Você está na página 1de 8

Journal of Food Engineering 103 (2011) 265272

Contents lists available at ScienceDirect

Journal of Food Engineering


journal homepage: www.elsevier.com/locate/jfoodeng

A nite element model for mechanical deformation


of single tomato suspension cells
E. Dintwa a, P. Jancsk a,b, H.K. Mebatsion a, B. Verlinden a, P. Verboven a, C.X. Wang b, C.R. Thomas b, E.
Tijskens a, H. Ramon a, B. Nicola a,
a
b

Flanders Centre of Postharvest Technology/BIOSYST-MeBioS, Katholieke Universiteit Leuven, Kasteelpark Arenberg 30, B-3001 Leuven, Belgium
School of Chemical Engineering, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK

a r t i c l e

i n f o

Article history:
Received 4 May 2010
Received in revised form 20 October 2010
Accepted 23 October 2010
Available online 2 November 2010
Keywords:
Finite element
Model
Single cell compression
Tomato cell
Micromanipulation
Texture
Fruit

a b s t r a c t
A nite element model was developed to simulate compression experiments on single tomato cells from
suspension cultures. The cell was modelled as a thin-walled liquid-lled sphere with a permeable wall
allowing ow of uid out in response to internal turgor increases due to the compression. The permeability of the cell wall/plasma lemma was considered to be constant throughout compression. The contact
between cell and compression probe was modelled using a soft contact boundary condition. The cytoplast
was represented as an internal pressure acting on the plasma lemma and cell wall. Assuming linear
elastic constitutive behaviour for the cell wall, and using previously determined cell wall material parameters, the model was found to be remarkably capable of reproducing the forcedeformation behaviour
of a single cell in compression, as well as its deformed shape, even for large strains. The model might be
used as a building block to construct more comprehensive tissue deformation models.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Fruit and vegetables are an important component of the human
diet and consumers usually expect such produce to be of premium
quality. Texture is a major quality attribute and inuences
consumer acceptance, shelf-life, resistance, and transportability
(Seymour et al., 2002). The texture of plant produce and its susceptibility to damage are determined by its mechanical properties.
There is an enormous body of the literature relating to this but
until recently a continuum approach has usually been adopted,
in which it is assumed that the properties of the material are independent of the spatial scale of observation (Ghysels et al., 2009). As
such, the classical techniques for determination of the basic
mechanical properties of these materials have been essentially
the same as those applied for standard engineering materials. Such
approaches, while they admittedly have been applied with some
success, have always been weak in legitimacy. The major concern
is that, unlike traditional engineering materials, plant tissue has a
highly complex hierarchical structure. For example, a fruit such as
a tomato consists of a complex conglomerate of different tissues
(e.g. skin, cortex, core, seeds, etc.), and each tissue has many microscopic constituents such as cells, the middle lamella and interstitial
spaces. The macroscopic mechanical properties of the fruit depend
Corresponding author.
E-mail address: bart.nicolai@biw.kuleuven.be (B. Nicola).
0260-8774/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2010.10.023

on the properties of these constituents, their arrangement and


their interactions. Mathematical continuum models ignore this
structure. A new view among practitioners is that the most realistic
hope for deriving generic, robust and durable constitutive relations
for the mechanical behaviour of this type of materials lies with a
multi-scale approach to the problem (Ghysels et al., 2009). The
contention is that only by a thorough understanding of the physics
at all the spatial scales within an individual fruit can we be able to
derive, in a quantitative way, the critical inuences by the different
properties at various levels. The ultimate objective is to derive constitutive relations that fully determine the mechanical behaviour
of the whole fruit by their microscopic properties. This will require
studies of tissues and cells at microscopic scales.
Studies of the mechanical properties of fruits and vegetables at
microscopic scales and beyond have understandably been
constrained by the lack of technology to conduct mechanical
experiments at such size scales. Recent developments in micromanipulation techniques (Mashmoushy et al., 1998; Shiu et al., 1999;
Blewett et al., 2000; Thomas et al., 2000; Wang et al., 2005),
however, have allowed accurate measurements to be performed
at micro-scales, hence opening up new possibilities in this area.
As arguably the most important primary unit of vegetative tissue,
the plant cell represents an obvious starting point for the study
of the mechanical behaviour of fruit tissue that takes into account
the multi-scale structure of the material. Working on single cells,
whether isolated from tissue or taken from suspension cultures,

266

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

is admittedly a reductionist approach, but it should provide a


sound basis for the subsequent introduction of complexity. One
possibility to study the cellular mechanics of single cells is the
use of micromanipulation to perform compression tests between
two plates on single cells (Mashmoushy et al., 1998; Shiu et al.,
1999; Thomas et al., 2000; Wang et al., 2005). Data (applied forces,
displacements, bursting forces) acquired from these single cell
compression experiments can be used with cell models to determine the intrinsic properties of the cell wall. The micromanipulation technique has been applied in measurements involving a
variety of types of single cells such as mammalian cells (Zhang
et al., 1991), yeast cells (Mashmoushy et al., 1998; Smith et al.,
2000a,b,c; Stenson et al., 2009), bacterial cells (Shiu et al., 1999)
as well as tomato fruit and suspension-cultured cells (Blewett
et al., 2000; Wang et al., 2004, 2006).
Historically, early attempts to model the mechanical behaviour
of vegetative tissue have predominantly been in the form of simple
mechanistic theoretical models geared at simulating the behaviour
of the tissue to external loads (Nilsson et al., 1958; Pitt, 1982; Pitt
and Chen, 1983; McLaughlin and Pitt, 1984; Gao and Pitt, 1991;
Gates et al., 1986). Alternative approaches to modelling cellular tissue involved the use of the nite element method (FEM) (Akyurt
et al., 1972; Pitt and Davis, 1984). In general these models were
lacking in quantitative accuracy or even validation. However, qualitatively, they were highly useful in studying various features of the
tissue. Driven by technological advances in measuring mechanical
behaviour of single cells, more complex models have appeared in
the literature. Wu and Pitts (1999) developed a FE model of a single
apple parenchyma cell compressed between two parallel plates.
They used intricate serial micro-sectioning and image analysis to
acquire a realistically shaped 3-D FE model of the cell from a series
of parallel slices of sample tissue. The cell was modelled as a thin
shell with surface pressure inside to simulate turgor pressure. Fluid
movement out of the cell and volume changes during compression
were not accounted for. Smith et al. (1998) developed a comprehensive FE model of the compression of an inated yeast cell, with an
explicit objective to use it to assess the uniqueness or otherwise
of the various crucial parameters involved in the model. The cell
was modelled as a thin-walled liquid-lled sphere with permeable
walls that allow expulsion of uid out in response to internal turgor
increases. Volume loss during compression was accounted for. The
cell wall was modelled as an isotropic elasticplastic material.
Using data derived from micromanipulation experiments, Smith
et al. (2000a,b,c) applied the single cell compression model of Smith
et al. (1998) to extract a variety of mechanical properties of the
yeast cell walls. Blewett et al. (2000) and Wang et al. (2006) performed single cell compression micromanipulation measurements
on, respectively, suspension-cultured tomato cells and tomato fruit
cells. In the absence of a FE model of these types of cells, they used
an analytical model of the cell derived from the theory of Yang and
Feng (1970) as adapted by Lardner and Pujara (1978, 1980) to extract the mechanical properties of the cells (Wang et al., 2004,
2006). The analytical model, which assumes a permeable cell with
linear elastic constitutive behaviour, yielded very good ts to
experimental force deformation data after parameter optimisation
for fractional deformations up to 20%. The curve ts were achieved
with two free parameters: the initial stretch ratio and the Youngs
modulus (i.e. both were estimated at the same time, choosing the
best least squares t to the data). Also, in Wang et al. (2004) the cell
wall hydraulic conductivity was assumed to be small enough that
the effects of water loss could be neglected for small deformations,
whilst in Wang et al. (2006) it was claimed that the compressions
were so rapid that there was insufcient time for signicant water
loss to occur.
An inherent drawback of analytical models such as the one used by
Wang et al. (2004, 2006) is the difculty of including more complex

features such as non-linear constitutive behaviour of the cell and


accounting for large strain deformation of the cell wall. Also, it is difcult to use an analytical model for a single cell to predict the mechanical behaviour of plant tissues composed of many individual cells. The
latter is of particular importance, as a better understanding of the
forcedeformation behaviour of fruit and vegetable tissue might
guide to better postharvest handling practices or inspire the development of novel techniques to measure their texture properties.
The aim of this work was to develop a nite element model to
simulate the compression of a single suspension-cultured tomato
cell, using data from Wang et al. (2004). This model could serve
as a basic building block for more complex models for tissue deformation under mechanical loading.
2. Materials and methods
2.1. Cell cultures, physical measurements and compression data
Experimental compression data described by Blewett et al.
(2000) and Wang et al. (2004) were used. In summary, single tomato (Lycopersicon esculentum L.vf36) cells were obtained from a suspension culture derived from a root radicle callus. The suspension
culture resulted in nearly synchronous growth, and hence the cells
produced were of similar age. The cells were suspended in 0.03 M
mannitol at pH 5, the same osmolality and pH as the medium at
harvest. During compression testing, the cells were held at
27 1 C. The micromanipulation experiments involved the individual cells being compressed between the at end of an optic bre
probe and a glass surface whilst measuring the forcedeformation
curve and video-imaging the cell compression. The probe speed
was 23 lm s1. Data on the probe displacement and the compression force from one such experiment were analysed here.
The initial thickness of the cell wall (i.e. the thickness of cell wall
when the cell is at zero turgor) was determined as the mean cell
wall thickness of several cells from the culture using freeze-fracture scanning electron microscopy. The mean initial turgor pressure
of the cells was determined earlier by Wang et al. (2004) from direct osmotic pressure measurements (the difference between the
measured external osmolality and the internal osmolality measured by freezing point depression). The initial stretch ratio of the
cell before compression was not measured but the diameter of
the cell before compression was known.
2.2. Image acquisition, analysis and digitization
In order to allow comparison of the shape of the deformed cell
according to the FE model to that of the real cell, the cell deformation
during the compression experiment was obtained from the video.
Still images from several stages of the compression were then further processed using MATLAB (Mathworks, Natick, MA, USA) image
analysis programs, digitized versions of the contours of the
deformed cell were obtained. Fig. 1 gives a summary of the procedure used to obtain these digitized images. First, a series of still
images was obtained from the movie of the experiment (Fig. 1a),
the boundary of the cell was then carefully marked manually by
means of a mouse pointer on the images (Fig. 1b). The cell image
was then further enhanced to allow easy boundary detection by
the program (Fig. 1c). The digitized deformed cell contour is as
shown in Fig. 1d.
2.3. Finite element model
2.3.1. Model denition
Fig. 2 gives a schematic description of the cell compression process. An uninated (zero turgor pressure) cell of a specied outer

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

267

Fig. 1. Steps in the acquisition of digitized contours of the deformed cells. First, a series of still images was obtained from the movie of the experiment (a), the boundary of
each cell was then carefully marked on the images (b). The cell image was then further enhanced to allow easy boundary detection by the image analysis program (c). The
nal digitized contour was eventually acquired using the MATLAB image analysis programs (d).

the osmotic pressure. In the equation, DP refers to the hydrostatic


pressure difference between the inside and outside of the cell and
Dp refers to the osmotic pressure difference. In the model, the outside pressure is held constant throughout (at a reference value of
0 Pa) and the compression is assumed to be fast enough to neglect
biological and physiological processes of the cell that might affect
the wall properties. Furthermore, it was assumed that water ows
are small both during ination and compression, so that the concentration of the cellular uid can be considered constant and at osmotic equilibrium with the medium. Thus, Dp in Eq. (1) vanishes
and the equation simplies to

bar
Inflated cell
2ri

cell
r0

Fig. 2. Schematic view of the cell compression. An uninated (zero turgor pressure)
cell with outer radius r0 is rst inated to a radius ri by raising its internal pressure
to a prescribed initial turgor pressure. The ination is then followed by compression
under a at rigid (probe) surface at compression velocity v.

radius r0 and initial wall thickness h0 is rst inated by raising its


internal pressure to a prescribed initial turgor pressure P0 and initial
outer radius ri. Because the uninated cell is assumed to be in equilibrium with the surrounding medium, it has the same osmotic
pressure. The inated condition is the state of the cell at the start
of the experimental compression. The ination is then followed
by compression under a at rigid (probe) surface at compression
velocity v. The initial stretch ratio ks of the cell is dened as r i =r0 .
The hydraulic conductivity Lp of the wall is considered to be constant throughout. The ow of uid from the cell was not explicitly
modelled in this case, rather it was represented as an internal pressure change. Volume loss during the compression of a cell can be
described as follows (Kedem and Katchalsky, 1958; Smith et al.,
1998):

dV Lp
SDP  Dp
dz
v

where V is the volume of the cell, z is the vertical displacement of


the compression surface, S is the surface area of the cell available
for ow (i.e. excludes the contact areas between cell and probe,
and cell and glass surface), P is the hydrostatic pressure while p is

dV Lp
SP
dz
v

In principle it is possible to determine the osmotic pressure as


well as the diameter of the uninated cell based on visual observation of cell volume changes in solutions of different osmolality but
these data were not available.
The compressed sphere is modelled as axisymmetric around the
axis that lies along the line of motion of the compression probe (zaxis). In addition, the problem symmetry in the horizontal plane
through the centroid of the cell means that only half of the model
needs to be analysed. The cell wall is modelled as isotropic and
purely elastic. Large strains are considered in the model while
the tangential/shear forces on the wall surface are neglected. The
structural mechanics module of the commercial package, COMSOL
Multiphysics 3.2 (COMSOL Group, Stockholm, Sweden) was used
for modelling.
2.3.2. Contact model
In the model, the compression surface is not explicitly modelled
and meshed. Instead a soft contact procedure is implemented
(Pennec et al., 2007). In this procedure the contact is modelled with
a force that increases rapidly as the rigid object approaches the
edge of the mesh. The contact is represented by a pressure of the
type

pcontact AeB:g

268

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

where g is the gap between the contacting surfaces (the at bar and
the surface of the cell). The constants A and B have no physical
meaning and should be selected carefully. The value of A should
be of the same order as the expected contact pressure. The constant
B must be large enough not to cause any signicant forces over
physically important gap distances; however, excessively large values would increase the nonlinearity of the problem, which might be
detrimental to the convergence. The soft contact procedure was
implemented directly as a boundary expression to the boundary
where contact is expected. For these models, after examining the local maxima of the pressures on the cell wall for several combinations of the two parameters, a value of 9 MPa was used for A and
B was set to 5  107 m1.
2.3.3. Material properties
The cell wall was modelled as isotropic and elastic (Hookean)
material with the following material properties: cell wall density
qw 1000 kg m3 (by assumption); hydraulic conductivity Lp =
4.64  1013 m (s Pa)1 (Maggio and Joly, 1995) Youngs modulus
E = 2.3  109 Pa (Wang et al., 2004); Poissons ratio m 0:4 (Wang
et al., 2004). The liquid inside the cell was assumed to be
incompressible and with similar properties to water (density
qf 1000 kg m3 ; bulk modulus Kf = 2  109 Pa).
2.3.4. Constraints and loads
During the ination phase of the model, a pressure load is applied to the internal boundary equal to the prescribed initial turgor
pressure. The initial turgor was adopted from Wang et al. (2004) as
0.363 MPa. During the compression stage, the soft contact conditions are applied to the external boundary. The rigid probe surface
is lowered in a series of downward displacement steps of 0.2 lm.
The active area and pressure are recalculated at every step and
the volume loss is accounted for through a boundary expression
on the internal wall boundary according to Eq. (2). Careful consideration should be given to the fact that not only the volume loss
depends on the pressure resulting from the direct FE solution, this
pressure in turn is also inuenced by the calculated volume loss.
So, instead of determining the volume loss in a single step, one
must do this iteratively (Smith et al., 1998). Hence, an iteration cycle was implemented in this model with a set minimum difference
between two subsequent values for the pressure as a stop criterion.
2.3.5. Implementation
The nite element mesh for the FE model was produced using
quadratic Lagrange elements (Fig. 3). The model contained a total
of 795 elements. Fig. 3a shows a portion of the mesh from the
top of the cell model. In order to determine the adequateness of
the mesh density, preliminary analysis results were compared to
the results obtained with a ner mesh (Fig. 3b). The results were

Fig. 4. Typical deformed shape plot of the cell after compression. Cell initial
diameter = 62 lm. A surface plot of the von Mises stresses is also included (grey
scale). The von Mises stresses are highest at the edge of the contact surface (approx.
0.5 GPa) and lowest at the centre of the contact surface (approx. 13 MPa).

identical, and therefore the mesh density was judged to be


adequate.
The model analysis was performed using a non-linear parametric solver with a direct linear equation system solver and large
deformation. The span of the load stepping parameter was chosen
to provide similar deformation as the corresponding experiments.
2.4. Simulation
The model was used to simulate data for a single cell from the
experiment reported by Wang et al. (2004). Hence the cell model
was developed with the dimensions and material properties as
determined (or estimated) in that work. The FE model analysis
was then performed and the results were compared to the experimental data. Comparisons were made for forcedeformation data
as well as cell deformed shapes at different stages of the compression. The main properties of the cell from that particular experiment were as follows: cell diameter before compression = 62 lm;
initial turgor pressure = 0.363 MPa; initial thickness of the cell
wall = 126 16 nm. The other constants were as given in Section
2.3.3. The speed of the cell compression was 23 lm s1.
3. Results
Fig. 4 shows the deformed shape of the cell after a simulated
compression with a at rigid bar to a prescribed maximum vertical

Fig. 3. A zoom view of the mesh of the cell model from the top end. The mesh in (a) was used in the formulation while the ner mesh in (b) was used to assess the
adequateness of the mesh density.

269

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

against d, the fractional deformation of the cell. The fractional


deformation is dened as follows:

d1

r i  z
ri

where z refers to the vertical displacement of the cell surface in contact with the at object and ri is the initial outer radius of the cell
after ination and before compression. There is a remarkable agreement between the FE model and experiment in the lower deformation ranges, up to a fractional deformation of approximately 0.2. At
this stage a systematic departure clearly occurs with the real cell
prole appearing to atten and follow a more straightened prole
until the cell bursts.
3.1.2. Deformed shapes of the cell
Fig. 6 shows a comparison of the deformed shapes of the cell in
the compression experiment (obtained from image analysis and
digitization of real cell images) to those generated with the FE
model for a series of stages of the compression (indicated by the
fractional deformations). In agreement with forcedeformation
curve discussed above, the FE model makes a very accurate prediction of the deformed shapes of the cell compression in the lower
deformation ranges. From the gure, cell shapes of fractional deformations up to 0.23 are predicted reasonably accurately by the
model. At higher deformations the model predictions start to depart systematically from the measured data; the latter show a
higher lateral expansion than the former.
The validation results show that the FE model is capable of
accurately predicting the force deformation behaviour of a cell
undergoing small to moderate compression and hence would be
useful for determining the mechanical properties of the cell. The
discrepancy between the model and the actual cell experiments
at larger deformations is understandable, because at that stage
the cell is nearing bursting. A possible interpretation of this
behaviour would be that, at these rates of compression, the cell

Fig. 5. Comparison of the forcedeformation proles determined from the FE model


and those obtained from experiment. The experiment and material properties are
3
according to Wang et al. (2004): r 0 62 nm; h0 126 nm; qw 1000 kg m ;
Lp 4:64  1013 m s Pa1 ; E 2:3  109 Pa; P 0 3:63  105 Pa; m 0:4 and
v 23 lm s1 .

displacement. The undeformed (but inated) shape of the cell is included for comparison. The von Mises stresses are calculated from
the stress components and may be used to predict whether the
(ductile) material may yield (Barber, 2002). Here the von Mises
stresses appear to reach a maximum at the edge of the contact surface with the at rigid manipulator, suggesting that cell bursting
would occur at this edge (assuming the cell wall material is
ductile).
3.1. Validation of the FE model
3.1.1. Forcedeformation curves
Fig. 5 shows a comparison of the results of the FE model to those
of experiment. In the gure the external applied force is plotted
-5

-5

x 10

4.5

4.5

-5

x 10

4.5

3.5

3.5

3.5

2.5

2.5

2.5

1.5

1.5

1.5

= 0.00

= 0.085

0.5
0.5

1.5

2.5

3.5

4.5

0.5

0
0

0.5

1.5

2.5

3.5

4.5
-5

-5

0
0

x 10

4.5

3.5

3.5

2.5

2.5

2.5

1.5

1.5

1.5

= 0.230

0
0

= 0.394

0.5
0.5

1.5

2.5

3.5

4.5

2.5

3.5

0
0

0.5

4.5

x 10

1
0.5

1.5

-5

2.5

3.5

4.5
-5

x 10

x 10

simulation

-5

3.5

1.5

-5

4.5

0.5

x 10

-5

-5

x 10

0.5

x 10

x 10

4.5

= 0.162

0.5

0
0

x 10

0
0

= 0.443
0.5

1.5

2.5

3.5

4.5
-5

x 10

experiment

Fig. 6. Comparison of the deformed shapes of the real cells to those of the FE model at different stages of the cell compression. d is the fractional deformation of the cell.
r0 = 62 lm; h0 = 126 nm; qw = 1000 kg m3; Lp = 4.64  1013 m (s Pa)1; E = 2.3  109 Pa; P0 = 3.63  105 Pa; m = 0.4 and v = 23 lm s1. Dimensions in the gure are metres.

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

wall constitutive behaviour is characterised by an elasticplastic


response where a plastic yield point might be identied at the
point of substantial departure from the FE model. The yield point
would then be followed by a portion of softening before the cell
nally bursts. However, the validation of this hypothesis would
require tensile tests on isolated cell walls which is currently not
feasible.

0.0045
Turgor Pressure = 0.363 M Pa

0.004

Force (N)

270

0.0035

Turgor Pressure = 0.2 M Pa

0.003

Turgor Pressure = 0.5 M Pa

0.0025
0.002
0.0015

3.2. Sensitivity of the FE model to the thickness of the cell wall,


Poissons ratio, hydraulic conductivity and initial turgor pressure

0.001
0.0005
0

In order to assess the robustness of the cell model developed to


various crucial variables and properties of plant cells, the model as
described above was analysed for its sensitivity to each of the variables by repeating the analysis at different values of each variable
while holding the rest and comparing the resultant forcedeformation curves. Four parameters were investigated, namely: the initial
thickness of the cell wall; the Poissons ratio of the cell wall; the
hydraulic conductivity of the cell wall and the initial turgor pressure of the cell. All the other cell properties and experimental variables were held as described in the validation above. The inuence
of the cell wall elastic modulus has been addressed by others before (Smith et al., 1998; Wang et al., 2004) and hence was not
investigated.
3.2.1. Sensitivity to cell wall thickness
Fig. 7 shows the forcedeformation response of the cell model
for three values of the initial cell wall thickness, based on the range
of values found by Wang et al. (2004). There is a clear inuence of
this variable as a thicker wall results in a stiffer response of the cell.
This inuence is to be expected as a thicker cell should provide
more structural rigidity to the cell.
3.2.2. Sensitivity to cell wall Poissons ratio
The effect of for three different cell wall Poissons ratios (0.3, 0.4
and 0.49) was investigated (results not shown). No substantial
inuence by the Poissons ratio of the cell wall on the overall
forcedeformation behaviour of the cell could be observed. A similar nding was reported by Wang et al. (2004).
3.2.3. Sensitivity to the initial turgor pressure
In the analytical model of Wang et al. (2004), a pre-determined
value of the initial turgor pressure was used to avoid the need to
treat this as an adjustable parameter. Wang et al. (2004) determined a mean value experimentally. Because individual cells
may vary, a sensitivity analysis was done using the FE model.
Fig. 8 shows a plot of the forcedeformation curves of the cell
model at three different values of initial turgor pressure. Clearly
6.0E-03
1.42E-07 m
1.26E-07 m
1.10E-07 m

5.0E-03

Force (N)

4.0E-03

0.05

0.1

0.15
0.2
0.25
Fractional Deformation

0.3

0.35

Fig. 8. Inuence of the initial turgor pressure on the forcedeformation of the cell
FE model. r0 = 62 lm; h0 = 126 nm; qw = 1000 kg m3; Lp = 4.64  1013 m (s Pa)1;
E = 2.3  109 Pa; m = 0.4 and v = 23 lm s1.

this has an inuence on the forcedeformation behaviour, with


higher turgor pressures resulting in stiffer responses. In addition
to the direct effect of pressure, there will be an additional effect
through consequential changes in the initial stretch ratio of the cell
wall. It is consistent that Wang et al. (2004) found that higher initial stretch ratios led to stiffer forcedeformation responses. It is
clear that changes in initial turgor pressure from 0.2 to 0.5 MPa,
do not have a huge effect on the simulated forcedeformation
curves.
3.2.4. Sensitivity to the hydraulic conductivity of the cell wall
The range of values of hydraulic conductivity Lp for plant cells in
general is reported to be in the range of 2  108 to 105 ms1 MPa1
(Wang et al., 2004; Maurel, 1997) while for tomato cells the value is
reported to be around 107 ms1 MPa1 (Wang et al., 2004; Maggio
and Joly, 1995; Hukin et al., 2002). In the model described herein the
value of Lp used for the tomato cell is: 4.64  107 ms1 MPa1. To
investigate the sensitivity of the cell model to Lp, the simulation
was repeated for a number of different values of Lp and comparisons
were made not just for the forcedeformation responses of the cell,
but also for the volume versus deformation curves. The values of Lp
used ranged from 0 to 0.01 ms1 MPa1 (i.e. way outside the range of
not just tomato cells, but plant cells in general). For all 4 values of Lp,
there was no noticeable effect on the forcedeformation curve of the
cell models (Fig. 9a). For the volume-deformation curve (Fig. 9b),
still there was no noticeable difference between the simulated results for realistic values of Lp. A signicant effect on the volume of
the cell only appeared when a value of 0.01 ms1 MPa1 (which is
5 orders of magnitude higher than the normal 107 ms1 MPa1
and 3 orders of magnitude higher than the range for all plants)
was used. This extreme value however still did not show any effect
on the forcedeformation curve.
In view of the foregoing, it can be concluded that, for tomato
cells being compressed under the conditions described here, the
hydraulic conductivity does not signicantly affect the force
deformation behaviour and thus is not a crucial parameter. A similar nding was reported by Wang et al. (2004).

3.0E-03

4. Discussion

2.0E-03
1.0E-03
0.0E+00
0.00

0.05

0.10
0.15
0.20
0.25
Fractional deformation

0.30

0.35

Fig. 7. Inuence of initial cell wall thickness to the forcedeformation behaviour of


the cell. r0 = 62 lm; qw = 1000 kg m3; Lp = 4.64  1013 m (s Pa)1; E = 2.3 
109 Pa; P0 = 3.63  105 Pa; m = 0.4 and v = 23 lm s1.

The nite element model developed in this manuscript is equivalent to the model of Wang et al. (2004) and incorporates the same
physics. However, unlike the latter which essentially is an analytical model, the main advantage of the nite element model is that
other constitutive equations for the mechanical behaviour of the
cell wall material can be incorporated easily. Further, it is sufciently exible to serve as a building block for a more comprehensive numerical model of tissue deformation. Such a model may

271

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

0.50

b 5.776E-14

0.45

5.774E-14

0.40

5.772E-14

0.35

5.770E-14

0.30

4.81E-11
4.81E-11

5.768E-14

Volume (m)

Dimensionless Force

4.64E-13
4.64E-13

0.25

0.00E+00

0.20

1.00E-08
1.00E-08

0.15

5.766E-14
5.764E-14

4.81E-11

5.762E-14

4.64E-13

0.10

5.760E-14

0.00E+00

0.05

5.758E-14

0.00
0.00

1.00E-08

0.05

0.10

0.15

0.20

0.25

0.30

0.35

5.756E-14
0.00

0.05

Fractional deformation

0.10

0.15
0.20
0.25
Fractional deformation

0.30

0.35

Fig. 9. Inuence of the hydraulic conductivity on the forcedeformation (a) and the volume (b) of the cell FE model. r0 = 62 lm; h0 = 126 nm; qw = 1000 kg m3;
E = 2.3  109 Pa; P0 = 3.63  105 Pa; m = 0.4 and v = 23 lm s1.

allow to understand how fruit behaves under mechanical loading.


This knowledge may guide towards adapting postharvest handling
procedures to minimise damage or to development novel methods
to measure fruit texture.
The model appears to reproduce remarkably well the experimental forcedeformation behaviour of the cell from experiments
based on the parameters as determined by Wang et al. (2004). It
was discussed earlier that the mechanical properties as determined by Wang et al. (2004) lack certainty. This is because the
method used to extract the elastic modulus of the cell relied on
least squares curve ttings that were achieved with two free
parameters: the initial stretch ratio and the Youngs modulus. In
addition, the cell wall hydraulic conductivity was not explicitly
measured for these specic cells but adopted from the literature.
In view of the observations of Smith et al. (1998) regarding the
non-uniqueness of solutions for compression of single cells there
was uncertainty as to the values of the parameters since several
combinations of parameters could still have produced good ts.
With the FE model, the initial stretch ratio does not need to be
re-estimated as a tting parameter since it is xed by the model
once the initial turgor pressure and the cell dimensions are known.
This is a major advantage as the (mean) initial turgor pressure and
the cell dimensions are easy to measure accurately (Mashmoushy
et al., 1998; Shiu et al., 1999; Blewett et al., 2000; Thomas et al.,
2000; Wang et al., 2005), although the initial turgor pressure of a
specic cell under test is not known. This simplication is achievable by assuming the osmotic pressure of the cells does not change
markedly from the uninated state to the compressed state. This is
reasonable as the volume changes during this processes are small
(<4%). From the studies on the sensitivity of the FE model of the cell
to the hydraulic conductivity, it has emerged that the hydraulic
conductivity does not have any noticeable effect of the force
deformation, at least for a practical range of values. Since all the
other parameters that could inuence the forcedeformation
curves (initial cell wall thickness and the (mean) initial turgor
pressure) were determined accurately from experiments, it can
be concluded that the elastic modulus of the cells as determined
by Wang et al. (2004) was correct.
In spite of the foregoing, it is worth remarking that the elastic
modulus of the cell as determined by Wang et al. (2004) and implemented in the FE model (E = 2.3  109 Pa) is surprisingly high,
especially when placed against the elastic moduli of fruit tissue.
Apple fruit tissues, for example, have elastic moduli in the region
of 5 MPa while tomato fruit tissue has its elastic moduli in the region of 0.5 MPa. The modulus of the cell wall used here is therefore
at least 3 orders of magnitude higher than that of tissue made of

similar cells. This is possible since the tissue, as a higher order


composite material, has its mechanical properties inuenced by
many other factors, such as the properties of the binding matrix,
the structural architecture of the constituent cells, and water ows
through the structure. A review of the literature on mechanical
properties of plant cell walls reveals a range of elastic moduli from
as high as 15 GPa for the primary cell walls of wood cell walls,
through 0.4 GPa, for cell walls of the shoots of seedlings of white
spruce, to 0.3 GPa, for cell walls of green algae (Marshall and
Dumbroff, 1999; Gindl and Gupta, 2002; Wei and Lintilhac,
2007). Hiller et al. (1996) established a value of 3 GPa for the cell
wall of potato tuber parenchyma tissue cells. In a more recent paper, Wang et al. (2006) found values of 3080 MPa for the cell wall
of inner pericarp tissue of tomato.
Regarding the prediction of the constitutive behaviour of the
cell material, the results of the FE model suggest that an elastic
constitutive relation with large strains is sufcient to explain the
forcedeformation behaviour of the cell at fractional deformation
of up to 0.2. At higher fractional deformations, it would appear that
elasticplastic constitutive behaviour with softening and an ultimate bursting point would be more appropriate. It is an advantage
of the FE model that it might eventually be possible to incorporate
such elasticplastic behaviour. The FE model developed here has a
similar level of accuracy as the model of Smith et al. (1998). However, it is much less computationally intensive. The implementation of the soft contact algorithm for this model, rather than the
fully meshed at compression bar implemented by the former, signicantly simplies this model without sacricing modelling
accuracy.
5. Conclusions
A nite element model was developed to simulate micromanipulation compression testing of single plant cells. It was suitable for
situations with large strains as found in cell compression experiments and used a soft contact algorithm to reduce computational
intensity. The model was validated using data from a real single
cell compression experiment performed on a suspension cultured
cell. Assuming linear elastic constitutive behaviour of the cell wall,
the model was found to be capable of reproducing the forcedeformation behaviour as well as the deformed shapes of the cell up to
fractional deformations of about 0.2.
Unlike the analytical model of Wang et al. (2004), the model
might be used to simulate single cells of any plant or any similar
cells by adjusting the relevant parameters and experimental variables. Because of its computationally simplicity it is believed that

272

E. Dintwa et al. / Journal of Food Engineering 103 (2011) 265272

this nite element model can serve as a building block for a more
comprehensive numerical model of tissue deformation. Such a
model can be used to improve our understanding of the
forcedeformation behaviour of fruit and vegetable tissue under
mechanical loading. This might guide to better postharvest handling practices for fruit and vegetables or inspire the development
of novel techniques to measure their texture properties.
Acknowledgements
The authors wish to thank the Research Council of the K.U. Leuven (OT 08/023), the Flanders Fund for Scientic Research (Project
G.0603.08; 3E060094) and the Engineering and Physical Sciences
Research Council, UK for nancial support.
References
Akyurt, M., Zachariah, G.L., Haugh, C.G., 1972. Constitutive relations for plant
materials. Transactions of the ASAE 15, 766769.
Barber, J.R., 2002. Elasticity second ed. Kluwer Academic Publishers, Dordrecht, The
Netherlands. 413 p.
Blewett, J., Burrows, K., Thomas, C., 2000. A micromanipulation method to measure
the mechanical properties of single tomato suspension cells. Biotechnology
Letters 22, 18771883.
Gao, Q., Pitt, R.E., 1991. Mechanics of parenchyma tissue based on cell orientation
and microstructure. Transactions of the ASAE 34, 232238.
Gates, R.S., Pitt, R.E., Ruina, A., Cooke, J.R., 1986. Cell wall elastic constitutive laws
and stressstrain behaviour of plant vegetative tissue. Biorheology 23, 453466.
Ghysels, P., Samaey, G., Tijskens, E., Van Liedekerke, P., Ramon, H., Roose, D., 2009.
Multi-scale simulation of plant tissue deformation using a model for individual
cell mechanics. Physical Biology 6, Art. No. 016009.
Gindl, W., Gupta, H.S., 2002. Cell-wall hardness and Youngs modulus of melaminemodied spruce wood by nano-indentation. Composites Part A: Applied Science
and Manufacturing 33, 11411145.
Hiller, S., Bruce, D.M., Jeronimidis, G., 1996. A micro-penetration technique for
mechanical testing of plant cell walls. Journal of Texture Studies 27, 559587.
Hukin, D., Doering-Saad, C., Thomas, C.R., Pritchard, J., 2002. Sensitivity of the cell
hydraulic conductivity to mercury is coincident with symplasmic isolation and
expression of plasmalemma aquaporin genes in growing maize roots. Planta
215, 10471056.
Kedem, O., Katchalsky, A., 1958. Thermodynamic analysis of the permeability of
biological membranes to non-electrolytes. Biochimica Biophysica Acta 27, 229
246.
Lardner, T.J., Pujara, P., 1978. On the contact problem of a highly inated spherical
nonlinear membrane. Journal of Applied Mechanics 45, 202203.
Lardner, T.J., Pujara, P., 1980. Compression of spherical cells. Mechanics Today 5,
161176.
Maggio, A., Joly, R.J., 1995. Effects of mercuric chloride on the hydraulic conductivity
of tomato root systems. Plant Physiology 109, 331335.
Marshall, J.G., Dumbroff, E.B., 1999. Turgor regulation via cell wall adjustment in
white spruce. Plant Physiology 119, 313319.

Mashmoushy, H., Zhang, Z., Thomas, C.R., 1998. Micromanipulation measurement of


the mechanical properties of bakers yeast cells. Biotechnology Techniques 12,
925929.
Maurel, C., 1997. Aquaporins and water permeability of plant membraines. Annual
Review of Plant Physiology 48, 399429.
McLaughlin, N.B., Pitt, R.E., 1984. Failure characteristics of apple tissue under cyclic
loading. Transactions of the ASAE 27, 311320.
Nilsson, S.B., Hertz, C.H., Falk, S., 1958. On the relationship between turgor pressure
and tissue rigidity. Physiologia Plantarum 11, 818837.
Pennec, F., Achkar, H., Peyrou, D., Plana, R., Pons, P., Courtade, F., 2007. Verication
of contact modeling with COMSOL multiphysics software. In: Proceedings of the
Sixth EUROSIM Congress on Modelling and Simulation, Ljubljana, Slovenia,
2007, September 913, 2007, 6p.
Pitt, R.E., 1982. Models for the rheology and statistical strength of uniformly
stressed vegetative tissue. Transactions of the ASAE 25, 17761784.
Pitt, R.E., Chen, H.L., 1983. Time-dependent aspects of the strength and rheology of
vegetative tissue. Transactions of the ASAE 26, 1275.
Pitt, R.E., Davis, D.C., 1984. Finite element analysis of uid-lled cell response to
external loading. Transactions of the ASAE 27, 19761983.
Seymour, G.B., Manning, K., Eriksson, E.M., Popovich, A.H., King, G.J., 2002. Genetic
identication and genomic organization of factors affecting fruit texture.
Journal of Experimental Botany 53 (377), 20652071.
Shiu, C., Zhang, Z., Thomas, C.R., 1999. A novel technique for the study of bacterial
cell mechanical properties. Biotechnology Techniques 13, 707713.
Smith, A.E., Moxham, K.E., Middelberg, A.P.J., 1998. On uniquely determining cellwall material properties with the compression experiment. Chemical
Engineering Science 53, 39133922.
Smith, A.E., Moxham, K.E., Middelberg, A.P.J., 2000a. Wall material properties of
yeast cells: Part II. Analysis. Chemical Engineering Science 55, 20432053.
Smith, A.E., Zhang, Z., Thomas, C.R., 2000b. Wall material properties of yeast cells:
Part I. Cell measurements and compression experiments. Chemical Engineering
Science 55, 20312041.
Smith, A.E., Zhang, Z., Thomas, C.R., Moxham, K.E., Middelberg, A.P.J., 2000c. The
mechanical properties of Saccharomyces cerevisiae. Proceedings of the National
Academy of Sciences 97, 98719874.
Stenson, J.D., Thomas, C.R., Hartley, P., 2009. Modelling the mechanical properties of
yeast cells. Chemical Engineering Science 64, 18921903.
Thomas, C.R., Zhang, Z., Cowen, C., 2000. Micromanipulation measurements of
biological materials. Biotechnology Letters 22, 531537.
Wang, C.X., Cowen, C., Zhang, Z., Thomas, C.R., 2005. High speed compression
of single alginate microspheres. Chemical Engineering Science 60, 6649
6657.
Wang, C.X., Pritchard, J., Thomas, C.R., 2006. Investigation of the mechanics of single
tomato fruit cells. Journal of Texture Studies 37, 597606.
Wang, C.X., Wang, L., Thomas, C.R., 2004. Modelling the mechanical properties of
single suspension-cultured tomato cells. Annals of Botany 93, 443453.
Wei, C., Lintilhac, P.M., 2007. Loss of stability: a new look at the physics of cell wall
behaviour during plant cell growth. Plant Physiology 145, 763772.
Wu, N., Pitts, M.J., 1999. Development and validation of a nite element model of an
apple fruit cell. Postharvest Biology and Technology 16, 18.
Yang, W.H., Feng, W.W., 1970. On axisymmetric deformations of nonlinear
membranes. Journal of Applied Mechanics 37, 10021011.
Zhang, Z., Ferenczi, M.A., Lush, A.C., Thomas, C.R., 1991. A novel micromanipulation
technique for measuring the bursting strength of single mammalian cells.
Applied Microbiology Biotechnology 36, 208210.

Você também pode gostar