with oxygenated blood is injected in
coronary arteries with a pump. The general principles underlying the efficacy of
cold chemical cardioplegic solutions are well established. Tissue preservation is
achieved by conserving energy stores through rapid arrest, decreasing the ongoing
metabolic rate, and combating changes induced by ischemia with specific pro-
tective agents. Besides the prevention of intraoperative myocardial damage
depends also on the completeness of delivery of cardioplegic solution. To maintain
an adequate delivery, the right delivery pressure should be used. If it is too low or
too high, it can lead to severe damage of the myocardium. Therefore, it is nec-
essary to establish the right pressure of the cardioplegic solution in coronary
arteries.
It is known that myocardial ischemia and reperfusion are associated with
unfavorable alterations in the physiologic mechanisms that regulate intracellular
and interstitial fluid balance. Certain aspects of these mechanisms, including the
Starling forces governing tissue fluid movement, lymphatic drainage and cell
membrane function, are altered, leading to the development of myocardial edema.
The composition of cardioplegic solutions (onconicity, hemodilution) and condi-
tions of delivery (hypothermia, high delivery pressure) are known to exaggerate
the development of edema resulting from ischemia or systemic inflammatory
responses.
Edema may increase microvascular resistance to a point of impending blood
flow and increase diffusion distance to myofibrils, leading to inadequate oxygen
delivery. It can arise with the use of cardioplegic solutions, especially in ischemic
myocardium, because of high delivery pressure, particularly in severely damaged
myocardium; hemodilution and hypo-osmolarity from crystalloid primes of car-
diovascular bypass or crystalloid cardioplegic solutions; physiologic changes in
ionic pump systems (i.e. Na–K-ATPase) or Donnan equilibrium for chloride ions
induced by hypothermia [1–3]. Although normal myocardium tolerates relatively
high infusion pressures, the myocardium within and surrounding ischemic seg-
ments is vulnerable to edema induced by high delivery pressure, because ischemia
not only causes a reduced autoregulation but also renders the capillary bed more
susceptible to inappropriately high pressure [4, 5].
There has been a number of researches done stating, how important is to
measure the delivery pressure of cardioplegic solution, when giving it into the
coronary arteries. It is often given at high flow and pressure following aortic
clamping to ensure rapid diastolic arrest, and it is easy to exceed a delivery
pressure of as high as 200 mmHg with standart setup in clinical practice (Fig. 1).
Irtun and Sorlie demonstrated the importance of cardioplegic solution delivery
pressure for myocardial protection in an in vitro rat heart study. They found that
moderate (106.5 and 73.5 mmHg) pressures gave good myocardial protection.
A considerably higher (175 mmHg) gave significantly higher coronary resistance
Biomechanical Properties of Coronary Arteries Neonates 113

Fig. 1 Standard setup for cardiopulmonary bypass and delivery system of cardioplegic solution
(The pump (1) takes oxygenated blood from the oxygenator (2) to the heat exchanger (3). There is
an Y-connector (4), where the cadioplegic solution (5) has been mixed with the oxygenated
blood. After the cooling of the blood in the heat exchanger (3) cold blood cardioplegia through
the line (6) has been delivered into the coronary arteries)

during the cardiac arrest period and reduced the coronary flow as well as left-
ventricle-developed pressure during the reperfusion period. The high delivery
pressure caused lower myocardial contents of adenosine triphosphate and creatine
phosphate at the end of reperfusion period [5–7]. In another study in vitro of pig
heart done by Irtun and Sorlie, they confirmed their results on rats and stated that
the delivery pressure of cardioplegic solution as high as 175 mmHg, which is
easily achieved in the aortic root with presently used cardioplegia delivery devices,
is harmful to the normal pig heart. They compared their model to a clinical setting,
stating that it is good reason to assume that harmful effects of high delivery
pressure also pertains to human clinical situations, as in many centers it is accepted
that the cardioplegia must be delivered quickly in order to induce a swift and
diastolic cardiac arrest without transitorial fibrillation and with, presumably,
maximum conservation of the high-energy nucleotids. Instead they showed that
even though this high delivery pressure gives a fast cardiac arrest, it gives a faster
breakdown of the high-energy nucleotids than a more physiological pressure. That
means that the myocardium would have less energy to work properly after aortic
occlusion and the contractility would be much worse [4].
It is known that autoregulation ensures approximately constant coronary flow, if
the coronary pressure is kept between 60 and 140 mmHg [4, 8]. The precapillary
sphincters regulate the vasomotion and this makes the capillary pressure nearly
constant. When they are gradually relaxed and the capillaries exposed to
inappropriately high pressures, this results in endothelial and myocardial damage.
114 N. Sikora et al.

There are also studies showing that not only high delivery pressure of the
cardioplegic solution may damage the myocardium. Aldea et al. show on a dog
model that at low perfusion pressures, not only its mean flow is reduced, but a
greater number of regions receive limited amounts of cardioplegic solution. These
observations may explain the patchy nature of subendocardial damage seen with
inadequate myocardial protection [9].

1.1 Neonates

There have been very few investigations done regarding the delivery pressure of
cardioplegic solution even in adults not to mention neonates. There are many
structural, functional, biomechanic and metabolic differences; therefore it is cru-
cial to give a cardioplegic solution with adequate delivery pressure, which still
remains undefined, especially in neonatal cardiac surgery [1].
It is widely accepted that the immature heart has a greater tolerance to ischemia
than the adult or mature heart [2, 3]. However, most of these results have been
obtained from normal hearts and it is unclear what the tolerance is when cyanosis
or accidosis are present. It has to be remembered that in majority of cases in
pediatric cardiac surgery the myocardium had been suffered from ischemia before
and is more vulnerable than the normal one. What is specific for neonatal
myocardium, it is more resistant to hypoxia, but less to the increased amount of
water, which is an issue for those using the crystalloid cardioplegia [5]. Besides it
stresses out how important is to clarify the right delivery pressure of the cardio-
plegic solution into the coronary arteries of neonates, because high pressure leads
to the development of myocardial edema.
Antegrade cardioplegia is often delivered without directly monitoring the
delivery pressure. The surgeon or perfusionist can only measure the actual per-
fusion pressure, which can result in cardioplegia being delivered at higher or lower
pressure than desired. Besides the optimal cardioplegia infusion pressure is still
unknown, especially in neonates. Due to the structural, functional, biomechanic
and metabolic differences, it may be more prone to a pressure injury in pediatric
cardiac surgery. Even though it is stated that high cardioplegic delivery pressure is
deleterious, especially to ischemic tissue, it still remains undefined [8, 10].
However, an adequate delivery pressure is still needed to ensure the distribution to
all areas of the myocardium [8, 9]. What is the right delivery pressure of cardi-
oplegia and what are the consequences of elevation of this pressure, especially in
the hypoxic heart, still remains an issue in neonatal cardiac surgery.
Another important issue, being investigated by Ishiyama et al., is the influ-
ence of coronary perfusion pressure on neonatal heart function. The change in
cardiac function following the change of coronary perfusion pressure or coro-
nary blood flow is called the Gregg phenomenon, well investigated in adults, but
not examined in neonates. During neonatal cardiac surgery, the coronary
Biomechanical Properties of Coronary Arteries Neonates 115

perfusion pressure depends on cardiopulmonary bypass, empirically kept low

compared to adult cases. They conclude that due to the immaturity of coronary
autoregulation neonatal hearts did not show it in any perfusion pressure range,
and both low and high perfusion pressure caused deterioration in the ventricular
function attributable to the immaturity of coronary autoregulatory capacity.
Therefore attention is needed to control the infusion pressure of the cardioplegic
solution [11, 12].
Cardioplegic infusion pressure must be adequate to ensure distribution to all
areas of the myocardium, but not as high as to cause cellular damage. The prin-
ciple is very simple, but what pressure is adequate and what causes myocardial
damage is still undefined. Even though most surgeons usually avoid excessively
high cardioplegic infusion pressures, the pressure must be also high enough to
ensure adequate myocardial distribution [9, 10]. Kronon et al. have investigated
the delivery pressure on an animal model and concluded that it is crucial in
neonatal cardiac surgery to maintain it under 100 mmHg. What is more, taking
into account, that majority of neonatal patients undergoing primary repair of
congenital heart disease are hypoxic, hypoxia profoundly alters the effect different
cardioplegic infusion pressures has on myocardium. The authors conclude that a
low cardioplegia infusion pressure does not protect the heart from further damage,
but allows the cardioplegia to facilitate repair of the injury caused by hypoxia and
reoxygenation, resulting in complete preservation of myocardium and vascular
endothelial function. They mention that this supports the safety of a cardioplegic
infusion pressure of 30–50 mmHg and implies it is high enough to ensure adequate
myocardial distribution, without what myocardial protection is poor. Conversely,
hypoxia alters the myocardium, resulting in an increased cellular injury, when the
cardioplegic infusion pressures were slightly higher (80–100 mmHg). This injury
manifests by post bypass myocardial and vascular dysfunction, increased edema
and decreased ATP levels [3, 10].

1.2 Biomechanics of Cardiovascular System

The mechanical properties of the arterial wall are very important because they
influence the arterial physiology. Furthermore, stresses and strains in the arterial
wall are extremely important factors in the understanding of the pathophysiology
and mechanics of the cardiovascular system [13, 14].
The primary function of all blood vessels is to circulate blood, but they are not
merely highways or pipes, as they have a passive capacitance function that
maintains the blood pressure in diastole as well as an active auto-regulatory
control that allows the organ to respond to local demands. Arteries are generally
subdivided in two types—elastic (e.g. aorta) and muscular (e.g. coronary artery).
The wall structure of both consists of intima, media and adventitia. The intima
layer consists of relatively acellular fibrous tissue and ground substance covered
116 N. Sikora et al.

by a monolayer of endothelial cells. The media is composed of multiple layers of

smooth muscle cells separated by collagen, ground substance and elastic fibres.
The adventitia has bundles of collagen and loose bands of elastic tissue. The
intima, media and adventitia are separated internal and external elastic laminae.
The mechanical properties of blood vessels depend on collagen and elastic fibres
as well as on smooth muscle cells and ground substances.
The pressure-diameter relation has been very popular, because it plays an
important role in the pressure-flow relationship of blood flow through the blood
vessel. The compliance of the vasculature (slope of the pressure-diameter relation)
is an important determinant of the non-linearity of the pressure-flow relationship.
Furthermore, the pressure-diameter-length can be transformed into biaxial (cir-
cumferential and longitudinal) stress–strain relation where the mean circumfer-
ential Cauchy stress is computed from pressure, diameter and wall thickness as per
Laplace’s equation and the strain is computed from circumference (or diameter)
measurements in reference to the zero-stress state [14–16].
The blood pressure is primarily opposed by the forces of elastin, collagen and
smooth muscle cells that are orientated to form defined layers. Thick elastin bands
form concentric lamellae while finer elastin membranes form networks between
lamellae, and the collagen fibres are distributed circumferentially in the interstices.
Kassab et al. has been observed that the wall thickness-to-radius ratio increases in
proportion to the increase in pressure such that the circumferential average wall
stress is restored after some period of growth and remodeling. In addition, they
found that the strain reaches its peak sooner and normalizes faster than stress. They
conclude that the vessel appears to be more sensitive to changes in strain than in
stress [16].
There are experiments showing that the physical nonlinearity of arterial
material is characterized by an increasing stiffness as strain increases. The original
of this behavior is found to be in the mechanical properties of the basic structural
components of the artery—elastin and collagen—as well as in the architecture of
the arterial wall as explained above [13].
As for neonates, it has to be taken into account that after birth coronary arteries
are immature. The wall of the vessels is much thinner comparing to the coronary
arteries of adults, influencing the biomechanical properties of the blood vessels.
These arteries are much more fragile and easier to damage. In early life, elastic
arteries increase in length, diameter and wall thickness in concert with the changes
in body weight and length associated with growth and development. The intima
becomes thicker due to the migration and proliferation of vascular smooth muscle
cells followed by synthesis of scleroprotein and extracellular matrix.
Knowing how important it is to have adequate myocardial protection in neo-
natal cardiac surgery and how easy it is to damage neonatal myocardium and
coronary arteries, when giving the cardioplegic solution into coronary arteries with
excessive delivery pressure, the aim of our study is to establish the pressure, which
is not harmful for neonates, taking into account the biomechanical properties of
coronary arteries of neonates.
Biomechanical Properties of Coronary Arteries Neonates 117

Fig. 2 The view of experimental strand

2 Materials and Methods

After receiving the permission of the Ethical Committee, between May and
December, 2009 eight samples of neonatal coronary arteries and one sample of
adult coronary artery, retrieved at the autopsies, were used as experimental
materials. The mean age of neonates was 12.3 ± 13.7 days and the mean weight
4.1 ± 0.9 kg. The length of the specimens was approximately 4 cm. After
resection the specimens were stored in Custodiol Perfusion Solution no longer than
24 h until the mechanical tests were performed. A special device was used to
measure the internal pressure, axial force, longitudinal and circumferential
deformation of the coronary artery (Figs. 2, 3).
One end of the artery was clamped to support to which a pressure transducer
and specially designed inductive force transducer were connected. The other end
was clamped to a support to which a pressure bottle containing fluid was con-
nected. The force transducer recorded the force necessary to maintain the vessel at
its in situ length. Axial stretch was introduced by a slide mechanism to which the
balance arms were fixed. The axial deformation of the artery was measured with a
specially designed inductive strain transducer connected to one of the arms of the
balance. Diameter changes in the specimen were sensed optically with a video-
dimensional analyzer coupled with a suitable lightning system for high contrast.
The changes in diameter were tracked and recorded continuously.
A sample of coronary artery was gradually loaded by internal pressure from 0 to
220 mmHg while maintaining the length of the sample constant at L0, the length in
situ. The pressure was elevated in 20-mmHg steps. The initial external diameter at
inner pressure p = 0 mmHg and at in situ axial length L0 was noted as D0. The
diameter D was recorded at each pressure level (Figs. 4, 5).
118 N. Sikora et al.

TV
Camera
Load Cell
Pressure Displacement
Transducer Transducer

Servo Motor
Out In

DC
Right Left
Handed Screw Specimen Chamber Handed Screw

Wafer Couple

Fig. 3 The scheme of experimental strand

Fig. 4 Neonatal coronary

artery pressurized with the
pressure of 0 mmHg

The value of wall thickness h was calculated as follows:

h ¼ h0 x k3; ð1Þ

where
1
k3 ¼ ; ð2Þ
ðk1 xk2 Þ
D
k2 ¼ ; ð3Þ
D0
The circumferencial stress was calculated as:
pxR
r¼ ; ð4Þ
h
Biomechanical Properties of Coronary Arteries Neonates 119

Fig. 5 Neonatal coronary

artery pressurized with the
pressure of 220 mmHg

Fig. 6 Determination of σ
elastic modulus
σ
E=
ε

80 mmHg
σ
60 mmHg
ε
ε

where p, inner pressure; R, radius.

 
L
k1 ¼ ¼ 1:0 ð5Þ
L0
In these equations, h0 is the initial thickness of the specimen and k1, k2 and k3
are, respectively, the stretch ratios in the axial, circumferential, and radial direc-
tions. Because the length of the artery was maintained constant at L0, the value of
k1 was 1. The initial wall thickness h0 was measured with a cathetometer
to ±0.0001 mm accuracy.
The flexibility and stiffness of arteries have been frequently characterized by
the values of compliance, pressure-strain elastic modulus and stiffness parameters.
Compliance is the fractional change in external diameter with change in pressure.
Because the pressure-diameter relation of an arterial wall is generally nonlinear,
compliance and pressure-strain elastic modulus are not usually material constants
but change with the internal pressure (Fig. 6).
120 N. Sikora et al.

250

200
Stress (kPa)

2
150

100

50
1

0
0 10 20 30 40 50
Strain(%)

Fig. 7 Relationship between stress and strain in coronary arteries; 1-adults, 2-neonates

The ultimate stress and ultimate strain were expressed as mean values plus a
standart deviation. A p value of less than 0.05 was considered statistically sig-
nificant. When two groups were compared, data were analyzed using student t test
with a p value of less than 0.05 indicating statistical significance.

3 Results

We observed that the relationship between pressure and strain, and stress and
strain in neonates was non-linear. There was a big difference between neonates and
adults in terms of elastic modulus (Fig. 7).
There was a rapid increase of strain until the inner pressure reaches 80–
100 mmHg and not as rapid regarding to the stress in the arterial wall. When the
internal pressure exceeds 100 mmHg the strain of the arterial wall increases much
slower but at the same time the wall stress and modulus of elasticity begin to
increase rapidly.
The strain in the wall of neonatal coronary artery, when the inner pressure is
80 mmHg, reaches 36.50 ± 3.75%, which is more than twice higher comparing to
the adult coronary artery -17.6%. As for stress, it reaches 68.61 ± 10.17 kPa with
the same inner pressure in neonatal coronary artery and increases rapidly, when
increasing the pressure (Table 1).
When it achieves 120 mmHg, the strain is 108.52 ± 15.19 kPa, which is more
than twice higher, comparing to the strain with inner pressure of 80 mmHg. The
strain in the wall of adult coronary artery was 45.84 kPa (p \ 0.05).
The stiffness in the wall of neonatal coronary artery increases rapidly com-
paring to adults, when the inner pressure exceeds 80 mmHg (Table 2).
The modulus of elasticity of the wall of neonatal coronary artery is
735.70 ± 319.54 kPa, when the inner pressure is 80 mmHg (Table 1), but when it
Table 1 Stress and strain in neonatal and adult coronary arteries, when applying different inner pressure
r (kPa) r (kPa) r (kPa) r (kPa) e (%)(60 mmHg) e (%) e (%) e (%)
(60 mmHg) (80 mmHg) (120 mmHg) (220 mmHg) (80 mmHg) (120 mmHg) (220 mmHg)
Neonates 49.02 ± 7.34 68.61 ± 10.17 108.52 ± 15.19 215.25 ± 36.67 33.35 ± 3.35 36.5 ± 3.75 40.35 ± 4.68 46.49 ± 9.88
Adults 21.15 28.81 45.84 87.84 16.41 17.62 0.01 23.78
Biomechanical Properties of Coronary Arteries Neonates
121
122 N. Sikora et al.

Table 2 Modulus of elasticity in neonatal and adult coronary arteries, when applying inner
pressure from 60 until 120 mmHg
E(kPa) (60–80) mmHg E (kPa) (80–100) mmHg E (kPa) (100–120) mmHg
Neonates 735.7 ± 319.54 1203.82 ± 460.22 1,462 ± 717.74
Adults 638.33 662.5 756.67

reaches 100–120 mmHg to 1,462 ± 717.74 kPa (p \ 0.05). In adults modulus of

elasticity in the same pressure is 756.67 kPa, which is twice lower comparing to
neonates. When the inner pressure increases from 60 to 120 mmHg, it increases
from 638.33 to 756.67 kPa (Table 2). In neonatal coronary artery it increases from
735.70 ± 319.54 to 1,462 ± 756.67 kPa. This can indicate the possible damage
in the wall of neonatal coronary artery, when the inner pressure exceeds
100–120 mmHg.

4 Discussion

One of the most important goal in pediatric cardiac surgery is to maintain an

adequate myocardial protection. It can be achieved by the cardioplegic solution
mixed with oxygenated blood injected in coronary arteries with a pump. The
general principles underlying the efficacy of cold chemical cardioplegic solutions
are well established. Tissue preservation is achieved by conserving energy stores
through rapid arrest, decreasing the ongoing metabolic rate, and combating
changes induced by ischemia with specific protective agents. The prevention of
intraoperative myocardial damage depends also on the completeness of the
delivery of cardioplegic solution. If it is too low or too high, it can lead to severe
damage of myocardium. Therefore it is necessary to establish the right pressure of
the cardioplegic solution in coronary arteries, because too low would lead to
inadequate delivery, but too high would damage the coronary arteries and myo-
cardium. In any case this injury manifests by postbypass myocardial and vascular
dysfunction, increased edema and decreased ATP levels [10]. The composition of
cardioplegic solutions (onconicity, hemodilution) and conditions of delivery
(hypothermia, high delivery pressure) are known to exaggerate the development of
edema resulting from ischemia or systemic inflammatory responses [3].
It can arise with the use of cardioplegic solutions, especially in ischemic
myocardium, because of high delivery pressure, particularly in severely damaged
myocardium. Although normal myocardium tolerates relatively high infusion
pressures, myocardium within and surrounding ischemic segments is vulnerable to
edema induced by high delivery pressure, because ischemia not only causes a
reduced autoregulation but also renders the capillary bed more susceptible to
inappropriately high pressure [4–7].
Biomechanical Properties of Coronary Arteries Neonates 123

There are a number of researches done [4–10, 12] stating, how important is to
measure the delivery pressure of cardioplegic solution, when giving it into the
coronary arteries. But these investigations have been done basically on different
animal models. There have been very few researches done regarding the delivery
pressure of cardioplegic solution even in adults not to mention neonates. Ante-
grade cardioplegia is often delivered without directly monitoring the delivery
pressure. The surgeon or perfusionist can only measure the actual perfusion
pressure, which can result in cardioplegia being delivered at higher or lower
pressure than desired. Besides the optimal cardioplegia infusion pressure is still
unknown, especially in neonates. Due to the structural, functional, biomechanic
and metabolic differences, it may be more prone to a pressure injury in pediatric
cardiac surgery. Even though it is stated that high cardioplegic delivery pressure is
deleterious, especially to ischemic tissue, it still remains undefined [8, 10].
However, an adequate delivery pressure is still needed to ensure the distribution to
all areas of the myocardium [8, 9]. What is the right delivery pressure of cardi-
oplegia and what are the consequences of elevation of this pressure, especially in
the hypoxic heart, still remains an issue in neonatal cardiac surgery.
There is almost no evidence, how important it is to maintain the right delivery
pressure of cardioplegic solution, taking into account the biomechanical properties
of neonatal coronary arteries. Therefore, we decided to establish the right pressure
after having investigated the samples of neonatal coronary arteries. Our results
show that in the wall of these arteries there was a rapid increase of strain until the
inner pressure reaches 80–100 mmHg and not as rapid regarding to the stress in
the arterial wall. When the internal pressure exceeds 100 mmHg the strain of the
arterial wall increases much slower but at the same time the wall stress and
modulus of elasticity begin to increase rapidly. It means that the structural ele-
ments of the arterial wall have been straightened and a possible damage in the wall
of coronary arteries of neonates may appear.
Our results show that the modulus of elasticity of the wall of neonatal coronary
artery appears to be two times higher comparing to adults, when the inner pressure
reaches 100–120 mmHg. It can indicate the possible damage in the wall of neo-
natal coronary artery, when the inner pressure exceeds 100–120 mmHg.

5 Conclusions

Our first experimental results show that, taking into account the biomechanical
properties, the delivery pressure of the cardioplegic solution in neonatal coronary
arteries should not exceed 100 mmHg, because higher increases the risk of
structural damage of the vascular wall, leading to the injury of myocardium. Our
research is going to be continued, including additional investigations of
morphology of arterial wall of neonates and adults.
124 N. Sikora et al.

References

1. Castaneda, A.R., Jonas, R.A., Mayer, J.E.: Myocardial preservation in the immature heart.
In: Castaneda, A.R., Jonas, A.R., Mayer, J.E., Hanley, F.L. (eds.) Cardiac Surgery of the
Neonate and Infant, 1st edn. W.B. Saunders Company, USA (1994)
2. Jones, T., Elliot, J.M.: Perfusion techniques. In: Stark, J.F., de Leval, M.R., Tsang, V.T. (eds.)
Surgery for congenital heart defects, 3rd edn. Wiley, England (2006)
3. Vinten-Johansen, J., Ronson, S.R., Thourani, V.H.: Surgical myocardial protection.
In: Gravlee, P.G., Davis, R.F., Kurusz, M., Utley, J.R. (eds.) Cardiopulmonary bypass, 2nd
edn. Lippincott Williams & Wilkins, USA (2000)
4. Irtun, O., Sorlie, D.: High cardioplegic perfusion pressure entails reduced myocardial
recovery. Eur. J. Cardiothorac. Surg. 11, 358–362 (1997)
5. Irtun, O., Sorlie, D.: Delivery pressure of the cardioplegic solution influences myocardial
protection. Eur. J. Cardiothorac. Surg. 9, 139–142 (1995)
6. Lindal, S., Gunnes, S., Ytrehus, K., et al.: Amelioration of reperfusion injury following
hypothermic, ischemic cardioplegia in isolated, infarcted rat hearts. Eur. J. Cardiothorac.
Surg. 4, 33–39 (1990)
7. Lindal, S., Gunnes, S., Lund, I., et al.: Ultrastructural changes in rat hearts following cold
cardioplegic ischemia of differing duration and differing modes of reperfusion. Scand.
J. Cardiovasc. Surg. 24, 213–222 (1990)
8. Buckberg, G.D., Beyersdorf, F., Kato, N.S.: Technical considerations and logic of antegrade
and retrograde blood cardioplegic delivery. Semin. Thorac. Cardiovasc. Surg. 5, 125–133
(1993)
9. Aldea, S.G., Austin, E.R., Flynn, A.E., et al.: Heterogeneus delivery of cardioplegic solution
in the absence of coronary artery disease. J. Thorac. Cardiovasc. Surg. 99, 345–353 (1990)
10. Kronon, M., Bolling, K.S., Allen, B.S., et al.: The importance of cardioplegic infusion
pressure in neonatal myocardial protection. Ann. Thorac. Surg. 66, 1358–1364 (1998)
11. Dole, W.P.: Autoregulation of the coronary circulation. Prog. Cardiovasc. Dis. 29, 293–323
(1987)
12. Ishiyama, N., Morita, S., Nishida, T., et al.: Different response in adult and neonatal hearts to
changes in coronary perfusion pressure. Pediatr. Cardiol. 27, 13–18 (2006)
13. Hayash, K., Stergiopulos, N., Meister, J.J. et al.: Techniques in the determination of
the mechanical properties and constitutive laws of arterial walls. M Properties (2001).
doi: 166.111.30.161
14. Ozolanta, I., Tetere, G., Purinya, B., et al.: Changes in the mechanical properties,
biochemical contents and wall structure of the human coronary arteries with age and sex.
Med. Eng. Phys. 20, 525–533 (1998)
15. Gupta, B.S., Kasyanov, V.A.: Biomechanics of human common carotid artery and design of
novel hybrid textile compliant vascular grafts. J. Biomed. Mater. Res. 34, 341–349 (1997)
16. Kassab, G.S.: Biomechanics of the cardiovascular system: the aorta as an illustratory
example. J. R. Soc. Interface 3, 719–740 (2006)
Biomechanical Properties of the Aorta
in Neonates and Infants

Elina Ligere, Normunds Sikora, Valts Ozolins, Lauris Smits,

Inta Bergmane, Aris Lacis and Vladimir Kasyanov

Abstract The study to examine biomechanical properties of the aorta in neonates

and the biomechanical properties of infantile aorta in case of different anastomoses
used for surgical correction of aortic coarctation was started to detect the influence
of the surgical technique on the biomechanics of infantile aorta and, therefore, the
possible further changes in hemodynamics and blood flow. We analyzed seven
specimens of native aorta and three specimens with the anastomosis end-to-end.
We observed a non-linear relationship between stress and strain in the neonatal and
infantile aorta. The strain of the end-to-end anastomosis was much smaller than in
the native aorta. The modulus of elasticity of the aortic wall increased with the
increase of inner pressure. However in the case of anastomosis end-to-end
the modulus was almost constant and was relevant to the modulus of elasticity of

E. Ligere (&)  N. Sikora  V. Ozolins  L. Smits  I. Bergmane  A. Lacis

Clinic for Pediatric Cardiology and Cardiac Surgery, University Hospital for Children,
Vienibas Street 45, Riga, Latvia
e-mail: [email protected]
N. Sikora
e-mail: [email protected]
V. Ozolins
e-mail: [email protected]
L. Smits
e-mail: [email protected]
I. Bergmane
e-mail: [email protected]
A. Lacis
e-mail: [email protected]
V. Kasyanov
Biomechanical Laboratory, Riga Stradins University, Riga, Latvia
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 125
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_10,
Ó Springer-Verlag Berlin Heidelberg 2012
126 E. Ligere et al.

the aorta with the inner pressure 100–120 mmHg. The results show that the
stiffness of the anastomosis did not change with the changes in inner pressure
which might affect hemodynamics.

Keywords Biomechanics  Mechanical properties  Neonatal aorta  Coarctation

1 Introduction

The incidence of congenital heart diseases is approximately 10/1000 in newborn

infants, but the coarctation of the aorta (narrowing of the descending aorta usually
just distal to the origin of the left subclavian artery) accounts for 8–10% of all
congenital heart defects. There is a prevalence of infantile type coarctation with
variable degree of the hypoplasia of aortic arch in infants. It leads to the development
of severe left heart failure after the closure of the ductus arteriosus with subsequent
circulatory shock, acidosis, renal insufficiency and death if left untreated.
The coarctation of the aorta might be associated with other congenital heart diseases
in up to 89% in neonatal age. Due to different anatomy and possible associated
anomalies, there are several methods of surgical correction of the disease. The
surgical correction of the coarctation of the aorta is the standard method in neonates
and infants (surgical techniques used are anastomosis end-to-end, subclavian flap
angioplasty and extended end-to-end anastomosis) [1, 2]. The results of primary
balloon angioplasty are debatable in early age and more associated with the risk of
aneurism formation, recoarctation and possible injury of femoral arteries and sub
sequent stenosis of femoral arteries. There is high incidence of recoarctation-up to
20–40% in cases of coarctation repaired within the first year of life with the need for
reinterventions- balloon angioplasty of recoarctation site. Coarctation of the aorta is
associated with increased risk of arterial hypertension in further life despite suc-
cessful repair and shortened life expectancy. According to the literature, an early
correction of the coarctation preserves the vessels of the postcoarctation zone from
structural changes but the precoarctation zone remains structurally changed with the
thickened intima and media and increased amount of collagen and elastin [1, 3–5].
Complications such as recoarctation or secondary hypertension, probably related to
the loss of arterial elasticity, frequently occur after aortic coarctation surgery. This
may be due to the fact that the aorta becomes enclosed in scar tissue, thus resulting in
future stenosis. Local loss of the natural aortic elasticity may be responsible for blood
pressure abnormalities. A hypothesis has been that increased arterial stiffness persists
despite satisfactory correction of the coarctation. There is an increased rigidity of
large arteries in children who had undergone surgery for coarctation of the aorta
compared with the patients without aortic disease. There are still considerable
debates regarding the best surgical approach of aortic coarctation repair [3]. There
are still lots of disputable questions according to the choice of the treatment and there
are almost no studies of biomechanical properties of the neonatal and infantile aorta.
Biomechanical Properties of the Aorta 127

The mechanical properties of the arterial wall in normal infants and in children after
the operation of coarctation influence arterial physiology and progression of arterial
disease via effects on blood flow [6].

2 Aim of the Study

To investigate the biomechanical properties of the aorta in neonates and infants

and to compare the biomechanical properties of the aorta in case of different
techniques of anastomoses.

3 Materials and Methods

The study was approved by ethics committee of The University Hospital for
Children. During the period of time from April 2009 till April 2010 we acquired 10
specimens (40 mm in length) of the upper part of the descending aorta during the
autopsies of neonates and infants (patients without the diagnosis of coarctation,
age of 2 days to 5 months, weight of 2.0–6.7 kg (mean weight 4 ± 1.6 kg). The
vessels were marked before resetting to identify the in situ axial extension. They
were preserved in Custodiol Perfusion Solution (CustodiolÒ HTK Solution is
intended for perfusion and flushing donor organs) not longer than 24 h at tem-
perature 2–48C. As a crystalloid solution, CustodiolÒ HTK Solution is considered
an intracellular solution, i.e., electrolyte concentrations similar to an intracellular
concentration, containing low concentrations of the electrolytes sodium, calcium,
potassium and magnesium and high concentration of an amino acid buffering
agent, histidine/histidine hydrochloride, the amino acid tryptophan a-ketoglutarate,
and the osmotic agent mannitol. A special device with video camera connected to
the laptop was used to measure the internal pressure, axial force, longitudinal and
circumferential deformation of the aorta (Fig. 1).
An aortic sample was gradually loaded by internal pressure from 0 to
200 mmHg while maintaining the length of the sample constant at L0, the length
in situ. The pressure was elevated in 20 mmHg steps with pressure held constant in
each step for 1 min. The initial external diameter at inner pressure p = 0 mmHg
and at in situ axial length L0 was noted as D0. The diameter D was recorded at each
pressure level. The value of the wall thickness h was calculated as follows:
h ¼ h0 x k3 ; ð1Þ

where
k3 ¼ 1=ðk1 x k2 Þ; ð2Þ

k2 ¼ ðD=D0 Þ; ð3Þ
128 E. Ligere et al.

Fig. 1 Common view of the

experimental set-up: 1–video
camera, 2–computer,
3–pressure transducer,
4–specimen (aorta)

Fig. 2 Schematic picture for σ

the calculation of the
modulus of elasticity
σ
E=
ε

80 mmHg

σ
60 mmHg
ε
ε

The circumferencial stress was calculated as:

r ¼ ðpxRÞ=h; ð4Þ

where p—inner pressure, R—radius.

k1 ¼ ðL=L0 Þ ¼ 1:0 ð5Þ
In these equations, h0 is the initial thickness of the specimen wall and k1, k2,
and k3 are, respectively, the stretch ratios in the axial, circumferential, and radial
directions. Because the length of the artery was maintained constant at L0 ; the
value of k1 ð¼ L=L0 Þ was 1. The initial wall thickness h0 was measured with a
cathetometer with ±0.001 mm accuracy. The artery was preconditioned before
tests by subjecting it to cyclic loading to bring it to a stable state, which could give
Biomechanical Properties of the Aorta 129

Fig. 3 Relationship between stress and strain (1-anastomosis end-to end, 2-native aorta)

a more reproducible mechanical response. During this process, the vessel was
pressurized from 0 to 200 mmHg in 20 steps five times with pressure held constant
for 1 min at each step. The initial curves were markedly hysteric, but the third or
fourth cycle gave reproducible curves with minimal hysteresis. A modulus of
elasticity E was calculated as incremental modulus between two values of internal
pressure (Fig. 2).
In the period of time from April 2009 to April 2010, we analyzed seven
specimens of native aorta and three specimens with anastomosis end-to-end. The
anastomoses were performed by the surgeon operating the patients with the
coarctation of the aorta using identical suture techniques

4 Results

We observed a non-linear relationship between stress and strain in neonatal and

infantile aorta (Fig. 3).
The strain of the aorta with an inner pressure of 60 mmHg is 25.15 ± 9.37%,
for 80 mmHg 29.37 ± 11.62%, for 100 mmHg is 32.63 ± 13.37% and the
maximal strain with inner pressure 200 mmHg is 47.65 ± 15.41%. The stress with
inner pressure 60 mmHg is 45.32 ± 15.29 kPa, for 80 mmHg 65.37 ± 24.2 kPa,
for 100 mmHg is 86.29 ± 33.55 kPa, and for 220 mmHg-253.41 kPa. In the case
of anastomosis end-to- end, the relationship between stress and strain is linear. The
strain of the anastomosis decreases and with the inner pressure 60 mmHg it is
9.52 ± 5.01%, for 80 mmHg 11.01 ± 4.93%, for 100 mmHg 12.3 ± 5.01%, but
at the level of sutures, the maximum strain with an inner pressure of 220 mmHg
reaches only 19.58% ± 7.06% (Table 1).
 130

Table 1 Strain and the modulus of elasticity of aorta and the zone of anastomosis end-to-end with the changes of inner pressure
r (kPa) r (kPa) r (kPa) r (kPa) r (kPa) e (%) e (%) e (%) e (%) e (%) E (kPa) E (kPa) E (kPa)
(60 (80 (100 (120 (220 (60 (80 (100 (120 (220 mmHg) (60–80 (80–100 (100–120
mmHg) mmHg) mmHg) mmHg) mmHg) mmHg) mmHg) mmHg) mmHg) mmHg) mmHg) mmHg)
Native 45.32 ± 65.37 ± 86.29 ± 108.04 ± 253.41 ± 25.15 ± 29.37 ± 32.63 ± 34.93 ± 47.65 ± 516.08 ± 704.02 ± 967.45 ±
aorta 15.29 24.2 33.55 42.72 95.94 9.37 11.62 13.37 14.43 15.41 126.21 170.67 164.60
Anasto 36.15 ± 49.52 ± 63.41 ± 78.35 ± 158.53 ± 9.5 2 ± 11.01 ± 12.3 ± 13.73 ± 19.58 ± 902.39 ± 1138.16 ± 1151.29 ±
mosis 2.85 4.01 7.28 6.81 22.23 5.01 4.93 5.01 5.51 7.06 86.49 216.23 309.13
end-to-end
P 0.26 023 0.23 0.22 0.08 0.046 0.048 0.05 0.047 0.017 0.012 0.043 0.24
E. Ligere et al.
Biomechanical Properties of the Aorta 131

Comparing the strain of native aorta with anastomosis we come to conclusion

that the strain of the anastomosis is much smaller and with the inner pressure 220
mmHg it is twice as small as in native aorta. The modulus of elasticity of the aortic
wall increases with the increase of inner pressure, but in the case of anastomos it is
almost constant and is relevant to the modulus of elasticity of the aorta with the
inner pressure 100–120 mmHg. The results show that the stiffness of the anasto-
mosis does not change with the changes in inner pressure which might affect
hemodynamics.

5 Discussion

Arterial stiffness has been shown to be an independent risk factor for cardio-
vascular events such as primary coronary events, stroke and mortality [3, 5].
Several clinical trials have shown that even normotensive subjects after success-
fully repaired coarctation have markedly increased central aortic stiffness and
decreased central aortic dispensability and compliance as well as significantly
increased pulse wave velocities in echocardiographic and magnetic resonance
studies which are subsequently associated with increased left ventricle mass
despite successful early repair [3, 5, 7]. Our first data show a significant difference
between the strain of native aorta and that of the anastomosis, being strain of the
anastomosis much smaller. The modulus of elasticity of the aortic wall increases
with the increase of inner pressure, but in the case of anastomosis these changes
are limited.
According to the age specific percentile table the arterial pressure in neonates
within the first week is 58/37(mean 48 mmHg) (25th percentile) up to
71/45 mmHg (mean 58 mmHg) (75th percentile), at the age of 2–3 weeks 71/42
(mean 56 mmHg) to 84/63 mmHg (mean 73 mmHg) and at the age of 1–5 months
88/53 mmHg (mean 62 mmHg) to 102/64 mmHg (mean 83 mmHg) [1]. The
differences between the biomechanical qualities of the native aorta and that of the
anastomosis end-to-end are less prominent within the ranges of physiological
arterial pressures, but more obvious with the increase of inner pressure. These
findings are consistent with the findings of elevated arterial pressures during
physical activities in otherwise normotensive subjects following the successful
coarctation repair.
There are almost no studies on the biomechanical qualities of human neonatal
and infantile aorta. There are several previous studies on animals where mor-
phological, structural and biomechanical changes of the aorta during development
as well as the relation between these changes and the mechanics of aorta have been
investigated, where mice models show greater changes in arterial pressures within
first two weeks of life (mean pressure increases from 30–80 mmHg) which are
associated with aortic wall thickening during the development. However these
studies on animals could not be completely attributed to relevant results in neo-
nates and infants [8].
132 E. Ligere et al.

The present study shows our first data where native infantile aorta and the aorta
with anastomosis end-to-end are investigated. Our further aim is to investigate and
compare the biomechanical properties of different types of anastomoses with the
native aorta. These differences may also further influence the possibilities of
successful balloon angioplasty in the case of recoarctation.
Limitations of the study is that the in vitro study maintain a constant longitu-
dinal stretch ratio during inflation, which is not physiological, and there is only a
small number of specimens investigated with consequent need to implement in a
larger group.

6 Conclusions

The first experiments showed that neonatal and infantile aorta has significant strain
properties which provide physiologic hemodynamics. During the operation with
the anastomosis the formation of local changes appear which can be the reason of
residual gradient and recoarctation. We have to prolong the study to investigate the
changes in biomechanical properties with the different types of anastomosis to get
statistically significant data and to compare the data with the data from clinical and
echocardiographical study of the children with coarctation corrected within the
first year of life.

References

1. Park, M.K., Troxler, G.: Pediatric Cardiology for Practitioners, 4th edn. Mosby Elsevier,
Philadelphia (2008)
2. Lacis, A., Volokolakovs, J.: Coarctation of the aorta- necessity for the operation and
perspectives in the early age group. Latvijas Ārsts 1, 404–415 (1994)
3. Bassareo, P.P., Marras, A.R., Manai, M.E., Mercurio, G.: The Influence of Different Surgical
Approaches on Arterial Rigidity in Children After Aortic Coarctation Repair. Pediatr. Cardiol.
30(4), 414–418 (2009)
4. Sehested, J., Baandrup, U., Mikkelsen, E.: Different reactivity and structure of the prestenotic
and poststenotic aorta in human coarctation. Circulation 65, 1060–1065 (1982)
5. Ou, P., Celermajer, D., Jolivet, O., Buyens, F., Herment, A., Sidi, D., Bonne, D., Mousseaux, E.:
Increased central aortic stiffness and left ventricular mass in normotensive young subjects after
successful coarctation repair. Am. Heart J. 155(1), 187–193 (2008)
6. Hayashi, K., Stergiopulos, N., Meister, J.J., Greenwald, S.E., Rachev, A.: Techniques in the
determination of the mechanical properties and constitutive laws of arterial walls, in
biomechanic systems and applications. In: Leondes, C.T. (ed.). Gordon and Breach, Tokyo,
pp. 6.1–6.61 (2001)
7. Ou, P., Bonnet, D., Auriacombe, L., et al.: Late systemic hypertension and aortic arch
geometry after successful repair of coarctation of the aorta. Eur. Heart J. 25, 1853–1859 (2004)
8. Kassab, G.H.: Biomechanics of the cardiovascular system: the aorta as an illustrator example.
J. Royal Soc. Interface 3, 719–740 (2006)
 Part III
Fluid and Gas
Numerical Study of Blood Flow Pressure
Drop in Aorta Coronary Sinus Conduit

Siti Aslina Hussain, Tan Hong Tat, Mohd Ismail Abdul Hamid,
Norhafizah Abdullah and Azni Idris

Abstract Numerical studies of the blood flow system of aorta coronary sinus
conduit were carried out using ANSYSTM CFD simulation. The pressure inside the
conduit was investigated to ensure a pressure drop from 80 to 15 mmHg. It was
aimed to model a coronary sinus conduit in three-dimension using ANSYSTM
CFD. The simulation involved pre-modeling, modeling and simulation stages
where the model will undergo each section of program in ANSYSTM CFD such as
design modeler, meshing, pre-processing, solver and post-processing. From the
analysis of coronary sinus conduit, it is found that a narrow tube needs to be
incorporated into the conduit produce. This is to induce a venturi effect to reduce
the pressure of blood from 80 to 15 mmHg within a specific throat length. A model
of 3 mm inlet and throat diameter of 1.13 mm throat diameter show the best result
for pressure reduction from 80 to 15 mmHg. The model gives a uniform pressure
drop along the throat section of the conduit.

S. A. Hussain (&)  T. H. Tat  N. Abdullah  A. Idris

Department of Chemical and Environmental Engineering, Faculty of Engineering,
Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia
e-mail: [email protected]
T. H. Tat
e-mail: [email protected]
N. Abdullah
e-mail: [email protected]
A. Idris
e-mail: [email protected]
M. I. A. Hamid
211, Jalan Palas, Southern Park, 41200 Klang, Selangor, Malaysia
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 135
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_11,
 Springer-Verlag Berlin Heidelberg 2012
136 S. A. Hussain et al.

 
Keywords Computational fluid dynamics Blood flow Pressure drop simulation 

Aorta coronary sinus conduit Coronary artery bypass

Nomenclature
Symbol Definition (unit)
c_ Strain rate (s-1)
‘ Length of tube (m)
D Diameter (m)
Dm Major diameter (m)
Dn Minor diameter (m)
K Consistency index (Dimensionless)
L Length (m)
m Plane number (Dimensionless)
n Power law index (Dimensionless)
N Number of part of the tube (Dimensionless)
P1 Pressure at point 1 (Pa)
P2 Pressure at point 2 (Pa)
Q Volumetric flow rate (m3/s)
Re Reynolds Number (Dimensionless)
T Temperature (K)
To Reference temperature (K)
V Velocity (m/s)
a Angle of bending in radian (rad)
k Friction coefficient (Dimensionless)
l Viscosity (Pa s)
q Density (kg/m3)

1 Introduction

1.1 Physiology of Coronary Venous System

Coronary artery bypass grafting (CABG) was first developed between 1967 and
1968. Its safety and immediate benefits are well established [1–3]. However there
are a subset of patients who have diffuse coronary artery disease and are symp-
tomatic. These patients show global cardiac enlargement with diffuse myocardial
involvement. Coronary atherosclerosis in these patients is severe and involves
multiple vessels. Prognosis is grave with a five and seven year survival rate of 45
and 34%, respectively [4].
The coronary venous system is a dense network whose volume is significantly
greater than the arterial vasculature. The system is spared in coronary athero-
sclerosis and indeed in chronic disease states, its growth is enhanced. Ratajczyk-
Pakalska [5] and associates have demonstrated that there are a larger number of
Numerical Study of Blood Flow Pressure Drop 137

Thebesian veins and other venous channels in patients with coronary atherosclerosis
with 70% of these patients having increased coronary arteriovenous shunts.
Baroldi and Scomazzoni [6] noted that there was widespread anastomosis
between the vessels of the heart and these were marked with the arterial system of
the heart. In summary except perhaps for the right marginal vein which may drain
directly into the right atrium, all the other veins of the heart including the great,
middle and small cardiac veins, and the posterior vein of the left ventricle drain
into the coronary sinus.
In the 1890s, Pratt reported his initial experimental observations of retroper-
fusion of arterial blood through coronary veins in isolated canine and feline hearts
[7]. Beck in 1948 operated on patients with diffuse coronary atherosclerosis by
connecting the great cardiac vein with the ascending aorta using a saphenous vein
and simultaneously ligating the coronary sinus orifice. This operation was known
as the Beck II operation [8, 9].
In his experiments, Beck’s greatest technical setbacks were thrombosis and
occlusion of the anastomosis. Some late deaths could also have possibly been due
to the fact that the coronary sinus was subjected to diastolic blood pressures, which
could have exceeded the safe coronary sinus pressure limits of 30–40 mmHg [8].
This is reflected when he noted in later experiments that dogs that had anastomosis
between the aorta and coronary sinus frequently died of heart failure while those
with a carotid artery—coronary sinus anastomosis survived. He discussed the
possibility of ‘‘delivering too much blood to the heart’’.
Experimental and clinical data suggest that methods such as pressure controlled
intermittent coronary sinus occlusion (PICSO) and synchronized retroperfusion
(SRP) can reduce ischemia, salvage jeopardized myocardium and can reduce infarct
size [10–12]. Today, retrograde coronary sinus cardioplegia delivered through the
coronary sinus is attributed to produce better myocardial protection because of more
homogeneous distribution of cardioplegia [13, 14]. With these facts in mind, a
conduit to deliver blood between the aorta and coronary sinus was developed where
blood delivered through the coronary venous vasculature would remain at appro-
priate low pressures without causing endothelial damage to the system.

1.2 Application of Numerical Method for Blood

Flow Simulation

A conduit is to be designed for the purpose of connecting the aortic root and
coronary sinus, transatrially. By harnessing the diastolic pressure at the aortic root,
sufficient blood could be delivered to the coronary venous system which supply the
myocardium section. The diastolic pressure can vary anywhere between
60–100 mmHg. The conduit is designed such that the targeted outlet pressure is
15 mmHg whilst the inlet pressure is tailored according to the diastolic pressure
(as it is during the perfusion of the coronary arterial system takes place). It is
138 S. A. Hussain et al.

preferable to obtain a pressure drop from 80 mmHg at the inlet of the conduits to
15 mmHg at the outlet of the conduit which designing the conduit.
The sizing of the conduit is designed to accomodate the morphological structure
of the heart. The demonstration of the intra atrial blood flow was modelled and
simulated using computational fluid dynamic (CFD). CFD is a non-invasive
technique which can provide details on haemodynamic information. The infor-
mation is then used to generate numerical meshes for CFD simulation which
provide detailed 3-dimensional illustrations of velocity and pressure distributions
in the right artrial cavity.
The computational fluid dynamics technique was chosen in the simulation of
blood flow because it is one of the branches of fluid mechanics that uses numerical
methods and algorithms to solve and analyze problems that involve fluid flows. Its
applicability can be exploited for solving coronary blood flow problems. This
method can render relatively accurate results of flow system in which numerous
amount of calculations are performed computationally as per the requirement to
simulate the interaction between fluids with the complex surfaces used in engi-
neering applications.
ANSYSTM CFX software was selected to simulate the blood flow profile in the
aorta coronary sinus system. This software is based on the finite element method
with the application of the Navier–Stokes equations in the analysis of the flow
model. For solving the partial differential equations derived from the Navier–
Stokes equations, a numerical method approach was applied.

1.3 Theory on Blood Flow Pressure

Fluid dynamics is the branch of engineering that focuses on the fluid behaviour.
The application of fluid dynamics can be further extended to study the blood flow
behaviour in the cardiovascular system.
Poiseuille’s equation [15] is used for the analysis of the pressure drop in conduit
given that the flow is laminar. The rate of fluid flow in a conduit depends on the
fluid viscosity, pressure difference and the dimension of the conduit. For laminar,
non-pulsatile fluid flow through a uniform straight pipe, the flow rate is given by
Poiseuille’s equation as follows:

p  ð0:5DÞ4 ðP1  P2 Þ
Q¼ ð1Þ
8lL
where Q is the volumetric flow rate in m3/s; D is the tube diameter in meter; P1–P2
is the pressure difference between the tube ends in Pascal; L is its length in meter;
l is the coefficient of viscosity in Pa s.
However, when applying Poiseuille’s equation in the blood flow system in
coronary sinus, there are few uncertainties. First, the flow is essentially laminar
outside of the capillaries but it is definitely pulsatile throughout the arterial
Numerical Study of Blood Flow Pressure Drop 139

subsystem. Second, Poiseuille’s equation assumes a constant viscosity, whereas

the viscosity of blood actually changes with velocity, since blood is not a non-
newtonian fluid. Third, the coronary sinus is not a straight and uniform conduit.
Regardless, Poiseuille’s equation is used as an initial approximation for computing
the necessary dimensional conduits parameters in the cardiovascular system.
Reynolds number has a considerable importance in determining the type of flow
in a conduit. The Reynolds number is the ratio of inertial forces to viscous forces
and, consequently, its magnitude can be used to identify and predict different flow
regimes, such as laminar or turbulent flow [16]. Reynolds number is given by:
qVD
Re ¼ ð2Þ
l
where l is the fluid viscosity in Ns/m2; V is the fluid velocity in m/s; D is the
conduit diameter in meter; q is the fluid density in kg/m3.
Laminar flow, sometimes known as streamline flow, occurs when a fluid flows in
parallel layers, with no disruption between the layers. In fluid dynamics, laminar flow
is a flow regime characterized by high momentum diffusion, low momentum con-
vection, pressure and velocity independent from time. Reynolds numbers of less than
2,100 are generally considered to be of a laminar type. Turbulence is a flow regime
characterized by chaotic, stochastic property changes. Reynolds number above about
4,000 is identified as turbulent flow. The range between 2,100 and 4,000 is usually
known as the transition range. The typical value of Reynolds number for flow in
coronary sinus normally lies within the laminar region in the order of 103.
The difference of viscosity also will affect the value of Reynolds number
greatly. It is to be noted that the viscosity of blood is not constant. Blood is a non-
newtonian fluid. This means that the viscosity changes with the applied strain rate.
A power law is often used to define the viscosity of blood [17]. The power law for
viscosity of blood is defined as:

l ¼ kc_ n1 eTo=T ð3Þ

where k is the consistency index, n is the power law index, To is the strain rate in
Kelvin, c_ is the rate of shearing strain in s-1.
The power law index usually indicates the fluid behaviour. If n = 1, the fluid
behaves as newtonian fluid. For n less than 1, the fluid is pseudoplastic while for
n more than 1, it is dilatant. Figure 1 shows the relationship between shear stress
with sheer rate for pseudoplastic, dilatant and newtonian fluid.
Apparently, blood exhibits a pseudoplastic behaviour, with its apparent
viscosity decreases as the shear rate increases as in Fig. 1. However, when the
shear rate increases above 100 s-1, blood starts to behave as a Newtonian fluid
[17]. Hence, there exists a maximum and minimum value for viscosity of blood.
The typical range of the viscosity of blood is from 1.25 to 3.5 cP.
The insights of pressure drop within a conduit need to be understood.
The pressure drop in conduit is usually caused by friction, vertical height difference
140 S. A. Hussain et al.

Fig. 1 Shear stresses versus

rate of shearing strain [22]

and changes of kinetic energy. The pressure drop along the circular conduit with
constant diameter can be explained using the following equation:
kLqV 2
DP ¼ P1  P2 ¼ ð4Þ
2D
where DP is the pressure drop in Pascal; k is the conduit friction coefficient;
D is the conduit diameter in meter; L is the conduit length in meter; V is the flow
velocity in m/s and q is the fluid density in kg/m3.
In order to determine the pressure drop, the Reynolds number and friction
coefficient value need to be calculated. Assume that all the flow in coronary
circulation is laminar, the conduit friction coefficient can be determined using the
following equation
64
k¼ ð5Þ
Re
where k is the conduit friction coefficient; and Re is the Reynolds number (as
derived in Eq. 2).
In this blood conduit, the wall roughness, e is considered negligible by assuming
that the conduit surface is slip free. With the assistance from Eq. 1 through Eq. 5,
the model structure of the artificial coronary sinus is sketched with important
parameters such as length and diameter of the conduit can be estimated. The
equations will also be used in validating the simulation results mathematically.

2 Methodology

2.1 Derivation of Design Equations

To design a narrow sub-section portion of the internal structure, the plane diameter
is substituted with the diameter of which agrees in accordance to Eq. 1 which
Numerical Study of Blood Flow Pressure Drop 141

Fig. 2 Mathematical model sketch

Table 1 Explanation of Symbol Description

symbols used in Fig. 2
a Angle of conduit bending in radian
‘ Length of conduit
N Number of part of the tube
D‘ Change of conduit length over part
of the tube and is given as D‘ ¼ N‘
r Conduit bending radius and given as r ¼ a‘
b Angle of part of the tube and is defined
as b ¼ D‘ a
r ¼ N
Dm Major cross sectional diameter of the conduit
Dn Minor cross sectional diameter of the conduit
m Plane number (1, 2, 3, …, N ? 1)

lasted for a predefined length. From Eqs. 11 to 13, the value is defined as
OFFSET-X, OFFSET-Y and OFFSET-Z respectively. This value of offsets is used
to create a new plane from the reference of XY plane as origin.
The mathematical derivation sketch of the conduit is illustrated in Fig. 2.
The variables given in Fig. 2 are explained in Table 1.
A, B, C, D and E are the conduit design parameters as defined by the following
expression.
‘
A ¼ r  cos a ¼  cos a ð6Þ
a
‘
A  cos a
B¼ ¼ a a ð7Þ
cos b cos N
142 S. A. Hussain et al.

  "‘ #
‘ a  cos
 
a
C ¼rB¼  ð8Þ
a cos Na
(  " #)
‘
‘ a  cos
a
a a
D ¼ C  cos b ¼   cos ð9Þ
a cos N N

‘
E ¼ r  sin a ¼  sin a ð10Þ
a

The offset is defined as the displacement of sketching plane from the origin.
The three offset of x, y and z plane is described as the following equations:
2   3
  ‘ ‘
acos a
‘ 
OFFSETx ¼  cos a þ 4 a  cosðNa ðm1ÞÞ  5 ð11Þ
a
 cos Na  ðm  1Þ

OFFSETy ¼ 0 ð12Þ

‘ ð m  1Þ
OFFSETz ¼  sin a  ð13Þ
a N

2.2 Modelling

The design values calculated in Sect. 2.1 are used for creating a conduit model
by using ANSYSTM Design Modeler (DM). A narrower sub-section tube is
incorporated into the larger external tube as shown in the proposed model in
order to achieve the desired pressure drop. The specification of the inlet and
outlet diameters is fixed. Therefore, the diameter of the intermediate tube
section needs to be varied and last for a certain length for a satisfactory
pressure drop of about 65 mmHg. The internal structure of the tube is as shown
in Fig. 3.

2.3 Meshing

After a solid model (Fig. 3) of the flow domain has been created, meshing of the
solid flow domain was constructed in ANSYSTM Meshing. In this stage, the
domain was divided into many small cells. Meshing is an essential processing
stage as it provides a surface representation for a complex geometry with a few
basic geometry primitives. The mesh properties are summarized in Table 2.
Numerical Study of Blood Flow Pressure Drop 143

Fig. 3 a External and b Internal structure of the purposed coronary sinus conduit

2.4 Processing

The meshed model was processed for further analysis. The whole process at this
stage is done under ANSYSTM Advance CFD function. The inlet, outlet and wall
condition for the model are specified in detail in Table 2. The result was obtained
from the parameters entered into the processor. This is displayed as a compre-
hensible graphical representation. The pressure drop and velocity profile of the
blood flow through the conduit model that differ in their inlet diameter were
investigated. Here, the inlet diameter was set at 3, 4 and 5 mm for models 1, 2 and
3, respectively. The crucial parameters for these models are specified in Table 2.

3 Results and Discussion

3.1 Pressure

The aim for designing the blood conduit is to achieve a pressure drop from
diastolic pressure of the venous system from 80 to 15 mmHg, a safe limit for
coronary sinus perfusion without causing any endothelial disruption. The conduit
outlay is depicted in Fig. 4. The fluid entered the conduit at 80 mmHg (which is
marked as the maximum pressure in red colored section of the Fig. 5), which
simulate the coronary sinus drainage. The fluid pressure started to decrease along
the throat section to a magnitude of 15 mmHg at the outlet of the throat. This is
due to the venturi effect but note that the conduit is not straight and the diameter of
the inlet and outlet of the conduit is not uniform. The theoretical value for min-
imum throat diameter which can be designed for the purpose of blood flow was
calculated at 1.13 mm. The value was obtained based on simple mathematical
144 S. A. Hussain et al.

Table 2 Specifications and parameters data for simulation for three models
Parameter Model 1 Model 2 Model 3
Fluid properties
Equation for non-Newtonian Power law Power law Power law
viscosity of blood r = kcn-1eT/T r = kcn-1eT/T r = kcn-1eT/T
Power law index, n 0.4851 0.4851 0.4851
Consistency index, k (kg sn-2/m) 0.2073 0.2073 0.2073
Reference temperature (T) 37 37 37
Maximum viscosity limit, rmax 0.00125 0.00125 0.00125
(kg/m s)
Minimum viscosity limit, 0.003 0.003 0.003
rmin (kg/m s)
Reference pressure (atm) 1 1 1
Fluid temperature (T) 37 37 37
Turbulence Laminar Laminar Laminar
Density (kg m-3) 1050 1050 1050
Model properties
Minor diameter, Dn (mm) 3 4 5
Major diameter, Dm (mm) 15 15 15
Tube length, ‘ (cm) 11 11 11
Bending degree, a () 30 30 30
Part number, N 10 10 10
Throat diameter, DT (mm) 1.13 1.15 1.17
Throat length, ‘T (cm) 4 4 4
Mesh properties
Total number of nodes 77704 77926 78297
Total number of tetrahedra 415245 416515 418579
Total number of elements 415245 416515 418579
Maximum spacing (mm) 0.5 0.5 0.5
Angle resolution (degree) 30 30 30
Minimum edge length (mm) 0.01 0.01 0.01
Maximum edge length (mm) 0.8 0.8 0.8
Simulation properties
Inlet
Mass flow rate (kg.s-1) – – –
Pressure (mmHg) 80 80 80
Outlet
Mass flow rate (kg.s-1) 0.004375 0.004375 0.004375
Pressure (mmHg) – – –
Wall Free slip Free slip Free slip
No. Iteration 100 100 100

calculations, in which by taking the maximum size of a red blood cell is about
8 lm and with a diameter of 1.13 mm, it allows about 141 red blood cells to flow
parallel in the throat section. Downstream of the throat section, the pressure is
stabilised at 15 mmHg.
Numerical Study of Blood Flow Pressure Drop 145

Fig. 4 Schematic 3D
representation of conduit

The external pressure profile throughout the conduit length is shown in Fig. 5.
At the throat section (0.095–0.090 m) an observable pressure drop to -104 mmHg
is detected. The external wall pressure is important for design purpose, allowing
information whether the material of construction can withstand the significant
pressure drop/increase. In particular, for thin walled conduit design, a collapsible
material is not desirable over a more structure material for the conduit.
Figure 6 shows that the pressure at the inlet zone is uniform at about 80 mmHg.
At the throat section, a major difference is noticeable. In the core of the throat
section, the fluid pressure decreases steadily from 80 to 15 mmHg. In contrast, the
wall of the throat having a lower pressure which is part of the venturi effect.
The condition after the outlet section of the throat remains almost constant due to its
increasing conduit diameter, which produces little change in pressure difference.
Near the wall of the conduit in the throat region, the pressure value is less than
in the core region of the conduit. This is due to high velocity flow causing skin
pressure to appear along the wall of the throat. This is significant when fabricating
the conduit material, in which a thicker wall is desired, to prevent compression
from occurring in the region.
As comparison to other models, a plot of pressure distribution along the length
of the conduit is presented in the graph as shown in Fig. 7. By analytical analysis,
in Fig. 7, the slope of model 1 seems to be smoother than the other models. This
proves that model 1 experiences a steady decrease in pressure along the throat
section. In addition, model 1 gives an outlet pressure of 15 mmHg, which is in
accordance to the specification.
Model 3 shows a drastic pressure drop at the throat section (0.9–0.95 m), which
is undesirable as it may cause turbulence in the fluid flow. Model 1 (shown as pink
plot) does not show any sudden pressure drop, i.e., the drop in pressure is gradual
and hence is preferable and unlikely to cause turbulence. Model 2 is an inter-
mediate state of Models 3 and 1.
The generic trend of the pressure profile for all three models are, pressure enter
at 80 mmHg, stable for a certain length of the throat, then start to drop along the
146 S. A. Hussain et al.

Fig. 5 Contour plot on the external of conduit in model 1

Fig. 6 Pressure distribution contour plot inside conduit of model 1

throat (from 0.08 to 0.05 m) to a pressure of 15 mmHg. Beyond the throat, the
pressure is stable at 15 mmHg until the conduit exit.

3.2 Velocity

From the viewpoint of velocity distribution, it is noticed that the velocity is

stationary at 0 m/s along the wall. This is due to the no slip condition for the wall
boundary condition. The entrance region of the throat was observed to have a
Numerical Study of Blood Flow Pressure Drop 147

Fig. 7 Pressure distribution comparison for all models

Fig. 8 Contour plot of velocity for model 1

maximum velocity of about 6.5 m/s, as shown in Fig. 8, which can be explained
by noting the fluid was all bottlenecked into a narrower conduit and thus
contributing to higher velocity at the particular throat entrance zone.
The general profile which illustrates the velocity of the fluid flow along the
conduit shows a higher velocity at the throat section compared to the inlet and
outlet region. This corresponds to the venturi effect, in which a constant mass flow
of the fluid will speed up when it flows through a smaller region of the conduit.
A sharp increment in fluid velocity was observed at the throat region. This
corresponds well with the pressure drop profile of this region. The velocity slows
148 S. A. Hussain et al.

Fig. 9 Velocity distribution comparison for all models

Fig. 10 Graphical representation of pressure and velocity distributions for model 1

down to 5 m/s along the throat region, followed by dropping to slightly above 0 m/s
along conduit exit section.
The velocity distribution comparison shown in Fig. 9 give different peak
velocities for different models. A higher velocity will contribute to a higher
Reynolds number and causes a low pressure at the wall. This will hence cause the
wall need to withstand more pressure force caused by the atmosphere. Thus, it is
Numerical Study of Blood Flow Pressure Drop 149

important to have a lower peak velocity as much as possible. From the comparison
in Fig. 9, model 1 is preferable due to its low peak velocity.
As a conclusion, model 1 was selected as the best model. This is due to its
characteristic of stable pressure reduction along the throat and low peak velocity as
shown in Fig. 10. In addition, model 1 also provides a more uniform outlet
pressure of 15 mmHg, which is in accordance with the required specification.

4 Conclusion

In conclusion for the study of the blood flow system in the aorta coronary sinus
conduit by numerical method utilizing the concept of a throat in the design, all the
models successfully reduced the pressure from 80 mmHg at the inlet to 15 mmHg
at the outlet. The pressure reduction effective zone is mainly on the throat section.
This section was designed specifically to undergo the venturi effect as the fluid
flow pass through it. However, model 1 with 3 mm inlet diameter and 1.13 mm
throat diameter was chosen as the best model because the uniformity of the
pressure reduction along the throat section.
This particular design shows that this model could be utilized to achieve the
pressure drop necessary to prevent coronary venous vasculature damage. Although
the coronary venous system can tolerate pressures as high as arterial pressures,
permanent coronary sinus pressure elevation leads to arterial under perfusion,
edema, subsequent ischemia, myocardial hemorrhage and reflex hemodynamic
depression (Bezold-Jarisch effect). Ekstein has demonstrated that pressures above
40–60 mmHg are harmful [18, 19]. Similarly Gott and Lolley have presented
evidence that in retroperfusion, the distal perfusion pressure should be in the range
of 40 mmHg. Higher values cause venous damage and intramyocardial hemor-
rhage. A relationship between peak pressure and damage in the coronary venous
circulation has been observed both experimentally and clinically [20, 21]. With
this experimental simulation, it is demonstrated that a safe pressure limit could be
achieved with the proposed model.

Acknowledgments First and foremost, I would like to express my deepest gratitude, appreci-
ation and support to all the research team members. This research has been filed for patent by
Innovation & Commercialisation Center of Universiti Putra Malaysia on 16 November 2009. The
application number is PI20097028.

References

1. Grondin, C.M., Campeau, L., Lesperance, J., Solymoss, B.C., Vouhe, P., Castonguay, Y.R.,
Meere, C., Bourassa, M.G.: Atherosclerotic changes in coronary vein grafts six years after
operation. Angiographic aspects in 110 patients. J. Thorac. Cardiovasc. Surg. 77, 24 (1979)
150 S. A. Hussain et al.

2. Hamby, R.I., Aintablian, A., Handler, M., Voleti, C., Weisz, D., Garvey, J.W., Wisoff, G.:
Aortocoronary saphenous vein bypass grafts. Long term patency, morphology and blood flow
in patients with patent grafts early after surgery. Circulation 60, 901 (1979)
3. Grondin, C.M.: Results of invasive studies to assess the first decade of coronary artery bypass
operations. Presented at the American College of Surgeons, San Francisco (1981)
4. Pigott, J.D., Kouchoukos, N.T., Oberman, A., Cutler, G.R.: Late results of surgical and
medical therapy for patients with coronary artery disease and depressed left ventricular
function. J. Am. Coll. Cardiol. 5, 1036 (1985)
5. Ratajczyk-Pakalska, E.: Thebesian veins in the human heart with atherosclerotic lesions in
the coronary arteries. In: Mohl, W., et al. (eds.) The Coronary Sinus, pp. 141–145. Steinkopff
Verlag, Darmstadt (1984)
6. Baroldi, G., Scomazzoni, G.: Coronary Circulation in the Normal and the Pathologic Heart.
Office of the Surgeon General, Washington (1967)
7. Pratt, F.H.: The nutrition of the heart through the vessels of Thebesius and the coronary veins.
Am. J. Physio. 1, 86 (1898)
8. Beck, C.S.: Revascularization of the heart. Ann. Surg. 128, 854 (1948)
9. Beck, C.S., Stanton, E., Batiuchok, W., Letter, E.: Revascularization of the heart by a graft of
systemic artery into coronary sinus. JAMA 137, 436 (1948)
10. Mohl, W.: The development and rationale of pressure-controlled intermittent coronary sinus
occlusion: A new approach to protect is chemic myocardium. Wien. Klin. Wochenschr. 96,
20–25 (1984)
11. Mohl, W., Glogar, D., Mayr, H., et al.: Reduction of infarct size by pressure controlled
intermitted coronary sinus occlusion. Am. J. Cardiol. 53, 923–928 (1984)
12. Papp, L., Kekesi, V., Osvath, B.: The effect of pressure controlled intermitted coronary sinus
occlusion during reperfusion. In: Mohl, W., et al. (eds.) Clinics of CSI, pp. 339–344.
Steinkopff Verlag, Darmstadt (1984)
13. Gundry, S.R., Kirsh, M.M.: A comparison of retrograde cardioplegia versus antegrade
cardioplegia in the presence of coronary artery obstruction. Ann. Thorac. Surg. 38, 124–127
(1984)
14. Solorzano, J., Taitelbaum, G., Chiu, R.C.J.: Retrograde coronary sinus perfusion for
myocardial protection during cardiopulmonary bypass. Ann. Thorac. Surg. 25, 201–211
(1978)
15. Giancoli, D.C.: Physics. Prentice Hall, Englewood Cliffs (1998)
16. Cengel, Y.A., Cimbala, J.M.: Fluid Mechanics: Fundamentals and Applications. McGraw-
Hill Companies Inc., New York (2007)
17. Petkova, S., Hossain, A., Naser, J., Palombo, E.: CFD modeling of blood in portal vein
hypertension with and without thrombosis. 3rd International Conference on CFD in the
Minerals and Process Industries. Australia (2003)
18. Aviado, D.M.: The Bezold-Jarisch reflex: A historical perspective of cardiopulmonary
reflexes. Ann New York Acad Sci 2006 940, 48–58(1953)
19. Eckstein, R.W., Hornberger, J.C., Sano, T.: Acute effects of elevation of coronary sinus
pressure. Circulation 7(3):422–436
20. Gott, V.L., Gonzalez, J.L., Zuhdi, M., et al.: Retrograde per-fusion of coronary sinus for
direct vision aortic surgery. Surg. Gynecol. Obstet. 104, 319–328 (1957)
21. Lolley, D.M., Hewitt, R.L.: Myocardial distribution of asanguineous solutions retroperfused
under low pressure through the coronary sinus. J. Cardiovasc. Surg. 1980(21), 287–294 (1980)
22. Noel de Nevers.: Fluid Mechanics for Chemical Engineers. McGraw-Hill, Singapore (2005)
Biomechanical Modeling of Aneurysm
Growth and Rupture Using Fluid
Structure Interaction

M. Mazwan Mahat, A. Juliawati and Ishkrizat Taib

Abstract We developed a method to analyze aneurysm growth and rupture based

on idealized spherical shape from actual patient-specific geometry data. This study
was carried out to evaluate whether wall mechanics of soft tissue coupled with
blood flow dynamics can be used to provide the insight into the weakening phe-
nomena. In order to simulate the behavior of the system, the fluid structure
interaction method (FSI) was utilized using transferred data from the fluid
dynamics model to finite element wall mechanics. The FSI transferred these
dynamics loads to exert the aneurysms wall whose respective deformations were
then determined. The numerical modeling of aneurysms results the blood flow
parameter of pressure and velocity inside the aneurysm sac in the form of profile
correlations. These parameters generate a possible aneurysm rupture time during
the growth as a reasonable quantitative observation. The developed method allows
us to identify biomechanical factors that can influence the blood flow property
changes and wall stress distributions. As part of the computed maximum wall
stress to relate with growth and rupture, normalized velocity and pressure profiles
inside the aneurysm sac were correlated. This explains the effect of blood flow to
the weakening vessel wall and rupture behaviour due to variable flow conditions.
These results assist medical practitioners to the prediction of time and location of
ruptured aneurysm.

M. Mazwan Mahat (&)

University of Technology MARA, 40450 Shah Alam, Malaysia
e-mail: [email protected]; [email protected]
A. Juliawati
Universiti Malaysia Pahang, Lebuhraya Tun Razak, 26300 Gambang, Malaysia
e-mail: [email protected]
I. Taib
Universiti Tun Hussien Onn Malaysia, Parit Raja, 86400 Batu Pahat, Malaysia
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 151
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_12,
Ó Springer-Verlag Berlin Heidelberg 2012
152 M. Mazwan Mahat et al.

Keywords Abdominal aortic aneurysm  Numerical method  Fluid structure

interaction

1 Introduction

Abdominal aortic aneurysm (AAA) is a bulging and widening of the blood vessel
due the weakening of the aortic wall which may cause rupture of the aneurysm.
Normally, in clinical practice, surgical treatment of AAA is considered after the
maximal diameter exceeds 5–6 cm [1]. Nevertheless, the increasing of the aneu-
rysm region may eventually increase the risk of rupture [2] although, rupture could
occur in a small aneurysm [1]. To study deeper the detection of aneurysm rupture,
both non-geometric patient characteristic and geometric properties should be
considered [3]. Recently, computational modeling of the mechanical behavior of
AAA is the most prominent trend in AAA rupture risk assessment through the
improvement of imaging and segmentation [4].
A study of idealized geometry models show that the wall stress is significantly
reduced in the presence of intraluminal thrombus (ILT) but the size of the aneu-
rysm and the constitutive properties of the aortic wall were considered [5].
In addition, the effect of aneurysm wall was increasing due the presence of ath-
erosclerotic plaques inside the blood vessel [5]. Furthermore, the stress distribution
depends on the real AAA shape as well as the maximum diameter [4]. There is a
highly complex structure of the AAA in vivo shape which is far from the axi-
symmetric assumption [6]. Thus, for a specific wall stress assessment and finite
element models, the main approach needs to be based on the detailed AAA
geometry in form of a three-dimensional description. The complex wall stress
distribution was shown from models based on patient-specific geometries [7],
which may be influenced after the presence of ILT [8]. Furthermore, there was a
relation between the occurrence and location [9] of aneurysm rupture and the peak
wall stress value that was computed from this aneurysm model. From the obser-
vation, the peak wall stress was a better predictor of aneurysm rupture than the
diameter in AAA [7]. The boundary condition for the pressure load was applied at
the aortic wall which resulted from the coupling of blood velocity and pressure
Besides that, the method of generating a patient-mesh based on specific
hexahedral finite elements on AAA lumen and wall is presented to facilitate the
incorporating fluid/structure interaction for assessment of AAA wall stress. This
incorporation was illustrated by simulating the AAA characteristics.

2 Numerical Modeling

In this study, fluid structure interaction analysis (FSI) is used to model the effect of
blood flow on the weakening aorta wall. Sequential usages of commercially
available numerical modelling software were used to determine the flow structure
Biomechanical Modeling of Aneurysm Growth and Rupture 153

Fig. 1 Abdominal aortic

aneurysm (a) real aneurysm
(b) aneurysm side
(c) simplified aneurysm
model

(a) (b) (c)

and the behaviour of the aorta wall during aneurysm. Data were exchanged
between the models and analysis was done to determine the flow behaviour.
The aneurysm was simplified in a three-dimensional model which was con-
structed using Engineering Fluid Dynamic (EFD) software as shown in Fig. 1. The
thickness of the blood vessel was set to 1.5 mm throughout, and was gradually
reduced after the expansion of the aneurysm region. The idealized model of a
blood vessel with cylindrical symmetry has been commonly used in haemody-
namic studies because the deviations within a segment of nondiseased blood vessel
are too small to significantly alter the flow field. The geometry of the diseased
region is normally not symmetrical, but in the present study cylindrical symmetry
is assumed.
The viscosity of blood was assumed to be constant because changes in viscosity
are relatively small for vessel diameters larger than 0.5 mm [6]. This study also
focused on the basic flow dynamics, so Newtonian behaviour is sufficient for
approximation purpose [10]. For simplicity, the flow is considered to be incom-
pressible, homogeneous and non-pulsatile. The gravitational force was ignored and
the cavity was assumed to be filled with stagnant blood, i.e. no endoleaks were
assumed in stented aneurysm models.
In this study, the Navier–Stokes equations, which are formulations of mass,
momentum and energy conservation laws for fluid flows were solved. The equa-
tion was supplemented by fluid state equations defining the nature of the fluid, and
by empirical dependencies of the fluid viscosity and/or thermal conductivity on
temperature. A particular problem was finally specified by the definition of its
geometry, boundary and initial conditions. The mass, momentum and energy
conservation laws in a Cartesian coordinate system rotating at the angular velocity
X about an axis passing through the coordinate system’s origin can be written in
the conservation form as follows:
oq o
þ ðqui Þ ¼ 0 ð1Þ
ot oxi
oqui o   op o  
þ qui uj þ ¼ sij þ sRij þ Si ð2Þ
ot oxj oxi oxj
154 M. Mazwan Mahat et al.

where u is the fluid velocity, q is the fluid density, Si is a mass-distributed external

force per unit mass, p is the pressure, sij is the viscous shear stress tensor and the
subscripts are an expression to denote summation over the three coordinate
directions. For Newtonian fluids the viscous shear stress tensor is defined as
 
oui ouj 2 ouk
sij ¼ l þ  dij ð3Þ
oxj oxi 3 oxk

where

l ¼ l l þ lt ð4Þ
Here dij is the Kronecker delta function (equal to unity when i = j, and zero
otherwise), ll is the dynamic viscosity coefficient, lt is the turbulent eddy viscosity
coefficient and k is the turbulent kinetic energy. Note that lt and k are zero for
laminar flows. In the frame of the k-e turbulence model, lt is defined using two
basic turbulence properties, namely, the turbulent kinetic energy k and the tur-
bulent dissipation e,

Cl qk2
lt ¼ fl ð5Þ
e
Here, fl is a turbulent viscosity factor. It is defined by the expression
 
 2 20:5
fl ¼ 1  expð0:025Ry  1 þ ð6Þ
RT

where
pffiffiffiffiffi
qk2 q ky
RT ¼ ; Ry ¼ ð7Þ
le l
and y is the distance from the wall. This function allows us to take into account a
laminar-turbulent transition.
Two additional transport equations were used to describe the turbulent kinetic
energy and dissipation,
  
oqk o o lt ok oui
þ ðqui kÞ ¼ lþ þ sRij o  qe þ lt PB ð8Þ
ot oxi oxi rk oxi oxj
    
oqe o o lt oe e R oui qe2
þ ðqui eÞ ¼ lþ þ Ce1 f1 sij þ lt CB PB  Ce2 f2
ot oxi oxi re oxi k oxj k
ð9Þ
Here PB represents the turbulent generation due to buoyancy forces and can be
written as
Biomechanical Modeling of Aneurysm Growth and Rupture 155

gi 1 oq
PB ¼  ð10Þ
rB q oxi
where gi is the component of the gravitational acceleration in direction xi, the
constant rB = 0.9, and constant is defined as: CB = 1 when PB [ 0, and 0
otherwise;
 
0:05 3
f1 ¼ 1 þ ; f2 ¼ 1  expðR2T Þ ð11Þ
fl
The constants Cl, Ce1, Ce2, rk, re are defined empirically. In the EFD software,
the following typical values are used:

Cl ¼ 0:09; Ce1 ¼ 1:44; Ce2 ¼ 1:92; re ¼ 1:3

rk ¼ 1

These equations describe both laminar and turbulent flows. Moreover, transi-
tions from one case to another and back are possible. The parameters k and lt are
zero for purely laminar flows.

3 Results and Discussions

3.1 Velocity Behavior

Figure 2 shows the profile for the x-velocity along the centerline perpendicular to
the inlet flow in the aneurysm. The results agree with that of Khanafer et al. [11].
It is observed that a very weak recirculation or vortex is present at the distal area.
This is the sign that implies energy losses inside the aneurysm. The energy losses
may be transferred to the energy of pressure and stress to load on the pathological
aneurysm surface which the surface frequently shrunk and drawn [12]. It was
expected that the speed decreases as the blood travels from proximal to distal end
due the increase of aneurysm diameter, in line with the basic flow theory [11]. The
negative values in the profile depict the reverse flow of blood travel in the distal
end. The velocity behaviors for each stage of aneurysm are illustrated in Fig. 3 and
Fig. 4. Different diameters of the aneurysm were used in order to suit another
related study. From Fig. 3, where x-velocity profiles were compiled for three
different stages near the proximal area, it can be seen that no vortex can be
detected for all three stages and that they possess the same pattern of flow with
reduced values as the aneurysm grows. This phenomenon is in line with the
continuity equation which suggests that, when the dilation area expands, the
velocity reduces since the flow rate is constant. The unstable flow occurred in
aneurysm produces a vortex deformation. The curvature gradient in the x-velocity
component at distal neck depicted the strong formation of a vortex. The vortex
156 M. Mazwan Mahat et al.

Fig. 2 The x-velocity profile

for the centerline
perpendicular to the inlet flow

Fig. 3 The x-velocity profile

along the centerline,
perpendicular to the inlet flow

occurred in the aneurysm contributed to the energy losses of the fluid and is unable
to recover after the flow entering back to the normal artery. In summary, due to the
increment of diameter in the aneurysm, the blood velocity will experience some
deduction. This will apparently increase the pressure since the flow rate is
constant.

3.2 Pressure Distribution

The pressure gradient is affected by the deformation of the aneurysm wall in a

certain time period until ruptured if not surgically treated. The correlation between
the peak pressure and aneurysm deformation were investigated to prove that the
increment of the aneurysm wall occurred due the pressure raise. Figures 5 and 6
illustrates the development of the aneurysm from the early of the aneurysm
detection by a CT scan until it is assumed to be ruptured. There are 3 stages of
aneurysm deformation as shown in the curvature of Fig. 7 then comparison made
with Thompson et al. 2002 shows in Fig. 8. The pressure distribution pattern
Biomechanical Modeling of Aneurysm Growth and Rupture 157

Fig. 4 Velocity profile at the

distal in cross sectional of
aneurysm

Fig. 5 Pressure distribution

along the aneurysm wall

increases gradually for each stage due the expansion of the aneurysm wall. This
apparently influences the increase of pressure since the flow rate is constant. The
obvious exchangeable pressure gradient at the aneurysm wall explains the defor-
mation of the vortex at early proximal and late distal as shown in Figs. 9, 10 and
11 respectively while the pressure in the middle section shown in Fig. 10. Defi-
nitely, the presence of a vortex accelerates the aneurysm rupture. The pressure
losses have an effect on the aneurysm rupture where the blood takes a long time to
pass through an aneurysm and energy losses are higher than in an unruptured
aneurysm. Then, the energy losses are transferred to the energy of the pressure and
stress to load on the pathological aneurysm surfaces [12]. The polynomial cur-
vature has interpret the correlates well between the peak pressure and each stage of
aneurysm deformation as shown in Fig. 5 which means that the pressure rise is
proportional to the increment of dilation area. The obvious curvature discrepancy
between 2nd and 3rd stage can be explained by the higher pressure which is
158 M. Mazwan Mahat et al.

Fig. 6 Pressure distribution

along the centerline of
aneurysm

Fig. 7 Correlation between

peak pressure and each stages
of aneurysm deformable

Fig. 8 The percentages of

aneurysm growth for each
stage of aneurysm
deformable

applied at the 3rd stage where the pressure increases, and the rupture risk become
high. The aneurysm growth can be explained in millimeter (see Fig. 6) where the
1st stage is considered as a reference. A previous study stated based on clinical
diagnose the expansion rate is as high as 0.5 cm per year and more. The
Biomechanical Modeling of Aneurysm Growth and Rupture 159

Fig. 9 Pressure distribution

at the proximal in cross
sectional of aneurysm

Fig. 10 Pressure distribution

at the middle in cross
sectional of aneurysm

Fig. 11 Pressure distribution

at the distal in cross sectional
of aneurysm

recirculation flow inside the aneurysm is related to a dilation of the aneurysm

growth which is proportional to the reduction of the wall thickness. The obvious
increment from the 2nd stage to 3rd stage (from 3.823 to 5.847 mm) predicted the
strong vortex which occurred inside the aneurysm. Then, the steeply increase of
160 M. Mazwan Mahat et al.

Fig. 12 Shear stress

distribution along the
aneurysm region

the aneurysm growth may predict the ruptured which can occur at the aneurysm
wall.

3.3 Fluid Shear Stress Distribution

Figure 12 illustrates the shear stress profile between blood and the aneurysm wall.
The oscillatory curvature from proximal to distal neck can be explained by the
unstable flow which occurred. The unstable flow is predicted as a vortex defor-
mation inside the aneurysm. In early proximal, the deformation of the vortex was
detected. The deformation of the vortex influences the energy losses inside the
aneurysm. At the late distal, the strong deformation of the vortex is expected to
occur inside the aneurysm because of the immediate increment of curvature which
is shown in Fig. 12. The vortex deformation increases the rate of ruptured aneu-
rysm. The recirculation flow and vortex deformation near the aneurysm wall could
produce the oscillatory curvature along the aneurysm region. The shear stress
pattern was validated by Khanafer et al. [11] curvature which shows a similar
pattern with the present study.

4 Conclusion

In this study, there are several limitations which influenced the results. The AAA
was simplified and this is the underlying assumption in the present study. The rigid
wall model was assumed in a hemodynamic pressure field. Previous experimental
and computational studies have shown which arterial wall model has a minor
Biomechanical Modeling of Aneurysm Growth and Rupture 161

quantitative effect on the computed shear wall stress whereas the flow features are
preserved.
There is necessarily space for improvement in the accuracy of the computa-
tional models in order to establish a more reliable patient-specific index of AAA
rupture. Hence, a realistic boundary condition extracted from the patient in vivo is
required to be transferred to the computational mode which requires that the model
involves the coupling of solid fluid dynamics. The result demonstrates that when
the pressure is applied on the wall, an under-prediction in the computed AAA
occurs. However, this study has a limitation for the clinical study but it is still used
for the future study especially in fluid flow and fluid structure interaction to
establish the degree of error associated with the simplified model in the decoupled
approach. Finally, this study presents the flow pattern in blood vessels through the
modelling of the actual expanded that was designed in three dimensions.

Acknowledgments The support of the University of Technology Malaysia, under the Com-
putatonal Fluid Mechanics and Computational Solid Mechanics Laboratory is gratefully
acknowledged.

References

1. Lederle, F.A., Wilson, S.E., Johnson, G.R., et al.: Immediate repair compared with
surveillance of small abdominal aortic aneurysms. N. Engl. J. Med. 346, 1437–1444 (2002)
2. Szilagyi, D.E., Elliott, J.P., Smith, R.F.: Clinical fate of the patient with asymptomatic
abdominal aortic aneurysm and unfit for surgical treatment. Arch. Surg. 104, 600–604 (1972)
3. Hatakeyama, T., Shigematsu, H., Muto, T.: Risk factors for rupture of abdominal aortic
aneurysms based on three-dimensional study. J. Vasc. Surg 33(3), 453–461 (2001)
4. Stringfellow, M.M., Lawrence, P.F., String fellow, R.G.: The influence of aorta-aneurysm
geometry upon stress in the aneurysm wall. J. Surg. Res. 42(4), 425–433 (1987)
5. Inzoli, F., Boschetti, F., Zappa, M., et al.: Biomechanical factors in abdominal aortic
aneurysm rupture. Eur. J. Vasc. Surg. 7(6), 667–674 (1993)
6. Aenis, M., Stancampiano, A.P., Wakhloo, A.K., et al.: Modeling of flow in a straight stented
and nonstented side wall aneurism model. ASME J. Biomech. Eng. 119, 206–212 (1997)
7. Fillinger, M.F., Raghavan, M.L., Marra, S.P., et al.: In vivo analysis of mechanical wall stress
and abdominal aortic aneurysm rupture risk. J. Vasc. Surg. 36(3), 589–597 (2002)
8. Wang, D.H.J., Makaroun, M.S., Webster, M.W., et al.: Effect of intraluminal thrombus on
wall stress in patient-specific models of abdominal aortic aneurysm. J. Vasc. Surg. 36(3),
598–604 (2002)
9. Venkatasubramaniam, A.K., Fagan, M.J., Mehta, T., Mylankal, K.J., et al.: A comparative
study of aortic wall stress using finite element analysis for ruptured and non-ruptured
abdominal aortic aneurysms. Eur. J. Vasc. Endovasc. Surg. 28(2), 168–176 (2004)
10. Perktold, K., Peter, R., Resch, M.: Pulsatile non-Newtonian blood flow simulation through a
bifurcation with an aneurysm. J. Biorheol. 26, 1011–1030 (1989)
11. Khanafer, K.M., Gadhoke, P., Berguer, R., et al.: Modeling pulsatile flow in aortic
aneurysms: effect of non-Newtonian properties of blood. J. Biorheol. 43, 661–679 (2006)
12. Qian, Y., Harada, T., Fukui, K.: Hemodynamic analysis of cerebral aneurysm and stenosed
carotid bifurcation using computational fluid dynamics technique. LSMS 2007, LNBI 4689,
pp. 292–299. (2007)
Study and Mathematical Modeling
of Transient Gas Compositions
for Modified Atmosphere Packaging

A. Heydari, I. Alemzadeh and M. Vossoughi

Abstract A generic method for keeping quality and quantity of fruits and
vegetables named Modified Atmosphere Packaging is introduced. In this work, the
interactions between respiration rates, permeability coefficients and headspace gas
compositions are studied and the system is modeled mathematically. The
dynamics of the system were solved using a fourth order Runge–Kutta method.
A computer simulation package developed for analyses of mathematical aspects.
Time to achieve equilibrium, equilibrium conditions and transient patterns com-
pared under different permeation rates. A new performance index used as Integral
of Time for Absolute Error (ITAE) based on dynamics of gas compositions to
evaluate transient patterns for each gas. In this research, the permeability of some
polymers from literature and theoretical packaging materials were studied and it
was found that the behavior of each gas is different with others and the pattern of
achieving the equilibrium depends on packaging permeability coefficients.

 
Keywords Modeling Modified atmosphere packaging Equilibrium Transient 

gas compositions Fruits and vegetables Polymer 
A. Heydari
Department of Chemical and Petroleum Engineering,
Sharif University of Technology, Azadi Ave, Tehran, Iran
e-mail: [email protected]
A. Heydari
Department of Chemical Engineering, University of Mohaghegh Ardabili,
Daneshgah Ave, Ardabil, Iran
I. Alemzadeh (&)  M. Vossoughi
Institute for biotechnology and Environment (IBE),
Sharif University of Technology, Azadi Ave, Tehran, Iran
e-mail: [email protected]; [email protected]
M. Vossoughi
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 163
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_13,
Ó Springer-Verlag Berlin Heidelberg 2012
164 A. Heydari et al.

1 Introduction

Quality optimization and loss reduction in the post harvest chain of fresh fruits and
vegetables are the main objectives of packaging technology. Modified Atmosphere
Packaging (MAP) relies on modification of the atmosphere in headspace of the
package, achieved by the natural interplay between the respiration of the product
and the transfer of gases through the packaging that leads to an atmosphere richer
in CO2 and poorer in O2. This atmosphere can potentially reduce the respiration
rate [1, 2]. Proper packaging is essential to maintaining quality of fruits via
physical protection; reducing water loss and delaying softening if optimal modified
atmospheres are kept. The MAP effects are based on observed slowing of plant
respiration in low O2 environments. As the concentration of O2 inside the package
falls below about 10%, respiration starts to slow. The goal of MAP of fresh
products is to create an equilibrium packaging atmosphere with O2 low enough
and CO2 high enough to be beneficial to the produce and not injurious [3–5].
A new packaging technology, used for prolonging the shelf-life of respiring
products, fresh products (fresh-cut produce), by following the oxygen consumption
and carbon dioxide production of the products over time, the respiration rate of
vegetables, expressed as mL O2 consumed per (kg.h) or mL CO2 produced per
(kg.h), could be determined at 7°C (selected storage conditions for minimally
processed vegetables), consists of Equilibrium Modified Atmosphere Packaging
(EMAP) combined with chilling. In this preservation technique, the air around the
commodity is altered to a gas combination of O2 and CO2 (balance N2), [6].
Modified atmosphere packaging achieves an equilibrium based on thickness and
type of polymer, gas exchange surface and volume of headspace. Passive modified
package may require some days to reach the equilibrium while active modified
package accelerate this process with some additional costs. In this research, a
Microsoft Visual Basic computer simulation package is developed. The design of a
proper packaging depends on a number of variables such as the respiration charac-
teristics of the product, its mass; recommend atmosphere composition, permeability
of the packaging materials to gases and its dependence on temperature. This program
predicts gas composition dynamic until to reach equilibrium. A mathematical model
used to estimate oxygen, carbon dioxide and nitrogen composition in headspace of
the packaging. The performance index Integral of Time of Absolute Error (ITAE)
based on dynamic gas composition diagram is introduced to obtain optimal values of
packaging conditions. The main goal of this software is finding optimal value for
thickness and type of polymer, gas exchange surface and volume of headspace.

1.1 Post-Harvest Plant Physiology

Fresh fruits and vegetables continue to respire after harvesting. Respiration involves
a complex set of biochemical processes of which part is the oxidative breakdown
of sugars and organic acids into CO2 and H2O plus heat and metabolic energy.
Study and Mathematical Modeling of Transient Gas Compositions 165

The respiration rate and metabolic pathway may change naturally as the products go
through the internal processes of ripening, maturity and senescence or be changed
artificially by altering the external environment of storage, i.e. temperature and
surrounding gas composition. The inhibition was interpreted to be in a manner
consistent with an enzyme kinetic modeling where a substrate (O2) must bind to a
receptor before a dissociable activator (C2H4) can attach.
The objective of controlled or modified atmosphere storage is to lower the rate
of respiration through atmosphere management (in addition to temperature con-
trol) to reduce the rate of substrate depletion, O2 consumption, CO and the
associated heat release. Senescence begins as the stored starch and sugar is con-
sumed; the rate of substrate consumption is simply determined by the rate of
respiration. Temperature control and modification of atmosphere are two impor-
tant factors in prolonging shelf life. MA packages should be carefully designed, as
a system incorrectly designed may be ineffective or even shorten the shelf life of
the product. The design should take into consideration not only steady-state
conditions, but also the dynamic process. If the product is exposed for a long time
to unsuitable gas composition before reaching the adequate atmosphere, the
package may have no benefit. The design of an MA package depends on a number
of variables: the characteristics of the product, its mass, the recommended atmo-
sphere composition, the permeability of the packaging materials to gases and its
dependence on temperature and the respiration rate of the product as affected by
different gas composition and temperature. Thus, respiration rate modeling is
central to the design of MAP for fresh fruits and vegetables.

1.2 Respiration Rate

All the foods and vegetables continue to respire after harvesting with different
rates. The respiration rate and metabolic pathway may change by any change in
temperature and surrounding gas composition. The objective of controlled or
modified atmosphere storage is to lower the rate of respiration through atmosphere
management to reduce the rate of substrate depletion, O2 consumption, CO2
production, and the associated heat release [7]. By decrease in respiration rate, the
physical and chemical properties of fruits and vegetables will change by lower
rates. Both of O2 and CO2 are important to control and modify the metabolism of
live plants. The quality of fruits and vegetables can be discussed based on some
parameters such as color, odor, taste, softness or hardness and external form before
and after packaging. The risk of a loss in quality or value is the most important
factor to design of a proper package. Beaudry [8] discussed effects of O2 and CO2
partial pressures on selected phenomena affecting quality of fruits and vegetables.
Fresh fruits and vegetables are still alive and require oxygen for their metab-
olism. Different fruits and vegetables, and even different varieties of a given fruit
or vegetable will vary in their respiration rates. The chemical equation of respi-
ration is as below:
166 A. Heydari et al.

C6 H12 O6 þ 6O2 ! 6CO2 þ 6H2 O þ Energy ð1Þ

The respiration rate of fresh produce can be expressed as O2 consumption rate
and/or CO2 production rates as give in Eqs. 2, 3:
1 dVO2
RO2 ¼  ð2Þ
M dt
1 dVCO2
RCO2 ¼ ð3Þ
M dt
where V is the partial volume in headspace of package, R is the respiration rate
expressed in volume of gas generated/consumed per unit of time and weight of the
product (M) and t is time. Respiration rate models presented in the literature are
linear, polynomial, exponential and Michaelis–Menten type [4, 9, 10].
The proper balance of several variables that affect the package atmosphere is very
important. When fresh products are sealed inside a polymeric film package, respi-
ration will lower the in-package O2 level and increase the CO2 level. A major
challenge in designing MA packages is to match the rate of O2 uptake and CO2
production of the products, with the O2 and CO2 transfer rates through the package. In
a well-designed package, gas levels inside the package will equilibrate within a
range, which benefits the products. For most products, the therapeutic range for CO2
and O2 is between 2% and 10%. The required combinations of temperature, oxygen
and carbon dioxide levels vary with vegetable type, variety, origin and season [11].

1.3 Gases Exchange Dynamic

Carbon dioxide is important because of its biostatic activity against many spoilage
organisms that grow at refrigeration temperatures and oxygen inhibits the growth
of anaerobic pathogens and nitrogen is used as a filler gas to prevent pack collapse.
Each gas has its special properties such as molecular weight, behavior in matrix of
polymer and exchange properties through different polymers which must be
considered in study and modeling of packaging. In this research, the transmission
rates of gases were studied based on solubility and diffusivity. One side of the
polymeric package is in contact with the atmosphere and the other side is in
contact with a modified atmosphere. Gas composition of headspace is constant but
inside the packaging, both respiration and gas permeability will change the gas
composition until reach to equilibrium. In this system, gases will transfer through
the packaging in both sides based on partial pressures of gases inside the pack-
aging. The quantity of gas exchange rate can expressed by following equation:
A
Q ¼ Pe ðpin  pout Þ ð4Þ
L
where L is the polymer thickness, Pe is the permeability coefficient of gases in a
specified polymer which is different for each gas, A is the mass transfer area and
Study and Mathematical Modeling of Transient Gas Compositions 167

pin and pout are gas partial pressure inside and outside of packaging respectively.
Both of gas exchange rates and respiration rates are time depended while the gas
composition inside the packaging is not constant.

1.4 Modified Atmosphere Packaging

Free volume or headspace of packaging will change because of respiration and gas
exchange. Each increase or decrease in respiration rate is affecting the gas exchange
rate and vice versa. The respiration rate is depended on composition of carbon
dioxide and oxygen. On the other hand, gas exchange rate is dependent on the
solubility of gases in the headspace while gases solubility in polymeric films is gas
composition dependent. The gas composition considered in both of respiration rate
and gas transmission rate [12]. The molar balance of gases in headspace can be
expressed as below where n is the number of molecules of each gas in the headspace:
dnO2 A
¼ PeO2 ðpO2 ;out  pO2 ;in Þ  RO2 M ð5Þ
dt L
dnCO2 A
¼ PeCO2 ðpCO2 ;out  pCO2 ;in Þ þ RCO2 M ð6Þ
dt L
dnN2 A
¼ PeN2 ðpN2 ;in  pN2 ;out Þ ð7Þ
dt L
The total pressure (ptotal) is assumed constant and the partial volume of each gas
in the headspace at each time can be calculated as below:
ni
pi ¼ ptotal ð8Þ
nO2 þ nCO2 þ nN2
To solve Eqs. 5–7 simultaneously, a fourth-order Runge–Kutta method is used.
This method is summarized as below:
1
yiþ1 ¼ yi þ ðk1 þ 2k2 þ 2k3 þ k4 Þh ð9Þ
6
Where:

k1 ¼ f ðti ; yi Þ
 
1 1
k2 ¼ f ti þ h; yi þ k1 h
2 2
 
1 1
k3 ¼ f ti þ h; yi þ k2 h
2 2
k4 ¼ f ðti þ h; yi þ k3 hÞ
168 A. Heydari et al.

1.5 Equilibrium Conditions

Packaging parameters can be divided in two groups; the first are dynamic
parameters which are time depended such as gas compositions, respiration rates
and gas exchange rates and second are static parameters which are constant such as
permeability coefficients, polymer thickness, mass transfer area and mass of fruit.
At equilibrium conditions, all dynamic parameters will be constant based on static
parameters. The equilibrium must be obtained at the accepted time especially in
case of modified atmosphere packaging because respiration rate is optimum at
equilibrium. On the other hand, gas composition of oxygen and carbon dioxide at
equilibrium is very important to keep quality and prevent side effects such as
fermentation. In addition, the dynamics of gas compositions must be considered to
evaluate transient patterns for each gas. All fruits and vegetables have accepted
upper and lower levels for oxygen and carbon dioxide thus, average of these limits
is a good goal to achieve at equilibrium.

2 Software Description

In this research, a Microsoft Visual Basic computer package using Object Linking
and Embedding (OLE) technique was developed. The OLE technique easily
exchange data between source code of software and Access (as database) and
Excel (as spreadsheet and drawing tool). This comprehensive and graphic program
predicts the gas composition dynamic until to reach the equilibrium. The proposed
software has two upgradeable databases made by Access. The first database
contains properties of polymers such as permeability of gases. The second data-
base consists of respiration rates of some fruits and vegetables with their variety.
A user can define and add new data to each of these databases. For example a user
can add a new respiration rate for a native fruit or add properties of new polymers.
The term OLE is a protocol or standard, which Microsoft asks makers of Windows
applications to support. OLE provides easy ways to exchange data between
Windows applications to take advantage of other applications. The OLE embeds
data directly into another application, or stores information about a link to a
specific piece of data. An OLE client is an application that uses the documents
provided by an OLE server. An OLE server is an application that can provide
access to its documents by means of OLE.

2.1 Integral of Time of Absolute Error (ITAE)

The performance index Integral of Time of Absolute Error (ITAE) based on the
dynamic gas composition diagram is introduced to obtain optimal values of
packaging conditions. Integral of absolute value of error, integral of squared error
Study and Mathematical Modeling of Transient Gas Compositions 169

and integral of time of absolute error are common methods to report quantitative
measures of performance. In this software the ITAE is used because inclusion of
‘‘time’’ tends to reduce contribution of large initial errors and also emphasizes
errors near time of equilibrium. ITAE has been found to be preferable in opti-
mization because the minimum value of the integral is readily discernible. Mini-
mization of a performance index is often an objective in most of problems such as
cost, fuel economy etc.
For have a better comparison, a published paper by Salvador [13] was used.
In his paper, the rate of respiration for Burlat cherries is reported as Michaelis–
Menten type.
Summary of input and output of the software for specified conditions is presented in
Fig. 1. In Fig. 2 gas composition versus time is shown. After about 625 h, this system
reaches its equilibrium. It can be seen that the behavior of each gas is different with
others and the pattern to achieve the equilibrium is strongly depended on packaging
conditions. In this case, compositions of N2 and O2 will increase and decrease,
respectively, while composition of CO2 reaches a maximum after about 150 h. In
Fig. 3 errors are presented for each gas based on their equilibrium conditions.
In this software ITAE is used to report the performance of packaging. To inhibit
the effects of negative numbers in numerical calculation of integral, it was nec-
essary to use the absolute error instead of error. Fig. 4 shows the absolute error
versus time.
ITAE for each gas is defined as:
t¼t
Z Eq
 
ITAEi ¼ time(hr)*yi ðtÞ  yi ðtEquilibrium Þ dt ð10Þ
t¼0

Where function yi(t) is dependence of change in composition of each gas base

on time. Quantity of time multiply by absolute error is revealed in Fig. 5. All of
gases achieve their equilibrium in different times; therefore, term of equilibrium in
Eq. 10 is the maximum equilibrium time between O2, CO2 and N2. The quantity of
the integral is calculated by Simpson numerical integration method. The total
performance index, which is used to find the optimum design is defined by Eq. 11.
ITAEtotal ¼ ITAEO2 þ ITAECO2 ð11Þ
In the above equation, ITAE for N2 is not considered because the N2 compo-
sition has no effect on the respiration rate.

3 Discussion and Conclusion

In this research, the permeability of some polymers from literature and theoretical
packaging materials are studied to find their ITAE quantities. The best polymer
can be selected based on the minimum quantity of ITAE. On the other hand it must
170 A. Heydari et al.

Fig. 1 Results of software in given conditions

be considered that gas composition of equilibrium will change for each polymer
and must be in proper levels. Based on conditions of one set of Salvador exper-
iments [13]; free volume = 1.5 L, surface area = 0.1 m2, polymer thickness = 2.1
mil, temperature = 5°C, permeability = 9.1, 1.9 and 1.9 mol.mil.m-2.h-1.atm-1 for
O2,CO2 and N2 respectively. We multiplied permeability coefficients to some
numbers to generate theoretical polymers which named as P1.5, P1, P1.5, P2 and
P3 while the mentioned numbers refer to the multiplied number to the original
polymer. The results are shown in Table 1.
Increasing in the permeability of polymers, the time to reach the equilibrium
and carbon dioxide level decreased but oxygen level increases. If the oxygen level
equal to 2% considered as the minimum required oxygen to inhibit fermentation,
P0.5 and P1 are not suitable. On the other hand, P3 has a high oxygen level which
means high respiration rate, so P3 is not proper although this polymer has the
minimum values for total ITAE and equilibrium time. Both of P1.5 and P2 are
proper to use and we must choose the best one. P1.5 has the better level for oxygen
but P2 has the lower equilibrium time. Based on total ITAE, P2 is the best polymer
for this system. For better comparison between these polymers, transient condi-
tions of oxygen and carbon dioxide are compared in Figs. 6, 7.
Based on the oxygen transient behavior, both P1.5 and P2 have very similar
pattern and approximately they reach to their equilibrium in the same manner but
in different times and gas compositions. Therefore, we cannot select the best
polymer.
Study and Mathematical Modeling of Transient Gas Compositions 171

100

O2
80
Gas composition (%) CO2
N2

60

40

20

0
0 100 200 300 400 500 600 700
Time (hr)

Fig. 2 Molar gas composition(%) versus time

20

O2
15
CO2
N2
10
E rror (% )

-5

-10

-15
0 100 200 300 400 500 600 700
T im e (hr)

Fig. 3 Error(yi(t) - yi(Eq)) versus time

The change in carbon dioxide level versus time for P1.5 and P2 are very
different. Both polymers make a maximum in carbon dioxide composition in
headspace. Difference of these maxima for P1.5 and P2 is the decision parameter.
It means that the greater maximum of P1.5 makes a greater ITAE for P1.5.
Therefore, P2 is better than P1.5 and is the best polymer for the considered
packaging conditions.
172 A. Heydari et al.

20

16 O2
Absolute error (%) CO2
N2
12

0
0 100 200 300 400 500 600 700

Time (hr)

Fig. 4 Absolute error versus time

18

16 O2
CO2
14 N2

12

10

0
0 100 200 300 400 500 600 700

T im e (h r)

Fig. 5 Times* absolute error versus time

Table 1 Effects of permeability coefficients on equilibrium

Polymer O2 [%] Eq CO2 [%] Eq Time [hr] Eq O2 ITAE CO2 ITAE ITAE total
P0.5 0.89 5.4 925 1.70 E3 11.40 E3 13.10 E3
P1 1.85 4.69 625 1.67 E3 3.93 E3 5.60 E3
P1.5 2.88 4.33 515 1.54 E3 1.95 E3 3.49 E3
P2 3.90 4.30 460 1.37 E3 1.12 E3 2.49 E3
P3 6.09 3.53 390 0.97 E3 0.46 E3 1.43 E3
Study and Mathematical Modeling of Transient Gas Compositions 173

Fig. 6 Comparison of change in oxygen level between P1.5 and P2 Versus time

Fig. 7 Comparison of change in carbon dioxide level between P1.5 and P2 Versus time

In this research, a Microsoft Visual Basic computer package using OLE tech-
nique was developed. This comprehensive and graphic program predicts gas
composition dynamics until to reach equilibrium. The performance index Integral
of Time of Absolute Error (ITAE) based on dynamic gas composition diagram is
introduced to obtain optimal values of packaging conditions. The main goal of this
software is finding optimal conditions for packaging based on the minimum of
ITAE. Using ITAE is a good alternative for equilibrium time in design of the best
passive modified atmosphere packaging.
174 A. Heydari et al.

References

1. Kader, A.A., Zagory, D., Kerbel, E.L.: Modified atmosphere packaging of fruits and
vegetables. Crit Rev Food Sci Nutr. 28, 1–30 (1989)
2. Saltveit, ME: A summary of CA and MA requirements and recommendations for harvested
vegetables. In: Saltveit, M.E. (ed.) Proceedings of the 7th international controlled atmosphere
research conference: Vol. 4, pp. 98–117. Davis, CA, USA (1997)
3. Zagory, D., Kader, A.A.: Modified atmosphere packaging of fresh produce. J Food Technol.
42, 70–77 (1988)
4. Day, B.: High oxygen modified atmosphere packaging for fresh prepared produce. Post
harvest News Inf. 7, 31–34 (1996)
5. Ballantyne, A., Stark, R., Selman, J.D.: Modified atmosphere packaging of shredded lettuce.
Int J Food Sci Technol 23, 267–274 (1988)
6. Jacxsens, L., Devlieghere, F., Debevere, J.: Validation of a systematic approach to design
equilibrium modified atmosphere packages for fresh-cut produce. Lebensm.-Wiss u-Technol.
32, 425–432 (1999)
7. Paul, D.R., Clarke, R.: Modeling of modified atmosphere packaging based on designs with a
membrane and perforations. J Membr Sci. 208, 269–283 (2002)
8. Beaudry, R.M.: Effect of O2 and CO2 partial pressure on selected phenomena affecting fruit
and vegetable quality. Postharvest Biol Technol 15. 293–303 (1999)
9. Beaudry, R.M., Cameron, A.C., Shirazi, A., DostalLange, D.L.: Modified atmosphere
packaging of blueberry fruit: effect of temperature on package O2 and CO2. Journal of the
American Society for Horticultural Science 117, 436–441 (1992)
10. Beaudry, R.M.: Effect of carbon dioxide partial pressure on blueberry fruit respiration and
respiratory quotient. Postharvest Biol Technol 3. 249–258 (1993)
11. Smyth, A.B., Song, J., Cameron, A.C.: Modified atmosphere packaged cut iceberg lettuce:
effect of temperature and O2 partial pressure on respiration and quality. J Agric Food Chem.
46, 4556–4562 (1998)
12. Chaudhary, B.I., Johns, A.I.: Solubilities of Nitrogen, Isobutane and Carbon Dioxide in
polyethylene. J Cell Plast 34. 312–328 (1998)
13. Salvador, ML, Jaime PJ, Oria, R.: Modeling of O2 and CO2 Exchange dynamics in modified
atmosphere packaging of burlat cherries. J Food Sci. 67, 231–235 (2002)
Kinetic Modeling of Biogas Generation
from Banana Stem Waste

N. Zainol, J. Salihon and R. Abdul-Rahman

Abstract A kinetic model for biogas generation from banana stem waste was
proposed on the basis of the obtained experimental results. The system consists of
an anaerobic sequencing batch reactor for the first stage and an anaerobic fixed bed
reactor for the second stage, which is operating at hydraulic retention times (HRT)
of nine days. The process was conducted at ambient temperature for the first stage
and thermophilic temperature for the second stage. Four differential equations
described the overall process. This study employed first order kinetics for
hydrolysis of non-soluble organic matter and a Michaelis–Menten equation type
for the soluble organic matter decomposition, total volatile acids consumption and
methane production. The following kinetics constants were obtained for the above-
mentioned anaerobic stages: (a) hydrolysis and solubilization of organic matter: k1
(kinetic constant for non-soluble organic matter degradation): 0.0037 day-1; k2
(maximum rate of soluble organic matter production): 0.0241 g soluble chemical
oxygen demand (SCOD)/l day; k3 (saturation constant): 0.0236 g SCOD/l;
(b) acidogenesis: k4 (maximum rate of soluble organic matter degradation): 0.0086
g SCOD/l day; k5 (saturation constant): 0.0189 g SCOD/l; and (c) methanogenesis:
k6 (maximum rate of acetic acid (TVA) consumption): 0.0092 g TVA/l day; and k7
(saturation constant): 0.0003 g TVA/l. The kinetic constants obtained and the
proposed equations were used to simulate the different steps of the anaerobic

N. Zainol (&)  J. Salihon

FKKSA, Universiti Malaysia Pahang, Lebuhraya Tun Razak,
26300 Kuantan, Pahang, Malaysia
e-mail: [email protected]; [email protected]
J. Salihon
e-mail: [email protected]
R. Abdul-Rahman
JKKP, Universiti Kebangsaan Malaysia, 43600 UKM Bangi,
Selangor, Malaysia
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 175
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_14,
Ó Springer-Verlag Berlin Heidelberg 2012
176 N. Zainol et al.

digestion process of banana stem waste and to obtain the theoretical values of non-
soluble and soluble CODs, TVA and methane production.

1 Introduction

The anaerobic digestion process is a good and attractive alternative for the
economical reduction of the organic matter concentration in vegetable solid wastes
[1–3]. Anaerobic digestion consists of a multitude of biochemical reactions in series
and in parallel that occur simultaneously [10, 14]. Due to the complexity of the
biochemical reactions, and the presence of inhibitor or delayer compounds in
TPOP, a more detailed study of the kinetics is necessary for understanding
and optimising the process. Mosey [12] and Kalyuzhnyi and Davlyatshina [8]
developed mathematical models describing the kinetics of acidogenesis, ethanol-
degrading acetogenesis, butyrate degrading acetogenesis, acetoclastic methano-
genesis, hydrogenothrophic methanogenesis, bacterial decay, pH and various
inhibitions of the mentioned steps. Mosche and Jordening [11] studied the acetate
and propionate degradation and inhibition. The inhibition caused by propionic acid
was only slightly at concentrations lower than 800 mg/l and pH higher than 6.5,
while propionate degrading bacteria were sensitive at pH lower than 6.6. Garcia
et al. [5] used first and second order models to describe the anaerobic digestion of
livestock manure. The results obtained showed that the second model had both
statistical and physical meanings in the parameter values obtained. The model took
into account a simplified reaction scheme formed by six reactions. Several sim-
plifications were made (lumping, pseudo-steady state for one compound, etc.)
yielding four key compounds to be analysed and fitted to the model as production-
rates expressions (total biomass, chemical oxygen demand (COD), volatile fatty
acid (VFA), and methane). Three main stages were considered in the process:
enzymatic hydrolysis, growth of acetogenic microorganisms and growth of meth-
anogenic microorganisms. Munch et al. [13] developed a mathematical model for
volatile acid production. The model was verified by comparing its prediction with
experimental data presented in the literature. Predicted and measured steady-state
effluent VFA, soluble COD and ammonia-N concentrations were compared for
varying hydraulic and solids residence times. Valentini et al. [15] compared
Michaelis–Menten, substrate first order, substrate and biomass first order and
substrate first order and biomass half order equations in the anaerobic degradation
of cellulose particles of known size. A general kinetic equation, which may include
all the four considered kinetics, was presented for a more accurate mathematical
representation of the hydrolysis process. Beccari et al. [1] studied the anaerobic
degradation of olive mill effluents by successive stages of acidogenesis and
methanogenesis and found that most of the lipids were degraded both in acido-
genesis and methanogenesis stages. On the other hand, polyphenol-like substances
were not degraded in acidogenic conditions, whereas they were partially removed
Kinetic Modeling of Biogas Generation 177

in methanogenic conditions. The aim of this work was to carry out a kinetic
evaluation of the different steps of anaerobic digestion (hydrolysis, acidogenic and
methanogenic) process of banana stem waste (BSW) using two stages system.

2 Materials and Method

2.1 Experimental Set-Up

Banana stem waste (BSW) was used as substrate in the experiments. These were
cut into small pieces (about 5 cm) and mixed with distilled water. All experiments
were done in a 20 L anaerobic sequencing batch reactor followed by a 10 L fixed
bed reactor with gas outlet. All the reactors were seeded with anaerobic accli-
matized banana stem sludge. The anaerobic digestion system was varied at reac-
tion temperatures using water bath. The hydraulic retention time (HRT) and
organic loading rate (OLR) for this system were 9 d and 4 gTS/l.d, respectively.
The process was conducted at ambient temperature for the first stage and ther-
mophilic temperature for the second stage. Daily withdrawal of an appropriate
volume from the reactor corresponding to the determined HRT or OLR was done
by a draw-and-fill method. Biogas evolved from the reactor was measured and
collected in a gas holder by water displacement. Samples were collected and
analyzed for performance evaluation.

2.2 Analytical Methods

COD concentration was spectrophotometrically analyzed using a spectrophotometer

and methods as in spectrophotometric instrument manual. The gas collection was
done using water displacement daily. Methane content was analyzed using gas
chromatography with thermal conductivity detector (GCTCD) and helium as the
carrier gas. Acetic acid concentration (TVA) was determined using high perfor-
mance liquid chromatography (HPLC). The substrate concentration was measured
as suspended solid according to standard methods for the examination of water and
wastewater. 20 ml well-mixed sample was filtered through a weighed standard
glass-fiber filter and the residue retained on the filter was dried to a constant weight at
103–105°C. The increase in weight of the filter represents the total suspended solids
[6]. The following parameters were determined: total chemical oxygen demand
(TCOD), soluble chemical oxygen demand (SCOD), total solids (TS), total volatile
solids (TVS), total suspended solids (TSS) and VSS. Non-soluble chemical oxygen
demand (NSCOD) was calculated by the difference between TCOD and SCOD, total
dissolved solids (TDS) were calculated by the difference between TS and TSS, and
volatile dissolved solids (VDS) were obtained by the difference between TVS
and VSS.
178 N. Zainol et al.

Fig. 1 Comparison between

the experimental hydrolysis
reactor effluent concentration
(NSCOD) and the theoretical
values

3 Results and Discussion

3.1 Kinetic Evaluation

Based on the literature review mentioned in Sect. 1, and taking into account the
characteristics of banana stem waste (BSW) and the experimental results obtained,
the following hypotheses were assumed:
1. The insoluble organic matter or volatile suspended solids were first transformed
to soluble organic matter (DVS) following a first-order kinetics.
2. The dissolved organic matter (VDS) resulting from the decomposition of the
insoluble organic matter and initially present in raw BSW, was transformed to
volatile acids following a Michaelis–Menten kinetic model.
3. The volatile acids resulting from the decomposition of VDS and initially
present in the raw BSW were transformed to methane and carbon dioxide
following a Michaelis–Menten kinetic model.
4. The digester was assumed to be a completely mixed reactor.
5. All concentrations of organic matter, including methane production, were
expressed as COD. With these considerations, the following kinetic model can
be established, using the following differential equations:

dSVSS
 = k1 SVSS ð1Þ
dt
 
dSVDS k2 SVDS
= k1 SVSS  ð2Þ
dt k3 + SVDS
   
dSTVA k4 SVDS k6 STVA
¼  ð3Þ
dt k5 + SVDS k7 + STVA
Kinetic Modeling of Biogas Generation 179

Fig. 2 Comparison between

the experimental
methanogenesis reactor
effluent concentration
(SCOD) and the theoretical
values

 
dSM k6 STVA
¼ ð4Þ
dt k7 + STVA
In Eq. 1:
dSVSS/dt is the removal rate of non-soluble organic matter (g NSCOD/l day)
k1: kinetic constant of the reaction (1/day)
SVSS: effluent concentration of biodegradable non-soluble organic matter equiva-
lent to biodegradable NSCOD concentration (g/l).
In Eq. 2
(dSVDS/dt) is the production rate of soluble organic matter (g SCOD/l day)
k2: maximum removal rate of soluble organic matter (g SCOD/l day)
SVDS: effluent concentration of soluble biodegradable organic matter (g SCOD/l)
and k3 is the saturation constant (g SCOD/l).
In Eq. 3:
(dSTVA/dt) is the rate of TVA production (g/l day)
k4: maximum rate of VDS uptake (g SCOD/l day)
k5: saturation constant (g SCOD/l)
k6: maximum rate of TVA consumption (g/l day)
STVA : effluent TVA concentration (g/l)
k7: saturation constant (g TVA/l).

Finally, in Eq. 4
dSM/dt is the rate of methane formation (g methane/l day).
180 N. Zainol et al.

Fig. 3 Comparison between

the experimental acetic acid
effluent (TVA) from
methanogenesis reactor and
the theoretical values

Fig. 4 Comparison between

the experimental methane
production and the theoretical
values

The model formulation demands a differential equation for biomass concentration

(X), which takes into account its accumulation, growth and losses with the effluent
due to the liquid flow. However, as is well known, part of the biomass is attached
onto the surface of the reactor walls, which is difficult to quantify. In this model, it
is assumed that the possible accumulation of biomass is muffled by its adhesion or
effluent loss. Therefore, the model assumes that biomass concentration within the
reactor remains practically constant. The values of the constants k1–k7 were
determined from the experimental results by mathematical adjustment using the
Matlab software (version 6.1) with the condition of obtaining a minimum value of
the sum of the squares of the differences between the experimental and theoretical
values.
Kinetic Modeling of Biogas Generation 181

Table 1 Comparison between predicted concentration and true concentration of NSCOD

No Predicted True concentration Squared error
concentration
1 4.5 4.5 0.0000
2 4.719 4.638 0.0066
3 4.8518 4.718 0.0179
4 5.0347 4.733 0.0910
5 5.2245 4.783 0.1949
6 5.4214 4.928 0.2434
7 5.6257 5.552 0.0054
8 5.8378 5.811 0.0007
9 6.0578 5.508 0.3023
10 6.2862 6.255 0.0010
11 6.5148 5.955 0.3134
Sum square error 1.1766
Mean square error 0.1070
Root mean square error 0.3271
Coefficient of determination (r2) 0.6884

Table 2 Comparison between predicted concentration and true concentration of SCOD

No Predicted True concentration Squared error
concentration
1 1.44 1.44 0.0000
2 1.3546 1.378 0.0005
3 1.3098 1.152 0.0249
4 1.256 1.281 0.0006
5 1.2093 1.074 0.0183
6 1.17 1.022 0.0219
7 1.1381 1.1 0.0015
8 1.1141 1.035 0.0063
9 1.0982 1.046 0.0027
10 1.0906 1.066 0.0006
11 1.0915 0.968 0.0153
Sum square error 0.0926
Mean square error 0.0084
Root mean square error 0.0917
Coefficient of determination (r2) 0.6166

3.2 Validation of the Kinetic Model

Prior to the determination of the kinetic constants (k1–k7), it is appropriate to

clarify the meaning of the variables SVSS and SVDS. The variable SVSS represents
the biodegradable VSS fraction, whilst SVDS represents the biodegradable VDS
fraction minus the TVA concentration (expressed as COD). With the aim of
obtaining the biodegradable suspended organic matter (SVSS) and biodegradable
182 N. Zainol et al.

Table 3 Comparison between predicted value and true value of TVA concentration
No Predicted True concentration Squared error
concentration
1 0.17 0.17 0.0000
2 0.161 0.168 0.0000
3 0.1558 0.1612 0.0000
4 0.1487 0.136207 0.0002
5 0.1416 0.1283 0.0002
6 0.1344 0.118 0.0003
7 0.1273 0.12 0.0001
8 0.1201 0.12845 0.0001
9 0.1129 0.1154 0.0000
10 0.1056 0.110322 0.0000
11 0.0987 0.1039 0.0000
Sum square error 0.0009
Mean square error 0.0001
Root mean square error 0.0088
Coefficient of determination (r2) 0.8506

Table 4 Comparison between predicted value and true value of methane yield
No Predicted yield True yield Squared error
1 0.13 0.13 0.0000
2 0.248 0.2717 0.0006
3 0.3168 0.268865 0.0023
4 0.4086 0.3156 0.0086
5 0.5005 0.4573 0.0019
6 0.5923 0.514 0.0061
7 0.684 0.746174 0.0039
8 0.7758 0.746174 0.0009
9 0.8676 0.695366 0.0297
10 0.9593 0.952156 0.0001
11 1.0479 0.952156 0.0092
Sum square error 0.0631
Mean square error 0.0057
Root mean square error 0.0758
Coefficient of determination (r2) 0.9229

soluble organic matter (SVDS) fractions, the corresponding non-biodegradable

fractions were previously calculated. Using Matlab software (version 6.1) with the
condition of obtaining a minimum value of the sum of the squares of the differ-
ences between the experimental and theoretical values, the kinetic constants
(k1–k7) of the model were calculated according to the following initial condition:
(SVSS)0 = 4.5 g NSCOD/l; (SVDS)0 =1.44 g SCOD/l; (STVA)0 = 0.17 g TVA/l; and
(SM)0 = 0 g COD/l. These initial concentration values were obtained from the data
Kinetic Modeling of Biogas Generation 183

of raw BSW, taking into account the abovementioned non-biodegradable con-

centration values. Using the software with the experimental values and the initial
values above described, the following values for the kinetic constants with their
corresponding standard deviations were obtained: (a) hydrolysis and solubilization
of organic matter: k1 (kinetic constant for non-soluble organic matter degradation):
0.0037 day-1; k2 (maximum rate of soluble organic matter production): 0.0241 g
soluble chemical oxygen demand (SCOD)/l day; k3 (saturation constant): 0.0236 g
SCOD/l; (b) acidogenesis: k4 (maximum rate of soluble organic matter degrada-
tion): 0.0086 g SCOD/l day; k5 (saturation constant): 0.0189 g SCOD/l; and
(c) methanogenesis: k6 (maximum rate of total volatile acids (TVA) consumption):
0.0092 g TVA/l day; and k7 (saturation constant): 0.0003 g TVA/l. Similar values
to those obtained in this work for the kinetic constant k1 were reported in the
literature [4, 7, 9, 10, 12, 14]. The proposed equations were validated by
comparing the theoretical curves obtained with the corresponding experimental data.
Figures 1, 2, 3, and 4 show the comparison of the experimental data of biodegradable
non-soluble organic matter (NSCOD), biodegradable soluble organic matter
(SCOD), TVA concentration and methane production, respectively, with the cor-
responding theoretical values obtained by the above equations and represented by
solid curves. Tables 1, 2, 3, and 4 show the predicted and true values of NSCOD,
SCOD, TVA concentration and methane production, respectively, with the corre-
sponding values obtained for sum square error (SSE), mean square error (MSE), root
mean square error (RMSE) and coefficient of determination (r2).

4 Conclusion

The values of r2 obtained in all cases (range from 0.6116 to 0.9229) demonstrate the
suitability of the kinetic model proposed. The first order and Michaelis–Menten
kinetic models can describe the hydrolysis, acidogenesis and methanogenic stages of
this anaerobic process and the kinetic parameters obtained represent the activity of
the different microorganism types effecting the anaerobic digestion of this waste in
two stages system.

References

1. Beccari, M., Bonemazzi, F., Majone, M., Riccardi, C.: Interaction between acidogenesis and
methanogenesis in the anaerobic treatment of olive oil mill effluents. Water Res. 30(1), 183–
189 (1996)
2. Borja, R., Rincon, B., Raposo, F., Alba, J., Martın, A.: A study of anaerobic digestibility of
two-phases olive mill solid waste (OMSW) at mesophilic temperature. Process Biochem. 38,
733–742 (2002)
3. Borja, R., Rincon, B., Raposo, F., Alba, J., Martín, A.: Kinetics of mesophilic anaerobic
digestion of the two-phase olive mill solid waste. Biochem. Eng. J. 13, 139–145 (2003)
184 N. Zainol et al.

4. Cuevas, G., Gonzalez, O., Gonzalez, S.: Wastewater fermentation and nutrient removal in
sequencing batch reactors. Water Sci. Technol. 38(1), 255–264 (1998)
5. Garcıa, F., Santos, V.E., Naval, L., Guardiola, E., Lopez, B.: Kinetic model for anaerobic
digestion of livestock manure. Enzyme Microbiol. Technol. 25(1–2), 55–60 (1999)
6. Greenberg, A.E., Clesceri, L.S., Eaton, A.D.: Standards methods for the examination of water
and wastewater, 18th edn. American Public Health Association, Washington (1992)
7. Gujer, W., Zehnder, A.J.B.: Conversion process in anaerobic digestion. Water Sci. Technol.
15, 127–167 (1983)
8. Kalyuzhnyi, S.V., Davlyatshina, M.A.: Batch anaerobic digestion of glucose and its
mathematical modeling. I. Kinetic investigations. Bioresour. Technol. 59 (1), 73–80 (1997)
9. Keshtkar, A., Meyssami, B., Abolhamd, G., Ghaforian, H., Asadi, M.K.: Mathematical
modeling of non-ideal mixing continuous reactors for anaerobic digestion of cattle manure.
Bioresour. Technol. 87, 113–124 (2003)
10. McCarty, P.L., Mosey, F.E.: Modeling of anaerobic digestion processes (A discussion of
concepts). Water Sci. Technol. 24(8), 17–33 (1991)
11. Mosche, M., Jordening, H.J.: Comparison of different models of substrate and product
inhibition in anaerobic digestion. Water Res. 33(11), 2545–2554 (1999)
12. Mosey, F.E.: Mathematical modelling of the anaerobic digestion process: regulatory
mechanisms for the formation of short-chain volatile acids from glucose. Water Sci. Technol.
15, 209–232 (1983)
13. Munch, V.E., Keller, J., Lant, P., Newell, R.: Mathematical model of prefermenters-I. Model
development and verification. Water Res. 33(12), 2757–2768 (1999)
14. Pavlostathis, S.G., Giraldo, E.: Kinetics of anaerobic treatment. Water Sci. Technol. 24(8),
35–39 (1991)
15. Valentini, A., Garuti, G., Rozzi, A., Tilche, A.: Anaerobic degradation kinetics of particulate
organic matter: a new approach. Water Sci. Technol. 36(6–7), 239–246 (1997)
 Part IV
Related Problems
Simulation of the Errors of Refraction
in the Human Eye

H. D. Silva, L. C. P. Dória, C. P. Dória, C. P. Dória, M. C. T. D. Silva,

H. D. Silva Jr. and M. T. D. Silva

Abstract This work is a review of the mechanism of refraction in the human eye,
to understand the deterioration of visual acuity. It is the conclusion of a work to
10 years for recovery of the presbyopia by the first author of this chapter doing
physical exercises with the extra-ocular and ciliary muscle. It was Scheiner in
1619 apud (Werner et al. Arquivos Brasileiros de Oftalmologia, São Paulo, Brasil
2000), who proved in his experiments, made with holes in a card, in which an
object is seen in each direction at a different distance. This corresponds to say that
the person sees an object as projection images multiple superimposed in different
positions on the retina, then, characterizes the effect resulting from the formation

H. D. Silva (&)
Universidade Federal de Pernambuco, Recife, PE, Brazil
e-mail: [email protected]
L. C. P. Dória
UPE, Recife, PE, Brazil
e-mail: [email protected]
C. P. Dória
H. D. Silva Jr.
M. T. D. Silva
UNICAP, Recife, PE, Brazil
e-mail: [email protected], [email protected], [email protected]
H. D. Silva Jr.
e-mail: [email protected]
M. T. D. Silva
e-mail: [email protected]
C. P. Dória
FBV, Recife, PE, Brazil
e-mail: [email protected]
M. C. T. D. Silva
UNIBRATEC, Recife, PE, Brazil
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 187
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_15,
Ó Springer-Verlag Berlin Heidelberg 2012
188 H. D. Silva et al.

of lenses, originating from the clustering of metabolic secretions trapped in the

cornea or lens without any organization. Through software, it is simulated the
monocular vision of an object perceived by a patient, under the effect of metabolic
secretions. The simulated object can be produced by a light source or the refraction
of light, which causes different effects. In the image simulation of a text written by
a pencil, on a lined paper, produced by the reflection of light, the text is erased and
only the guidelines are displayed, as could happen with a person. It is shown, for
monocular vision, the formation mechanism of the myopia, hyperopia, astigma-
tism and presbyopia due to the accumulation of metabolic secretions in the cornea
or lens, and for binocular vision, it is justified the possibility of an individual
having photosensitive epilepsy due to error refraction stimulated by colored beams
and variable intensity emitted by television.

Keywords Human eye Intraocular pressure
Errors of refraction
Photosen-


sitive epilepsy Seizure

1 Introduction

Explanations of how accommodation occurs have been speculated upon for cen-
turies. By reviewing the literature on accommodation and presbyopia one finds
much that is assumed to be known is still controversial [1].
The term presbyopia means ‘‘old eye’’ and is a vision condition involving the
loss of the eye’s ability to focus on close objects. It is a condition that occurs as a
part of normal aging and is not considered to be an eye disease. The process occurs
gradually over a number of years. Symptoms are usually noticeable by the age of
40–45. The symptoms begin with an annoying inertia of focus when gazing from
far to objects near and decline the ability to focus on near objects to carry out
prolonged work without stinging, smarting, or tearing, which eventually leads to
disinterest in reading. Fine print and small targets can no longer be resolved at the
costumary reading distance, and when the object is habitually brought nearer, the
blur strangely increases. Soon, even ordinary print begins to blur, smudge, smear,
run together, and disappear. These symptoms are intensified under inadequate
light, amplified by poor contrast, and exaggerated at the end of the day [1].
The extra ocular muscles, considering their relatively small size, are incredibly
strong and efficient. There are six extra ocular muscles that control the movements
of the (human) eye, which act to turn or rotate an eye about its vertical, horizontal,
and antero-posterior axis: medial rectus (MR), lateral rectus (LR), superior rectus
(SR), inferior rectus (IR), superior oblique (SO), and inferior oblique (IO). The
actions of the extra ocular muscles depend on the position of the eye at the time of
muscle contraction [2].
The contractions of the oblique muscles occur when the visual axis of the eye is
in the nasal direction, causing a prismatic effect in the cornea due the conical
projection, to compensate for the distance between the eyes. The relaxation of the
Simulation of the Errors of Refraction in the Human Eye 189

oblique muscles occurs when the visual axis of the eye is in the temporal direction
or when the eye is in rest. The oblique muscles movement changes the intraocular
pressure that is the mechanism of forced convection for the transport of nutrients
and drainage of secretions metabolic in cornea [3].
The shiny and transparent cornea, having slightly greater curvature than the rest
of the globe, constitutes the anterior one-sixth of the outer coat of the eyeball. The
horizontal diameter, 11.7 mm, is greater then the vertical diameter, 10.6 mm.
The cornea is more curved and thinner centrally 0.58 mm, while its peripheral part
is less curved and thicker, i.e. 1 mm. Anteriorly it looks elliptical, while it is
circular posteriorly. The radii of curvature of the anterior and posterior surfaces are
7.8 and 6.6 mm, respectively. The cornea can be divided into the cornea proper—
transparent and avascular and the limbus—1 mm transition zone, richly vascular
(Limbus, scam, i.e. the line of junction between two edges). The normal water
content of the cornea is between 75 and 80%. The nutrition of the cornea is derived
from three sources: (a) aqueous humour; (b) exudation from the perilimbal vessels;
and (c) precorneal tear film [4].
The aqueous humour comprises about 4% of the total volume of the eye and
represents the ocular tissue fluid. It maintains the intraocular pressure and supplies
nutrition to the avascular structures, namely the cornea and lens. The aqueous
humour formed by the ciliary body, comes to the posterior chamber and leaves the
eye by two modes of drainage: the conventional or the unconventional outflow.
The conventional outflow represents 85–90% of the drainage. The aqueous
humour comes from the posterior chamber, flows between the iris and lens into the
anterior chamber, and then to be filtered out of the eye via the trabecular
meshwork into the canal of Schlemm. This pathway is principally dependent on
the relationship between the intraocular pressure and pressure in the exit veins
situated at the angle of the anterior chamber. It is estimated that about 1% of the
fluid in the anterior chamber drains away per minute [4].
The unconventional outflow or uveoscleral outflow represents 5–15% of the
drainage. The drainage occurs through the stroma and vessels of the iris root and
ciliary body, and flows backward to leave the eye via supraciliary and supra-
choroidal spaces to finally reach the orbital vessels [4]. Figure 1 shows a sche-
matic sketch of the right eye.
The iris is a contractile structure, consisting mainly of smooth muscle,
surrounding the pupil. Light enters the eye through the pupil, and the iris regulates
the amount of light by controlling the size of the pupil. The iris contains two groups
of smooth muscles; a circular group called the sphincter pupillae, and a radial group
called the dilator pupillae. When the sphincter pupillae contracts, the iris decreases
or constricts the size of the pupil. The dilator pupillae, innervated by sympathetic
nerves from the superior cervical ganglion, causes the pupil to dilate when they
contract. These muscles are sometimes referred to as intrinsic eye muscles [4].
The crystalline lens is part of the anterior segment of the eye and it is trans-
parent, biconvex and semisolid, in the adult. Anterior to the lens is the iris, which
regulates the amount of light entering into the eye. The lens is suspended in place
by the zonular fibers, which attach to the lens near its equatorial line and connect
190 H. D. Silva et al.

Conventional
outflow
Iris

Unconventional Visual axis Optic Axis

outflow Anterior chamber
Ciliary (aqueous humour)
body
Iris Pupil
Zonules Lens Posterior
Lens
chamber
Aqueous flow Ciliary
muscle
Vitreous
Uveal humour Ora
meshwork Cribriform serrata
meshwork Retina
Schlemm’s
canal Choroid
Trabecular Hyaloid
meshwork canal
Fovea
Optic disc
Collector channel Optic nerve

Fig. 1 A schematic sketch of the right eye with details of the flow of aqueous humour

the lens to the ciliary body. Posterior to the lens is the vitreous body, which, along
with the aqueous humour on the anterior surface, bathes the lens. It is enclosed by
a capsule that has a biconvex ellipsoid shape. The radius of the anterior surface is
9 mm while that of the posterior surface is 5.5 mm. In the adult, the lens has a
variable diameter between 9 and 10 mm and has an axial thickness 3.6 mm [4].
The lens, by changing shape, functions to change the focal distance of the eye so
that it can focus on objects at various distances, thus allowing a sharp real image of
the object of interest to be formed on the retina.
The vitreous humour is the transparent, colorless, gelatinous mass that fills the
space between the lens of the eye and the retina lining the back of the eye. It is
produced by certain retinal cells. It is composed mainly of water (99% of its mass
of about 3.9 g), with salts, sugars, vitrosin (a type of collagen), a network of
collagen type II fibers with the mucopolysaccharide hyaluronic acid, and also a
wide array of proteins in micro amounts [5]. The vitreous humour has a viscosity
two to four times that of pure water, giving it a gelatinous consistency. It also has a
refractive index of 1.336 [5] and its volume is about 4 ml [4]. The gel in the
vitreous chamber is stagnant, therefore, if blood, cells or other byproducts of
inflammation get into the vitreous, they will remain there (floaters), unless
removed surgically. If the vitreous pulls away from the retina, it is known as a
vitreous detachment. Although the vitreous is in contact with the retina and helps
to keep it in place by pressing it against the choroid, it does not adhere to the
retina, except in three places: around the anterior border of the retina; in the
macula and at the optic nerve disc (where the retina sends about 1.2 million nerve
fibres (axons) to the brain). The vitreous keeps the format of eyeball.
The retina is a complex, layered structure with several layers of neurons
interconnected by synapses. The only neurons that are directly sensitive to light are
Simulation of the Errors of Refraction in the Human Eye 191

the photoreceptor cells. These are mainly of two types: the rods and cones. Rods
function mainly in dim light and provide black-and-white vision, while cones
support daytime vision and the perception of colour. A third, much rarer type of
photoreceptor, the photosensitive ganglion cell, is important for reflexive
responses to bright daylight. In vertebrate embryonic development, the retina and
the optic nerve originate as outgrowths of the developing brain, so the retina is
considered part of the central nervous system (CNS). It is the only part of the CNS
that can be visualized non-invasively [5]. Therefore, the retina, through their
optical sensors, captures and discretizes the continuous image that is projected on
it and transmits to the brain via the optic nerve. The region of the projected image
onto the optical disc is lost.
The retina is subdivided into two parts: the central or macular and the
peripheral [4].
Central Retina or Macula Lutea—the specialized region of the retina is 3 mm
or 2-disc diameter temporal to the optic disc. The central retina can be subdivided
into three areas, the fovea, parafovea and perifovea.
The fovea centralis is a central depression in the macula lutea. It shows a bright
reflex seen by the ophthalmoscope. It is the most sensitive part of the retina and
contains only cones.
The parafovea is a 2.1 mm wide area all round the fovea and contains both
rods and cones arranged alternately.
The perifvea is a 1.5 mm wide area around the parafovea and in this region are
two rods between each cone.
Peripheral Retina—may be divided into four zones: (a) the near periphery;
(b) the mid periphery; (c) the far periphery; and (d) the ora serrata or extreme
periphery.
The near periphery. The near periphery is the 1.5 mm wide area around the
macula characterized by the absence of Henle’s fibre layer and the presence of
thicker cones surrounded by a collar of rods.
The mid periphery. It is 3 mm wide. It is characterized by interrupted ganglion
cells and thicker cones, the cones being separated from each other by at least three
rods.
The far periphery. This is 9–10 mm wide temporally and 16 mm wide nasally.
At this region there are large and widely-spaced ganglion cells and the number of
cones reduced, the cones having shorter outer segments.
The ora serrata. The ora serrata is 2 mm wide temporally and 0.7 mm wide
nasally. There is gradual disappearance of the rods and replacement with mal-
formed cones, disappearance of the outer molecular layer and fusion of the rods
and cones. At 0.5 mm before the termination of the retina the rods and cones,
ganglion cells and nerve fibre layer tease.
The cornea has the optical power fixed by the lens [5]. This statement is
presented in all publications, but in [3] it is shown that the cornea has its dioptric
power variable with the direction of the visual axis of the eye, in turn, varies
transiently the intraocular pressure and the loss of motor coordination of the
oblique muscles can cause different deformations in the eyeball.
192 H. D. Silva et al.

Myopia, hyperopia, astigmatism and presbyopia are among the highest

incidence of refraction errors in the human eye. Without seeking to describe the
refractometric optical method, the analysis of the direction effect of oscillation of
the light beam produced on the retina with the direction of oscillation of the light
source incident on the eye, defines the conditions of refraction (if the condition of
observation is equal to the examiner at infinity) [5]:
(a) in a myopic eye, the light beam on the retina moves in the opposite direction of
the light source incident on the eye;
(b) in the hyperopic, the light beam on the retina moves in the same direction of
the light source incident on the eye;
(c) in emmetropic, it will not notice any movement, but a diffused illumination on
the retina will be noticed.
Astigmatism has diopter of different values for the refractometric study done to
a vertical plane and then to a horizontal plane [5].
Presbyopia is a universal and irreversible process [5] but in [3] it can be
reversible.

2 Objectives

This work indicates that the refraction error is caused by the photo-physical
problem of overlapping projection of the same image in different positions, as is
the understanding of the experience of Scheiner in 1619 apud [1] also indicates the
biophysical cause of this overlap projection on the retina, considering the team’s
experience in the treatment of presbyopia of the first author of this chapter, and
that the cornea, the aqueous humour, the iris, the lens and the vitreous humour
form a single biconvex lens of layers with significant differences in refractive
indices that always transmit a well focused image projection on the retina.
The lens has the ability to increase projection of the object of interest and the
photo-physical function of the pupil to avoid instability of focus that can occur
when there is diffused strong light or when the object is close to the cornea.

3 Method

The aim is to show that the error of refraction is caused by the photo-physical
problem of overlapping projection of the same image in different positions. This
was elaborated by a simple photo-physical scheme for studying the effects of the
projections and used a simulation software of overlapping images. Several ways of
overlapping were performed and the resulting images that best reproduced the
visual effects are reported here.
Simulation of the Errors of Refraction in the Human Eye 193

Fig. 2 Schematic sketch of the pinhole camera with formation of the hypermetropia and myopia

3.1 Photo-Physical Scheme

To understand what happens in the eye, consider Fig. 2. A box with a tiny hole is
placed at one end and a film or photographic paper is placed at the other. The flat
face with a hole is replaced by a calotte. Several holes are made in this calotte.
Each hole projects an image on different locations of the film without deformation
and with the same size. The curvature of the calotte determines the distance that
the projected images intersect at an imaginary plane.
The understanding of the experience of Scheiner in 1619 apud [1] can be
replicated if different diopter lenses are placed in the holes of the scheme in Fig. 2,
i.e. the holes are aimed in different directions and project images in different sizes
that can be interpreted at different distances.

3.2 Simulation

There are two situations to be simulated: in the first situation all regions are
projected on the retina and in the second situation only the important part is
projected on the retina without the influence of the surroundings. For ease of
explanation, the first situation described will be called ‘‘total projection’’ and the
second situation will be called ‘‘partial projection’’.

3.2.1 Total Projection

The same simulation methodology can be performed with any computer image
processor program that admits the use of a custom filter, enabling the form of overlap.
194 H. D. Silva et al.

The resulting image is defined mathematically as the convolution of the custom filter
with the initial image.
This simulation methodology was applied to two cases: the preview of a
landscape, and the display of writing.
The visualization of a landscape can be simulated from a color photograph that is
characterized as an image with various colors and different intensities whereas
writing is characterized as being monochromatic and having only one intensity,
however, the surroundings do not interfere in the text display. The resulting image is
defined mathematically as the convolution of the custom filter with the initial image.

3.2.2 Partial Projection

The simulation is simple, we used the Microsoft PaintÒ software without choosing
‘‘opaque drawing’’, the commands ‘‘select all’’, ‘‘copy’’ and ‘‘paste’’ moving
horizontally or vertically. The resulting image is mathematically defined as the
logical sum of the set of overlapping images.

4 Results

We conducted three sets of simulations using the overlapping of images, to

reproduce some symptoms described in interviews with people who suffer from
refractive errors.
Some lectures reported that when they are not wearing corrective lenses, they
have difficulties to understand the writing of their students in pencil, but had no
trouble to read the writing done in ink. The writing in pencil appears as if nothing
was written. The first set of simulations, shown in Fig. 3, presents the interference
of the background color on the text, described as being only the perception of the
lines in the writing with a thin, black pencil on a lined piece of paper. This set of
simulations shows that increasing the number of overlapping images can cause an
inability to understand the writing as described by some people who said that
without corrective lenses, they did not recognize if there was any writing, they
only recognized the lines.
The original text and four simulations are shown in Fig. 3, where they are
exposed image overlays, using the total projection in a system with 4, 16, 64, 256
and 1024 simulated lenses, respectively. The foil reflection is projected onto the
retina. The image that is projected depends on the layout of the simulated lenses,
i.e. not all people have this symptom.
Some computer users mentioned the difficulty to understand a text on a black
background which becomes very blurry. However, they were able to understand
without difficulty a black text on a white background. The second set of simula-
tions, shown in negative image, Fig. 4, presents the interference of the text on the
black background. In this case the text formed is the negative of the light source.
Simulation of the Errors of Refraction in the Human Eye 195

Fig. 3 Simulation of a text using total projection

25

50

100

Fig. 4 Simulation of a text using the particular projection (negative image)

The original text and the three simulations are presented in negative image,
Fig. 4, where they are exposed image overlays, using the particular projection, in a
system with 5, 25, 50 and 100 simulated lenses, respectively. In this case the text is
projected on the retina. The observation of a star can result in this difficulty. The
display depends on the arrangement of simulated lenses, i.e. not all people have
this symptom.
For people with difficulties in driving cars at night there were two types of
reports: a person who had undergone surgery for myopia reduction reported that
the lights of the vehicles coming from the opposite direction were impairing the
safe driving of their car while another person reported difficulty in driving safely
196 H. D. Silva et al.

Fig. 5 Simulation of a photograph using total projection

because of a loss in sharpness. Simulation was not performed for the first case,
because the light beams formed from the light source are not about image over-
lapping. However, the second report obtained from the third set of simulations,
shown in Fig. 5, presents the simulation of the interference between all elements of
photography. Therefore, the individual does not only see poorly at night, but also
requires a little time for their eyes to adjust to brightly lit areas as the contrast
vision may also be greatly reduced. The lens takes longer to change the curvature
because of the volume of accumulated secretions.
The original photograph and two simulations are shown in Fig. 5 which illus-
trates the simulation of the refractive errors in the photograph projected with
image overlays, using total projection. Two simulations are presented, in a system
with 1024 and 65536 simulated lenses, respectively. The simulations with 1024
simulated lenses shown in Figs. 3 and 5 have the same layout.
There were reports of people who claimed they only needed corrective lenses
for reading, because they had good vision for distances, though, the doctor has
prescribed corrective lenses to focus in both situations, near and far. This mistake
is explained in the simulations with 1024 overlapping images shown in Figs. 3
and 5. These simulations were performed with the same array of overlapping
images. The simulation shown in Fig. 5 may seem acceptable, due to the fact that
the image degradation occurs slowly, but reading is impossible in the simulation
presented in Fig. 3.
The superimposition of images was the understanding of the experiment of
Scheiner in 1619 apud [1], observing through a hole in a card that the image was
perceived at different distances in different directions. In the three sets of simu-
lations where images of the same size (same distance) were overlaid in different
directions there was a simplification of the model.
Simulation of the Errors of Refraction in the Human Eye 197

When myopia or hyperopia is not acute and the patient’s symptoms become
more pronounced at age 40, we can call it presbyopia.
The dilated pupil exposes a larger number of agglutinated metabolic secretions
that may prevent good viewing as another imaginary surface of intersection may
occur in which case the corrective lens will be different.

5 Discussion

The simulations presented in this chapter show that the projection of overlapping
images produces a visual effect similar to visual observations noted by patients
with refractive errors. A simplified photophysical scheme was also presented
which produces overlapping images and is compatible with the criteria for error
detection refraction. Biophysical interpretation for the eye to produce the over-
lapping images is described in [3].
The simulations presented in Fig. 4 show that 25 lenses, formed by aggluti-
nations of metabolic secretions are enough to reduce visual acuity. It is important
to note that the agglutination of metabolic secretions is a slow process of growth of
the agglomerations that form the lenses and when it causes a significant reduction
of visual acuity, the process of removing these lenses does not result in a gradual
recovery of the visual acuity. There is a hysteresis phenomena, i.e. the route of
recovery of visual acuity is not the opposite of the loss of visual acuity using the
volume of accumulated secretions as an independent variable. The visual acuity of
the first author of this chapter improved only after 10 years of work.
The analogy of the scheme is shown in Fig. 2, with the eye: the film corre-
sponds to the retina and the imaginary plane corresponds to an imaginary parallel
surface similar in form to the retina surface. To check the effect of myopia and
hyperopia in the scheme shown in Fig. 2, an intermittent light beam is first passed
only across the holes in a horizontal line and then in a vertical line. If the film is
after the imaginary plane of the intersection of projections, we have myopia and
the light beam on the film moves in the opposite direction of the intermittent light
source on the calotte. Otherwise if the film is before the imaginary plane of
the intersection of the projections, we have hypermetropia or emmetropia.
If the imaginary plane of the intersection of projections is not parallel to the plane
of the film, we have astigmatism. If the illumination is diffused on the retina there
is an accumulated secretion in the eye.
The evolution of refractive error may be perceived by the individual when a
small source of weak light is observed: to assess the evolution of treatment a red
light source in the form of a circle should be used, such as the brake indicator light
found in some cars, then the horizontal and vertical dispersion can be measured in
a quantitative diameters. Ten years ago, when the exercises began, the first author
of this chapter had a horizontal dispersion with three diameters and a vertical
dispersion with two diameters.
198 H. D. Silva et al.

The analysis of refractive error can be performed by passing a card close the
eye very slowly while looking at a round, green light. Move the card horizontally
from left to right, then in the opposite direction. Afterwards, you should swipe the
card vertically upward, then in the opposite direction. Afterwards, you should
swipe the card vertically upward, then in the opposite direction. If, with every
movement, suppression of the light source starts in the opposite direction you have
hyperopia, if the suppression of the light source begins in the same direction you
have myopia, however, if the suppression of the source starts in different directions
you have astigmatism.
It is important to mention that myopia causes a greater corneal curvature, while
hypermetropia causes a lesser corneal curvature. When the human eye sees objects
in the distance that appear smaller than objects close by this is known as per-
spective. It is important to note that myopia forms a smaller image, giving the
feeling that it is farther away, while hypermetropia forms a larger image giving the
feeling that it is closer. A person with myopia has difficulty seeing distant objects
because the number of photo sensors that capture the image projected on the fovea
is insufficient to produce information necessary for understanding to the brain.
When the person with myopia approaches the object, its projected image is larger,
therefore its image is captured by a larger number of photo sensors that may be
sufficient to produce the information necessary for the brain to comprehend them.
A person with hyperopia has a lot of difficulty seeing nearby objects because the
projected image can be outside of the fovea where there is an insufficient number
of photo sensors to produce the information necessary for the brain to understand.
When the person with hyperopia moves away from the object the image can be
projected onto the fovea and the number of photosensors can be sufficient to
produce information necessary for the brain to understand.
Figure 4 also shows what can occur in the evolution of the treatment of pres-
byopia. Initially the individual can distinguish only the paragraphs of a text, as
simulated with 100 overlaps, after the individual can move on to distinguish the
lines, as simulated with 50 overlaps. Next they can see the outlines of the words, as
simulated with 25 overlaps, then distinguish the outlines of the letters, as simulated
with 5 overlaps and finally they are able to read it.

6 Conclusion

The simulations presented in this chapter show that the projection of overlapping
images produces a visual effect similar to visual observations noted by patients
with refractive errors. A simplified photophysical scheme was also presented
which produces overlapping images and is compatible with the criteria for error
detection refraction. Biophysical interpretation for the eye to produce the over-
lapping images is described in publication [3].
The oblique muscles cause changes in the curvature of the eye about its antero-
posterior axis (the visual axis). The oblique muscles cause a prismatic refraction in
Simulation of the Errors of Refraction in the Human Eye 199

the cornea and they vary the IOP to the mass transfer in the cornea and anterior
chamber. In binocular vision, the angular motions of the visual axis causes a
prismatic refraction of variable-scale, base-out 30D to base-in 8D, in the cornea
and, consequently, varies the intraocular pressure [3]. The iris prevents the return
of aqueous humor to the posterior chamber when the intraocular transient pressure
increases and facilitates the passage of aqueous humor to the anterior chamber
when the intraocular transient pressure decreases. The iris is the check valve
diaphragm. Any problem that prevents the metabolic secretions forced convection
facilitates the agglutination of these secretions and results in the formation of small
bubbles that behave as lenses without a well-defined shape, which are translucent,
absorb light, reduce the visual acuity, increase the viscosity through dehydration
by decanting, cause instability in the focusing consequently causing movement of
the projecting light beam on the retina, and in addition produces the simplified
effects presented in the model of Fig. 2.
The television image is exposed on a plane and it is projected in the same size
on the two retinas of the human binocular system due to the focusing mechanism
of the eyes. However, the refractory error in each eye makes different projected
images. The images displayed on television or computer displays that form
patterns in time or space, such as flashing lights, bold, regular patterns, or regular
moving patterns and stroboscopic light cause the brain to look for a three-
dimensional understanding of a two-dimensional movement because the images
projected on the retinas are different. If the rate of change of events and the
focusing speed have the same order of magnitude, the process can trigger resonant
results (the instant of the command feedback in disagreement with the instant of
system action) then the progressive increase to the brain activity can cause
seizures, photosensitive epilepsy, a problem cited in [6]. The binocular visual
focusing system is a complex nonlinear process.
Metabolic secretion that is stagnant for a long time can dehydrate by decanting
increasing the absorption of light in the eye and may even form myodesopsia in the
cornea or in the lens also known as floaters. If the volume of metabolic secretions
accumulated in the lens hinders one’s accommodation, it is presbyopia.
The cornea, the aqueous humour, the iris, the lens and the vitreous humour form
a structure of layers with significant differences in refractive indices of a single
biconvex lens that has control of light energy input. Every image projected on the
retina is always well focused. To have an excellent binocular visual acuity, one
must have 100% of monocular visual acuity to close-up objects and to distant
objects, with the dilated pupil.

Acknowledgments To God who helped me, the first author, to overcome early childhood
problems and many years of suffering which caused me to become an analytical observer of the
events in my own body: To my brother Marcos, who 10 years ago recommended the first book
regarding the work of vision correction: To all my children for all their long and difficult
participation in this work: To my fellow members of ANDES-SN, Sindicato Nacional, and its
members of the Section ADUFEPE who, during the intervals of some of the society’s meetings,
provided important scientific information for the study: To those who were interviewed, to
200 H. D. Silva et al.

friends, colleagues and family members who directly or indirectly have contributed to this work.
The other authors are in agreement with the first author.

References

1. Werner, L., Trindade, F., Pereira, F., Werner, L.: Physiology of accommodation and
presbyopia. Arq. Bras. Oftalmol. 63(6) (2000)
2. Information on http://www.tedmontgomery.com/, on 04/03/2010
3. Silva, H.D., Dória, L.C.P., Dória C.P., Dória C.P., Silva, M.C.T.D., Silva, H.D. Jr., Silva,
M.T.D.: Nutrients and metabolic secretions transfer in cornea, Accepted for Defect and
Diffusion Forum (2011)
4. Ahmed, E.: In: A Textbook of Ophthalmology. Phi Learning and Prentice hall of india, India
(2001)
5. BICAS HEA: Oftalmologia fundamentos. Ed. Contexto, São Paulo (1991)
6. G. F.A., Harding and P. M. Jeavons.: Photosensitive Epilepsy, New Edition, Lavenham Press,
London (1994)
Human Gait: Kinematics Analysis
and Mechatronic Simulation

Alvaro Joffre Uribe, João Maurício Rosário

and José Tenreiro Machado

Abstract This work presents a kinematic and motion planning analysis for human
gait simulation using a mechatronic model. The forward and inverse kinematics for
calculating either positions or rotations of the lower member segments is based on
collected data from a walking person. The trajectories are reconstructed using spline
interpolation from the most representative motion points during gait in order to
define different parameterized positions for each articulation. Both kinematics and
motion planning allow obtaining suitable data for testing the mechatronic model
which is designed considering direct current motors as muscle actuators for the knee
and waist joints. Cylindrical objects are used for representing the thigh and shank
along with their configured mechanical properties. Finally, through the use of Matlab
r
SimMechanicsTM toolbox, interactions and dynamics between the lower member
mechanism and the ground were simulated with motion captured data and experi-
mental data from motion planning calculations. The motion planning results sug-
gested that statistical data lead to gait paths not suitable for reproducing on a person,
as each joint position does not match the subject’s real one. The simulation
results from SimMechanicsTM model, suggested that further components have to be

A. J. Uribe (&)  J. M. Rosário

Faculty of Mechanical Engineering, University of Campinas (UNICAMP),
Rua Mendeleiev, 200, Campinas, SP, Brazil
e-mail: [email protected]
J. M. Rosário
e-mail: [email protected]
A. J. Uribe  J. M. Rosário
Cidade Universitária ‘‘Zeferino Vaz’’, Barão Geraldo, Campinas, SP, Brazil
J. T. Machado
Department of Electrotechnical Engineering,
Institute of Engineering of Porto (ISEP),
Rua Dr. Antnio Bernardino de Almeida, 431, Porto, Portugal
e-mail: [email protected]

A. Öchsner et al. (eds.), Analysis and Design of Biological Materials and Structures, 201
Advanced Structured Materials 14, DOI: 10.1007/978-3-642-22131-6_16,
Ó Springer-Verlag Berlin Heidelberg 2012
202 A. J. Uribe et al.

considered in order to obtain more accurate trajectories thus, minimizing the error
when comparing the obtained path with the motion captured data.

Keywords Lower member  Kinematics  Motion planning  Mechatronics 

Simulation

1 Introduction

Human gait studies have yielded various research in motion capture, mechanics,
electronics and control techniques for developing humanoid robotics and orthotic
devices for assisting motion and aiding rehabilitation therapies by increasing some
human capabilities [1–4]. In some studies [5, 6], anthropometric statistic data
served as base for defining basic measures related to lower member parts and gait
parameters. Trajectories calculated for applied robotics in human gait were
initially based on passive walker models over slops or subdue to inertia for cre-
ating gait patterns. These approaches have led to posterior dynamic concepts [7],
like The Zero Moment Point presented in [8], which achieves stability in both
sagittal and frontal planes by measuring the Center of Gravity position of the body
within the sustentation polygon, while keeping track of inertias during gait. Walk
planning studies using techniques such as forward and inverse kinematics [1, 9],
Splines trajectory planning or polynomials interpolations allow obtaining mathe-
matical models for calculating each joint’s position and orientation during regular
or pathological gait [10–12]. The previous studies lead to the design and devel-
opment of various mechanisms and actuator for reproducing human lower member
motion. In Ruthenberg [13], a one degree of freedom (DOF) rotational mechanism
allowed the suitable motion without increasing the user’s weight, while generating
a torque similar to the one present during walking. In Kasaoka [14], a device based
on a ballscrew and a direct current motor with mioelectric inputs allow controlling
and predicting the motion to better assist the user. By considering the effects of
compensatory mechanisms as the main reason of exoskeletal robotic device
rejection due to the high energy required for motion, in Peter [15] a device based
on the knee’s extension and flexion without compensatory elements allowed
motion when no motor functions were present. In Agrawal and Agrawal [16], an
alternative passive solution using a hybrid parallelogram is presented in order to
balance the mechanism’s gravity and minimize torques during the motion.
The present chapter is focused on gait kinematic analysis, as well as joint tra-
jectory planning based on anthropometric measurements and collected data from a
walking person. This data allows developing a mechatronic Matlabr model using the
SimMechanicsTM toolbox for simulating and comparing how accurate normal gait
can be achieved. The purpose of this study is to implement a model for future
development of an assisting lower member device by considering the previous
mentioned parameters. The methodology followed for accomplishing this objective
Kinematics Analysis and Mechatronic Simulation 203

is described in the following sections. In Sect. 2 the kinematic analysis is presented

studying it from its forward and inverse perspective. In Sect. 3, motion planning is
studied for creating suitable trajectories for the lower member articulations.
In Sect. 4, a comparative study of the kinematics and motion planning allows the
selection of the analysis that fits best the reproduction of walking. In Sect. 5, the gait
mechatronics model is presented and its resulting motion compared to the human
one. Finally, in Sect. 6, the conclusions and findings are presented.

2 Kinematics Analysis

The kinematics problem can be solved using the forward or the inverse analysis,
either one has advantages and disadvantages depending on whether the system has
one or multiple solutions. For this project, both approaches are useful as during lower
limb motion, known rotations may lead to a target joint position (such as when
avoiding an obstacle), and known paths allow the calculation of suitable angles for
adjusting the positions during normal walk. Biomechanics characteristics are studied
for modeling a suitable kinematics representation of the lower member which is
composed of the thigh, the shank and the ankle, each having three DOF [17, 18].
During walk the most notorious movements are the extension and the flexion rota-
tions, although the abduction and the adduction rotations are internally present, they
are not considered in order to simplify the model. For solving the kinematics, the
lengths of each element of the mechanism (in this case the thigh and the shank) are
needed to define the positions and orientations of the joints (waist and knee). These
measurements can be calculated through anthropometric statistic tables where each
distance depends on a person’s height, or they can be measured directly from a
person. Table 1 presents the afore mentioned measurements, where i is a number
given to each segment and li is the calculated length for the respective i part.
Using the anthropometric data the kinematics problem can be solved through
the forward and inverse analysis. For achieving both solutions, a kinematic model
must be proposed according to the human data and anatomy. The support system
of the lower member is chosen to be represented as a serial kinematic chain with
two rotational joints for performing flexion and extension rotations at the hip (h)
and knee (f) articulations, as can be seen in Fig. 1. It can also be seen that the end
of the chain is identified as the ankle whose left and right positions are defined by
the U and V coordinates containing the vertical and horizontal information, where
xf and yf are the target positions during gait.

2.1 Gait Pattern

Gait is a cycling process that repeats itself every time any foot leaving the ground,
leaves it again after swinging a step forward. It can be divided in two main phases,
the double support which takes place when both feet are in contact with the
204 A. J. Uribe et al.

Table 1 Anthropometric Data

Segment i li Calculated values Captured data values
Pelvis 0 0.10059 1.95 –
Lower member 1 l3 þ l 4 0.8343 0.8343
Upper body 2 0.47 0.91 –
Thigh 3 0.23667 0.4631 0.4631
Shank 4 0.24556 0.3712 0.3712
ankle-foot base – 0.03846 0.075 0.0812
ankle-heel – 0.04853 0.0946 –
ankle-first metatarsal – 0.08901 0.1736 –

Fig. 1 Lower member kinematic model

ground, and the single support when only one foot is. During single support, one
leg remains in a stance position moving as an inverted pendulum and pivoting the
body around its ankle, while the swinging leg balances and behaves like a regular
pendulum. For analyzing how the x and y positions behave during the walking
period, gait data is captured using a motion capture system based on six infrared
cameras and four optical trackers placed around each joint and at the most sig-
nificant anatomic points around the waist, the knee and the ankle. The center of
each articulation is determined by the cross section where the position of each
marker intersects allowing the calculus of both the thigh and shank lengths and
also the lower member initial rotation and joint position, as presented in Fig. 2.
Anthropometric measurements from the motion capture information can be
obtained, as each x, y and z coordinates from the waist, knee and ankle are known.
Through trigonometric calculus in the form of (1), the lengths and initial rotations
of the shank and thigh (l3 and l4 , respectively according to the labels given in
Table 1), can be found.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
l3 ¼ ðwaistx  kneex Þ2 þ ðwaisty  kneey Þ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1Þ
l4 ¼ ðkneex  anklex Þ2 þ ðkneey  ankley Þ2
Kinematics Analysis and Mechatronic Simulation 205

Fig. 2 Motion capture system and initial captured position

2.2 Forward Kinematics

The forward kinematics allows calculating the horizontal (xf ) and vertical
position (yf ) of the ankle using the geometrical relations from each link depicted
in Fig. 2 and in equation (2). Another solution can be found using the trans-
formation matrix through the Denavit–Hartenberg notation (DH) [19], as it
relates positions and rotations for each link and joint. The main difference
between these approaches is the processing time needed to find the solution.
Complex systems take more time to be solved by DH due to the processing of the
zeros within the transformation matrix, while the geometric solution directly finds
the corresponding position [20].

xf ¼ l3  sinh þ l4  sinðh  fÞ
ð2Þ
yf ¼ l3  cosh þ l4  cosðh  fÞ

In order to find a solution using the DH method, the lengths and the rotations of
each segment with respect to a reference coordinate system are required. These
parameters can be seen in Fig. 2 and are presented in Table 2. From them can be
inferred the non existent abduction/adduction movements as the rotation (a) and
lengths (a) in the z axis remains zero for both links. Finally, it is worth noting that
by modifying the DH parameters, the number of DOF can be increased or
decreased as the model requires it, while if using the geometric solution any
additional degrees would require the recalculation of the equations.
The transformation matrix j1 Tj expresses the complete positioning of a specific
link in the space (translation and orientation) using the parameters in Table 2,
obtaining matrix T presented in Eq. (3), where j is the number of the segment,
being 0 the first link, 1 the second, and so forth. From T the positions and rotations
206 A. J. Uribe et al.

Table 2 DH parameters Segment h a d a

based on Fig. 2
Thigh h 0 l3 0
Shank f 0 l4 0

of each link can be known. To find the solution, the number of matrix operations
required is proportional to the number of links and DOF of the mechanism. For the
model presented in Fig. 2, two matrix multiplications per leg are necessary for
calculating the knee and ankle rotations and positions.
2 3
coshj cosaj  sinhj sinaj  sinhj aj  coshj
6 sinhj cosaj  coshj sinaj  coshj aj  sinhj 7
j1
Tj ¼ 6
4 0
7;
5 ð3Þ
sinaj cosaj dj
0 0 0 1

2.3 Inverse Kinematics

The inverse kinematics calculates angles from known positions through symbolic and
numeric methods. With this method, the closed form solutions offered by symbolic
approaches have the disadvantage, that the motion law is expressed as an invariant
polynomial expression that does not consider the time and space dependent con-
straints. Unlike the closed form solutions, numerical methods may be combined with
flexible motion strategies by introducing time dependent weighting factors. However,
the use of numerical methods can only success if the most efficient algorithms are
applied [20–22]. The use of numerical methods is facilitated by the incremental
equation using the vector of adjustment position (Dx) which is equal to the product of
the Jacobian (J) and the vector of joint adjustment (Dh) presented as Dx ¼ J  Dh.
This linear expression is only valid under the presumption that the joint adjustment is
small given that bigger values of (Dh), (Dx) no longer matches physical reality.

2.3.1 Gauss Elimination Method

When the number of external coordinates in Dx is lower than the number of joint
coordinates in redundant member configurations, the inverse kinematics problem
must be treated as an inhomogeneous linear equation system which can be
examined most efficiently by multiplying the rectangular Jacobian (J) with its
transpose (J T ), and performing a Gaussian elimination on the resulting square
matrix. Solutions only exist when the elimination process produces a diagonal zero
matrix, and when this condition is satisfied the Jacobian is classified as a linearly
independent full rank matrix [22]. For a linearly dependent system, solutions exist
Kinematics Analysis and Mechatronic Simulation 207

when the rank deficiency of the Jacobian is enlarged by a load vector having the
same rank deficiency, otherwise, there is no solution and the desired position is out
of range (Cronecker–Capelli theorem) [23].
The elimination method uses the J matrix components as shown in (4), using
d d
the known differential relations between angles and calculated data (ddhx ; ddxf ; dhy ; dyf ).
The variables are eliminated through each iteration until the solution is found
performing ð4  n3col  9  n2col  7  ncol Þ=6 arithmetic operations, where ncol is the
number of columns of the J matrix.
2 3
dx dx
dh df
J ¼ 4d dy
5;
y
dh df
dx
dh  Dy þ ddxf  Dx
f¼ ; ð4Þ
dx d d
df  dhy þ dyf  ddhx
Dx þ ddxf  Df
h¼ dx
;
dh

2.3.2 Greville Iterative Method

For overcoming the ambiguity of the Jacobian, a special limiting condition known
as the quadratic criterion or least squares principle is introduced thus, yielding a
simple no iterative algorithm that has been classified as the minimum norm
solution (G) of the generalized inverse [21]. This is presented in (5).
G ¼ J T ðJ  J T Þ1 ;
ð5Þ
h ¼ J T ðJ  J T Þ1  Dx ;
Although this solution can easily be computed in real time, its practical
application is limited to kinematic chains producing a full rank matrix. In order to
bridge the singularities of the generalized inverse and to find solutions for per-
manent rank deficiency systems, Moore and Penrose defined the principle of the
pseudoinverse [24], which represents the unique solution for: J  J T  J ¼
J; J T  J  J T ¼ J T ; ðJ  J T Þþ ¼ J  J T ; ðJ T  JÞþ ¼ J T  J, yielding to a complete
pseudoinverse matrix. Its application to matrices with more rows than columns is
meaningless because such cases do not represent any solvable kinematic problem.
The pseudoinverse can be calculated using the Greville method which allows
solving the inverse matrix without needing its rank. The advantage of not needing
the rank of a matrix is to avoid calculating the number of linearly independent
rows or columns of a matrix, along with the dimension of its image. The Greville
algorithm is presented in Fig. 3, where A1 is formed by the first column of J, K is
the number of iterations (related to the number of DOF), and finally Aþ K is the
pseudoinverse matrix, as explained in [21].
208 A. J. Uribe et al.

Begin

A1 = 0 Ck = 0

A1+ = (A1t A1)-1 A1t A1 = 0 Bk = (Ckt Ck)-1 Ckt Bk = (1+Dkt Dk)-1 Dkt Ak-1+

K=2
Ak+ = Ak-1+ - Dk Bk
Bk

Dk = Ak-1+ Ak

K=n
Ck = Ak - Ak-1 Dk
K = K+1 A+ = An+

End

Fig. 3 Greville iterative algorithm adapated from [21]

The solution is calculated after executing 3  ncol  nrow þ 4  ncol arithmetic

operations through two iterations according to the number of the DOF of the model,
where nrow is the number of rows and ncol the number of columns of the system.

3 Motion Planning

For trajectory generation, the most relevant points during walk are chosen by
analyzing the horizontal and vertical behavior of the ankle, as depicted in Fig. 4
where the highest and lowest points within the time for walking one step are
known. Ankle rotations when leaving and contacting back the surface are also
considered as they influence the motion on the opposite ankle. With the gathered
information, a suitable path for each lower member can be reconstructed using
various polynomial techniques [25].
The purpose of this motion planning analysis is to select an interpolation
technique that best fits the captured gait movement. For solving the planning
problem, the spline interpolation is chosen due to its piecewise polynomial
properties for satisfying the condition of passing through all the control points.
This approach considers the parameters presented in Fig. 4, where p is the step
number and T the time over the cycle, taken from anatomic relations or measures
from a person.
Kinematics Analysis and Mechatronic Simulation 209

Fig. 4 Gait points of interest. Left: Ankle motion during gait. Right: Ankle motion when leaving
the ground

3.1 Cubic Spline Interpolation

Third order or cubic splines respond accurately when following a sequence of points
due to their interpolation derivative and integrative characteristics which determine
the behavior at both ends of the curves and their continuity [26]. For calculating the
spline equations, a minimum of three points are required in order to obtain two sets of
third order polynomials to connect them. Their general form is presented in Table 3
where it can be seen how each set has its own parameters which will vary depending
on the chosen continuity conditions. For each cubic spline, three points during three
moments in time (t1  t  t2 ) of the knee and ankle trajectories are needed.
For finding the equation parameters a1 , b1 , c1 and d1 , smooth continuity conditions
defined by f ðtÞ} ¼ f ðt2 Þ} for x ¼ t1 and f ðtÞ} ¼ t2 at t ¼ t2 along with a zero slop
characteristic at both ends (first and second derivatives equal to zero), were chosen to
calculate the spline trajectory. These conditions were applied to the polynomials
presented in Table 3 for finding the spline parameters. For testing the spline per-
formance, the interpolation curve was calculated using three and five points taken
from the motion captured data allowing to compare how accurate the obtained curves
are in relation to the motion captured ones. In Fig. 5 the obtained spline curves show a
quality dependance of the gait motion fitting on the amount of control points. It can be
observed that interpolation with five points resulted in better and more alike repro-
duction of the target path proving to be better than choosing just three gait points.
It has to be noted that the interpolation is not constrained to any maximum
number of points as at least three are necessary to reproduce the trajectory with the
minimum possible error. This number may vary depending on various gaits from
different persons. The ankle motion of three persons with different heights is
presented in Fig. 6, illustrating how gait differs depending on the subject char-
acteristics such as anthropometry or health conditions. Figure 6 also shows the
effects of the counter ankle rotation when leaving the ground in all three samples
as the ankle analyzed gets closer to its single support phase.
210 A. J. Uribe et al.

Table 3 Spline polynomials

Spline 1 (t1  t  t2 ) Spline 2 (t2  t  t3 )
f ðtÞ ¼ a1 ðt  t1 Þ3 þ b1 ðt  t1 Þ2 þ c1 ðt  t1 Þ þ d1 f ðtÞ ¼ a2 ðt  t2 Þ3 þ b2 ðt  t2 Þ2 þ c2 ðt  t2 Þ þ d2
0 2
f ðtÞ ¼ 3a1 ðt  t1 Þ þ 2b1 ðt  t1 Þ þ c1 f ðtÞ0 ¼ 3a2 ðt  t2 Þ2 þ 2b2 ðt  t2 Þ þ c2
0
f ðtÞ ¼ 6a1 ðt  t1 Þ þ 2b1 f ðtÞ0 ¼ 6a1 ðt  t1 Þ þ 2b1

three main gait points

two additional gait points
three points based spline
Linear interpolation
five points based spline

Fig. 5 Cubic spline interpolation with 3 and 5 points

Fig. 6 Motion captured gaits

3.2 Cubic Hermite Splie Interpolation

This is a third degree order spline in which the polynomial is expressed using the
Hermite form having two control points and two control tangents. The equation
form and parameters are the same as those in Table 3 for the cubic spline.
Kinematics Analysis and Mechatronic Simulation 211

Fig. 7 Hermite interpolation curves

The main difference is that first derivatives are continuous at the data points, while
the second derivatives are not, as presented in (6) [27].

y0j ðxj Þ ¼ yj ; ;
ð6Þ
y0j1 ðxj Þ ¼ yjþ1 ; ;

After solving the equations with the Hermite conditions, the calculated poly-
nomials for vertical and horizontal motion of the ankle are shown in Fig. 7, where
can be seen similar curve shape in comparison of the trajectory presented in Fig. 6.

4 Comparative Study

The kinematic analysis and the trajectory generation were implemented consid-
ering the forward and inverse kinematics methods, along the cubic Hermite
polynomials respectively. This section presents a comparative study of the pre-
vious mentioned approaches, to analyze how they perform and behave when
planning motion in order to generate suitable parameterized information for later
use as input data in the mechatronic implemented model.

4.1 Forward Kinematics Implementation

For solving the forward kinematics, the absolute kinematic analysis through
geometric relations was implemented. It was chosen over the relative analysis
(DH method) due to its better performance with complex systems given the fact
that, when numerous matrix operations are performed, the amount of zeros within
the calculations produces higher computational time process for finding the system
solution.
212 A. J. Uribe et al.

Fig. 8 Forward kinematics developed program

The parametrization of the system is done by considering the lengths of the

thigh and the shank along with their rotation, which is chosen by the user for
solving the forward kinematics. The system’s legs are analyzed individually and
each workspace is constrained by anatomic flexion/extension rotations. A program
was developed for visualizing qualitative and quantitative results of the forward
problem solution. It allows the user to choose a target angle for both shank and
thighs on each leg, calculate the resulting targeted positions and analyze them
through a 3D representation and as transformation matrices. The developed gra-
phic user interface is shown in Fig. 8, where the circled 1 identifies the input boxes
for the h and the f angles. The circled 2 contains a 3D canvas where the calculated
positions from the kinematic analysis are plotted in the three axes, x (frontal), y
(sagittal) and z (transverse, where the ground is positioned). Finally the circled 3
shows the data from the transformation matrices for the thigh and shank.

4.2 Inverse Kinematics Approaches

The Gauss and Greville methods were successfully implemented for calculating
the J 1 and consequently the h and f angles from known positions. The resulting
flexion/extension rotations of the shank and thigh are presented in Fig. 9, where no
Kinematics Analysis and Mechatronic Simulation 213

Fig. 9 Inverse kinematics angles. Left: Calculated h and f. Right: f comparison between
experimental and theoretical

differences or error were found between the obtained h and f angles, due to
simplicity of the two DOF problem. To select which method suits best the analysis,
it has to be kept in mind that the system is not constrained to just two DOF, and
more can be added so further complexity can be achieved when considering full
human lower member motion. With the latter condition considered, the Greville’s
method was chosen due to its scalability and less computing power required, given
the number of needed arithmetic operations.
In order to define how accurate was the trajectory obtained through the Greville
method and to define its error when compared to the motion captured walk, a
statistical forecasting model based on estimation period such as the root mean
squared error (RMSE) is used. The main characteristic of the RMSE is that the
measurements and calculations are done in the same units as the input data.
Another feature is its sensitiveness, which is greater than other methods because of
the squaring process inside the equation. By analyzing the calculated curves with
the motion captured data, a h of 0.1435rad and a f of 0.224rad RMSE angles were
obtained. Fig. 9 shows the diferences between the calculated and motion captured
f angles, allowing to visualize the error. The calculated angle through inverse
kinematics fails to reproduce accurately the joint rotations, even though Greville
avoids the singularities, the resulting angle trajectories were not suitable for gait
reproduction.

4.3 Spline-Based Trajectories

Considering the positions of the anthropometric and measured data, each of the
spline parameters were calculated using the Hermite form producing the results
presented in Table 4 for the anthropometric statistic data and Table 5 for the
214 A. J. Uribe et al.

Table 4 Anthropomorphic-based y and x spline coefficients

Curve ay by cy dy ax bx cx dx
1 -2.36 1.42 0 0.075 -0.38 0.68 0 0
2 -39.67 9.29 0.27 0.10 -63.44 20.68 0.2 0.018
3 2.79 -1.94 0 0.22 -6.41 2.58 1.73 0.33
4 -0.001 0.013 -0.021 0.085 0.00 -0.024 0.036 0.057

Table 5 Measured-based y and x spline coefficients

Curve ay by cy dy ax bx cx dx
1 -0.031 0.105 0 0.081 -0.1 0.287 0 0.081
2 -2.76 1.51 0.1 0.12 0.23 0.77 0.25 0.102
3 11.75 -4.9 0 0.225 -36.6 14.52 0.97 0.33
4 0.004 0.086 -0.08 0.1 -0.57 -0.51 0.81 0.94

Fig. 10 Ankle Hermite spline trajectories during one step. Left: Vertical motion. Right:
Horizontal motion

measures taken from a subject. Each of the previous mentioned tables, present the
calculated cubic Hermite parameters a, b, c and d for both vertical (y) and hori-
zontal (x) motion, as presented in Sect. 3 and Fig. 5. With the Hermite form, only
four curves were needed, in order to obtain suitable alike trajectories based on
the most significant vertical and horizontal positions during normal gait. It can also
be noted from each cubic spline both vertical and horizontal parameters differ one
from the other, this suggests that the trajectories are not similar with each data set,
leading to a notorious difference between the statistic data and the measured,
experimental data.
To further analyze how well the spline interpolation follows the gait data, the
horizontal and vertical trajectories of the ankle calculated with the parameters
from Table 4 and Table 5, were compared during a single step of the gait cycle.
Figure 10 presents the results of plotting the Hermite splines from the motion
Kinematics Analysis and Mechatronic Simulation 215

captured data, anthropometric statistic data and measures taken from a person.
Qualitative can be seen, that the plotted curves are similar and that the main
difference is the lowest point of the ankle which is lower with the statistic data than
the real measurements. To quantitative analyze these differences, the RMSE was
applied between the motion captured data and the anthropometric parameters
obtaining a 0:072m vertical and a 0:3285m horizontal error. The error between the
measured person based spline and the anthropometric data was 0:0789m vertically
and 0:36m horizontally. Finally, between the measured person based spline and the
captured data the error was 0:0077m vertically and 0:0351m horizontally.

5 Mechatronics Model

For developing the mechatronics model, bones and muscular counterparts were
chosen to give the necessary support and actuator elements for performing normal
gait given known thigh and shank angles. The bioinspired chosen elements for
support are cylinders configured with inertia, length and volume similar to the leg
characteristics, while for the actuation system four direct current servomotors were
chosen for executing the rotations on each limb.
For the model implementation, the SimMechanicsTM Toolbox of Matlabr and
documentation available in Mathworks, offer tools for modeling 3D mechanical
systems within the Simulink environment. Instead of deriving and programming
equations, the multibody simulation tool allows building complete models com-
posed of bodies, joints, constraints, and force elements that reflect the structure of
the system generating automatically a 3D animation that gives a visualization of
the system dynamics.
To properly implement the mechatronics model, the mechanical behavior of the
lower member is define as similar to that of a double inverted pendulum. Two
double pendulums are configured to interact through a shared connection (waist) in
order to behave as regular pendulum during the balancing phase of gait and as
inverted pendulum when in contact to the ground with every step.

5.1 Implemented Blocks

Considering that the model integrates various subsystems such as, the thigh, shank
and ground interactions the Toolbox offers Body, Joint, Constraint, Actuator,
Sensor and Force blocks for configuring and creating the needed SimMechanicsTM
machine for reproducing suitable human gait.
For controlling the flexion/extension servo position, a SimMechanics
SimMechanicsTM model was implemented where the input angle for the joint was
converted to torque by the body’s actuator component in order to rotate the
216 A. J. Uribe et al.

Fig. 11 SimmulinkTM
pendulum model between hip
and thigh using a PI
controller with a known input
angle

articulation and provide feedback through the measures taken with a sensor.
This allows simulating an inverse pendulum-like behavior for each leg, where
an integrative-derivative (PI) position control is implemented, as presented in
Fig. 11. The very same principle was applied for simulating a horizontally regular
ground interaction where a fixed value and the sensed velocities at the tip of the leg
were established as reference values for model interaction.
The model for simulating the lower member results from integrating the
inverse pendulum and ground interaction blocks accordingly to each leg. The
system can either be solved through known sequencies of angles or through
those calculated from the inverse kinematics method. A total of six blocks are
needed for simulating both legs walking; four for the shank and thighs, and two
more for the ground interaction with each ankle, all of them based on the
model presented in Fig. 11. The SimMechanicsTM model is presented in
Fig. 12, where the components are group under the four gray areas. The first
area (from left to right) has the input component which reads the angles from a
database. The second and third area have the thigh and shank control blocks
respectively. Finally, the fourth area has the contact control between the ground
and the mechanism. The thigh and shank body blocks for interconnecting the
control models were configured accordingly to their length, joint position,
inertia and the location of the center of mass. To measure and monitor posi-
tions, rotations, velocities and torque, scopes were placed at various points of
the model.
Kinematics Analysis and Mechatronic Simulation 217

Fig. 12 SimMechanicsTM Mechatronic-based model composed of four main areas highlighted on

grey

5.2 Simulation Results

After simulating the SimMechanicsTM Mechatronic-based model with gait infor-

mation from motion captured data, the ankle vertical and horizontal trajectories were
taken from the shank scope situated at the left shank output of the control block as
presented on Fig. 12, obtaining the motion paths shown in Fig. 13. These simulated
trajectories were compared to the ones obtained from the motion capture system,
presented in Fig. 14. From the analysis of both Fig. 13 , 14, it can be seen that each
set of horizontal and vertical curves have a qualitative similarity, as their highest and
lowest points have the same values, however, there are two main differences with the
obtained results in the vertical axis. First, the smoothness of the curve when reaching
and leaving the lowest point due to the natural impact absorption of the lower
member biomechanics, is not present in the SimMechanicsTM results as there is no
cushion in the system thus, the ankle maintains its lower point constant. The second
difference is the effects on the mid-high point when the counter ankle rotates to leave
the ground. For the motion captured data, this value does not exceeds 12cm, while in
the SimMechanicsTM results reaches 13:8cm.
Taking advantage of the SimMechanicsTM capability of showing a virtual model
performing the simulation, Fig. 15 shows the simulated gait from the implemented
block system, where it can be seen how the swing and double support phases are
reproduced during a walking period. From it, the ankle’s highest and lowest points
can also be identified.
218 A. J. Uribe et al.

Fig. 13 SimMechanicsTM ankle positions obtained after simulation. Left: Vertical motion. Right:
Horizontal motion

Fig. 14 Ankle trajectories obtained through motion capture. Left: Vertical motion. Right:
Horizontal

Fig. 15 Mechatronic model walking

Kinematics Analysis and Mechatronic Simulation 219

6 Conclusions

The forwards kinematics successfully allowed the calculation of the thigh and
shank position using known rotation angles. The forward solution does not present
any singularities due to the natural constraint of the patella in the shank rotations
but, without this limitation two different positions may exist for a rotation set of
angles. It is in this case, when singularities are present, that the inverse kinematics
is used. The Greville method allows overcoming the singularities while increasing
the complexity of the model when adding more DOF. The results obtained from
the inverse analysis suggest that, in order to achieve more accurate motion paths,
the kinematics analysis needs to be executed along with a control strategy (pro-
portional, derivative and/or integral).
For the trajectory planning, the cubic spline in its simple form failed to follow
the targeted trajectory, while its Hermite variation successfully passed over the
control points as well as the ankle’s path during a step segment in the gait cycle.
The trajectory error can increase or decrease depending on which and how many
control points are chosen. Common Anthropometric statistic calculations failed to
offer accurate trajectories when compared to motion captured data, due to dif-
ferences in the measurements. This is why it is strongly recommended to carefully
choose the data directly from the chosen subject.
These two previous approaches complemented each other in the motion gen-
eration, as not only the angles, but also, the positions of the knee and waist can be
calculated during the gait cycle Finally, by using the kinematics and motion
planning together, parameterized information for reproducing gait on the mecha-
tronics SimMechanicsTM model can be calculated thus, obtaining suitable and
likely human gait cycle. The ground and position control parameters can be
changed in order to test the behavior of the model in different types of terrain such
as sloped or stepped surfaces and lower member motion over these.
The SimMechanicsTM simulated trajectories may be suitable for reproducing on
a biped walking robot, instead of applying them on a human motion assisted
device, as discomfort would be present due to the difference between the motion
captured and the simulated paths trajectories. These trajectories have specific knee
and ankle simulated positions which were not equal to their natural measured
positions, resulting in a poor motion coupling between the device and user.
Future work is focused on the improvement of the control strategy by studying
and analyzing thoroughly the dynamics involved when executing gait on various
scenarios, as well as the effects of the mechanical design of a suitable device for
assisting walking motion. If design to assist motion, the simulated trajectories
should match motion captured data and the error position should not exceed the
natural position of each articulation.

Acknowledgments The authors would like acknowledge the support of FAPESP (Fundação de
Amparo à Pesquisa do Estado de São Paulo), under process 2009/05396-4, the Instrumentation
Laboratory for Biomechanics (LIB) and the Automation and Robotics Laboratory of the
UNICAMP.
220 A. J. Uribe et al.

References

1. Kajita, S., Hirukawa, H., Harada, K., Yokoi, K.: Introduction à la commande des robots
humanoïdes. Springer, Paris (2009)
2. Grabowski, A., Herr, H.: Leg exoskeleton reduces the metabolic cost of human
hopping. J. Appl. Physiol. 107, 670–678 (2009)
3. Banala, S., Agrawal, S.: Gait rehabilitation with an active leg orthosis. In: Proceedings
of IDETC/CIE, ASME, International Design Engineering Technical Conferences and
Computers and Information in Engineering Conference (2005)
4. Sawicki, G., Gordon, K., Ferris, D.: Powered lower limborthoses: Applications in motor
adaptation and rehabilitation.In: Proceedings of the IEEE 9th International Conference on
Rehabilitation Robotics pp. 206–211 (2005)
5. Huang, Q., Yokoi, K., Kajita, S., Kaneko, K., Arai, H., Koyachi, N., Tanie, K.: Planning
walking patterns for a biped robot. IEEE Trans. Robotics Autom. 17(3), 280–289 (2001)
6. Bruderlin, A., Calvert, T.W.: Goal-directed, dynamic animation of human walking.
SIGGRAPH. In: Proceedings of the 16th annual conference on Computer graphics and
interactive techniques. New York, NY, USA: ACM, pp. 233–242 (1989)
7. Collins, S., Ruina, A., Tedrake, R., Wisse, M.: Efficient bipedal robots based on passive-
dynamic walkers. Sci. 30, 1082–1085 (2005)
8. Vukobratovic, M., Borovac, B.: Note on the article ‘‘zero-moment point - thirty five years of
its life.’’ Int. J. Humanoid Robotics 2(2), 225–227 (2005)
9. Yan, J.L., Huang, H.P.: A fast and smooth walking pattern generator of biped robot using
jacobian inverse kinematics. In: Advanced Robotics and Its Social Impacts, ARSO. IEEE
Workshop, pp. 1–6 (2007)
10. Thant, A., Kang, Y.: Application of cubic spline interpolation to walking patterns of biped
robot. World Acad. Sci. Eng. Technol. 50, 27–34 (2009)
11. du Plessis, L.J., Snyman, J.A.: ‘‘Trajectory-planning through interpolation by overlapping
cubic arcs and cubic splines’’. I. J. Numer. Methods Eng. 57(11), 1615–1641 (2003)
12. Chung, S.-K., Hahn, J.K.: Animation of human walking in virtual environments.
In: Proceedings of the Computer Animation IEEE Computer Society, Washington DC,
USA. p. 4.(1999)
13. Ruthenberg, B., Wasylewski, N., Beard, J.: An experimental device for investigating the
force and power requirements of a powered gait orthosis. J. Rehab. Res. Dev. 4, 203–213
(1997)
14. Kasaoka, K., Sankai, Y.: Predictive control estimating operator’s intention for stepping-up
motion by exoskeleton type power assist system hal. International Conference on Intelligent
Robots and Systems. In: Proceedings IEEE/RSJ, pp. 1578–1583 (2001)
15. Peter, G.J., Malcom, G.H.: A knee and ankle flexing hybrid orthosis for paraplegic
ambulation. Med. Eng. Phys. 25, pp. 539–545 (2003)
16. Agrawal, A., Agrawal, S.K.: Design of gravity balancing leg orthosis using non-zero free
length springs. Mech. Mach. Theory 40, 693–709 (2005)
17. Inman, V., Ralston, H., Todd, F.: Human Walking. Williams and Wilkins, Baltimore (1981)
18. Hamill, J., Knutzen, K.: Bases Biomecânicas do Movimento Humano. Manole, Tamboré SP
(2008)
19. Denavit, J., Hartenberg, R.: Kinematic Synthesis of Linkages. McGraw Hill, New York
(1964)
20. Barrientos, A.: Fundamentos de Robótica. McGraw Hill, Madrid (2007)
21. Gorla, B., Renaud, M.: Modles des Robots Manipulateurs: Application leur Commande, C.
ditions, Ed. Automatisation & Production (1984)
22. de Sa, C., Rosario, J.: Implementation of numerical algorithms for the resolution of the
kinematic inverse problem in robot manipulators. Nonlinear Dyn., Chaos, Control, Appl.
Eng. Sci. 1, 317–320 (1997)
23. Prasolov, V.: Problems and theorems in linear algebra. AMS Bookstore, USA (1994)
Kinematics Analysis and Mechatronic Simulation 221

24. Albert, A.: Regression and the Moore-Penrose pseudoinverse. Academic Press, London
(1972)
25. Press, W.H., Teukolsky, S.A., Vetterling, W.T., Flannery, B.P.: Numerical Recipes 3rd edn:
The Art of Scientific Computing, 3rd edn. Cambridge University Press. http://www.amazon.
com/exec/obidos/redirect?tag=citeulike07- 20&path=ASIN/0521880688 (2007)
26. Smith, A.R.: Spline tutorial notes, Lucasfilm Ltd, Tech. Rep. (1983)
27. Engeln-Mllges, G., Uhlig, F.: Numerical algorithms with C. Springer, New York (1996)