Journal of Biotechnology 119 (2005) 181–196

Flow modelling within a scaffold under the influence of

uni-axial and bi-axial bioreactor rotation
H. Singh a,1 , S.H. Teoh b,2 , H.T. Low b,3 , D.W. Hutmacher b,∗
a Tissue Engineering Laboratory E3-05-04, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore
b Division of Bioengineering, National University of Singapore, 9 Engineering Drive 1, EA 03-12, Singapore 117576, Singapore

Received 20 September 2004; received in revised form 17 March 2005; accepted 29 March 2005

Abstract

The problem of donor scarcity has led to the recent development of tissue engineering technologies, which aim to create
implantable tissue equivalents for clinical transplantation. These replacement tissues are being realised through the use of
biodegradable polymer scaffolds; temporary/permanent substrates, which facilitate cell attachment, proliferation, retention and
differentiated tissue function. To optimise gas transfer and nutrient delivery, as well as to mimic the fluid dynamic environment
present within the body, a dynamic system might be chosen. Experiments have shown that dynamic systems enhance tissue
growth, with the aid of scaffolds, as compared to static culture systems. Very often, tissue growth within scaffolds is only seen
to occur at the periphery. The present study utilises the Computational Fluid Dynamics package FLUENT, to provide a better
understanding of the flow phenomena in scaffolds, within our novel bioreactor system. The uni-axial and bi-axial rotational
schemes are studied and compared, based on a vessel rotating speed of 35 rpm. The wall shear stresses within and without the
constructs are also studied. Findings show that bi-axial rotation of the vessel results in manifold increases of fluid velocity within
the constructs, relative to uni-axial rotation about the X- and Z-axes, respectively.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Computational fluid dynamics; Tissue engineering; Bioreactor

1. Introduction

In a number of long-term tissue engineering stud-

∗ Corresponding author. Tel.: +65 6874 1036; fax: +65 6872 3069.
E-mail addresses: g0301488@nus.edu.sg (H. Singh),
mpetsh@nus.edu.sg (S.H. Teoh), mpelowht@nus.edu.sg
(H.T. Low), biedwh@nus.edu.sg (D.W. Hutmacher).
1 Tel.: +65 6874 8870; fax: +65 6777 3537.
2 Tel.: +65 6874 4605; fax: +65 6872 3069.
3 Tel.: +65 6874 2225; fax: +65 6872 3069.
ies under static conditions (e.g. Petri dish), it has
proven extremely difficult to promote the high-density
three-dimensional in vitro growth of cells that have
been removed from the body and deprived of their
normal in vivo vascular sources of nutrients and gas
exchange. To optimise gas transfer and nutrient deliv-

0168-1656/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jbiotec.2005.03.021
182 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
ery, as well as to mimic the fluid dynamic environment
present within the body, a dynamic system has to be
chosen. The bioreactor is one example that attempts
to fulfil these requirements, particularly in the study
of tissue engineering, by simulating the physiological
and fluidic conditions that occur in vivo (Freed and
Vunjak-Novakovic, 2002; Papoutsakis, 1991; Cherry
and Papoutsakis, 1986). Cells have often been seen to
grow well along the periphery of 3D scaffold, while
proliferation is often significantly affected at the centre
of the scaffold, where necrotic neo-tissue can some-
times be seen. This is partly due to poor fluidic transport
of media and scaffold design, among other reasons.
Therefore, simulations that provide such flow visual-
izations could greatly assist in the design of scaffolds
as well as bioreactor.
A number of experimental studies were previously
reported by using different bioreactor systems (Freed
et al., 1994; Martin et al., 2004; Vunjak-Novakovic et
al., 1996). However, limited information has been pub-
lished on studies that attempt to simulate the dynamic
fluid environment prior to commencing experimen-
tal studies. Advantages of computational simulations
include the ability to modify and study the effects of
bioreactor design with respect to the flow analysis,
without having to develop and construct actual phys-
ical models, or to run a large number of experiments.
This is coupled with the significant savings in time and
costs. Furthermore, visualisation of flow as enabled by
the simulation package is a key factor in determining
the efficacy of the system, and allows for design opti-
mization prior to bioreactor design and modifications.
NASA has taken significant strides relating to their
rotating bioreactor studies, under the influence of unit
and micro-gravity. While it is known that in the pres-
ence of body forces, density differences between the
cells attached to micro-carriers and the fluid medium,
cause relative motion resulting in mechanical shear
and increased mass transport. However, in the micro-
gravity environment, buoyancy effects are greatly
reduced. The gravity of Earth is replaced by centripetal
acceleration as the dominant body force. For a typ-
ical rotation rate of 2 rpm, within a 0.05 m diameter
vessel, the magnitude of the body force is reduced
(Kleis and Pellis, 1995) to approximately 0.001 m/s2 .
Kleis et al. (1996) proceeded further by carrying
out studies on mass transport, with regards to their
micro-gravity bioreactor model. Consequently Boyd
and Gonda (1990) mathematically modelled the motion
of particles within the NASA bioreactor model at both
unit and micro-gravity. However, the effects of gravity
can be felt by almost everything on Earth, and is not
ignored here for practical reasons. Rotating (Bursac
et al., 2003; Martin et al., 2004) and perfusion (Martin
et al., 2004; Bancroft et al., 2003; Darling and
Athanasiou, 2003; Sodian et al., 2002) bioreactors,
each with their own flow characteristics, are currently
being utilized for studies involving the culturing of
bone and cartilage. These bioreactor models induce dif-
ferent normal and shear forces that act on cells, with
varying consequences.
One aspect that this paper will focus on is that of
shear stresses acting on cells and tissues. A response
of an endothelial monolayer of cells subject to steady,
laminar shear stresses is that of an alteration in mor-
phology. A typical change would occur from an ini-
tial polygonal pattern, to one, which depicts an elon-
gated profile, being aligned to the direction of fluid
flow (Levesque and Nerem, 1985; Levesque et al.,
1989; Stathopoulos and Hellums, 1985). This change
is basically accompanied by the restructuring of the
cytoskeleton, or more specifically, the alignment of
microtubules, followed by the formation of actin stress
fibres. Similar experiments were conducted on bovine
endothelial cells. It was found that the stiffness of
endothelial cells exposed to shear stresses of 2 Pa,
increased with the duration time of exposure. However,
after 24 h of exposure, the stiffness of the endothe-
lial cells was similar all around the cell, indicating
the ability of the cells to adapt to the changes in the
environment. This concept includes that of stress fibre
orientation, as mentioned in the paper by Yamada et al.
(2000).
Olivier and Truskey (1993) showed that the flow
regime itself, and not just the magnitude of shear
stress, plays a critical part. Their experiments involved
turbulent flows as generated in a cone and plate vis-
cometer, which actually resulted in the detachment of
anchorage-dependent cells at shear stresses of only
1.5 dyn/cm2 .
At relatively higher levels of shear stress, rang-
ing from 26 to 54 dyn/cm2 , human endothelial cells
were found to detach from its surface and correspond-
ingly exhibited reduced viability (Stathopoulos and
Hellums, 1985). In contrast to this, no cell loss was
reported for bovine aorta endothelial cells, which were
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196 183

subject to a stress level (Levesque and Nerem, 1985)

of approximately 85 dyn/cm2 . Pritchard et al. (1995)
found that adhesion generally increases with decreas-
ing wall shear stress. It is also now well documented
that cell metabolism is affected by fluid stresses. Shear
stresses, within limits, tend to stimulate the release of
specific enzymes and growth factors, thereby enhanc-
ing cellular attachment and proliferation.
Stathopoulos and Hellums (1985) found that HEK
cells release urokinase, due to variations in shear stress
levels. Furthermore, human umbilical vein endothelial
cells were found to release a five-fold amount of prosta-
cyclin at 10 dyn/cm2 , as compared to near-zero stress
conditions. Similarly, Vunjak-Novakovic et al. (1996,
1998) studied the affects of dynamic seeding within
polymer scaffolds with respect to chondrocytes, and Fig. 2.1. Actual bioreactor prototype model.
found that these cells can be “conditioned” to grow
under the correct shear stress levels. Chondrocytes have ments of the flow solution. Flow solutions are based on
been known to proliferate better under the influence of the conservation of mass, momentum and energy equa-
shear stresses and are able to withstand higher loads as tions. GAMBIT and FLUENT were therefore used in
compared to chondrocytes cultured statically (Sucosky tandem, to model the flow of fluid within our prototype
et al., 2003). Both Porter et al. (2005) and Raimondi bioreactor model.
et al. (2004) utilised CFD techniques to model the
effects of perfusion, to better comprehend the influence 2.2. The bioreactor model
of perfusion and hence shear stresses, on 3D cultures.
Lappa (2005) on the other hand, developed a rigorous A novel bi-axial bioreactor system was developed
set of equations to model fluid flow and algorithms to by a team from the National University of Singapore
estimate soft tissue growth within a bioreactor. Redaelli and the Singapore Polytechnic. The unique design of
et al. (1997) on the other hand, attempted to model the entire system adds some flexibility that is some-
pulsatile flow within arteries. This intriguing article times found to be lacking in commercially available
describes how a FORTRAN algorithm was interfaced bioreactors. This novel design involves the bioreac-
to FIDAP, a commercially available CFD package, and tor being vertically upright, instead of the conven-
then simulated. tional horizontal-type vessel that is so often seen (see
Fig. 2.1). However, the fact that allows this design to
particularly stand out is that it allows for rotation either
2. Software and methodology about the vertical axis (spinning), the horizontal axis
(tumbling), or even rotation simultaneously about both
2.1. GAMBIT and FLUENT axes (gyroscopic action). Ideally, the utilisation of these
features would lead to enhanced cell culture media flow
GAMBIT (Fluent Inc.), being a powerful graph- throughout the entire vessel (see Fig. 2.2 and Table 2.1).
ical modeling tool, was utilised as a pre-processor
for the Computational Fluid Dynamics package, FLU- 2.3. The tissue engineering scaffold
ENT. This general-purpose CFD tool can solve many
fluid flow problems such as steady, unsteady, lami- The scaffolds utilised for the simulations are rou-
nar and turbulent flows, heat transfer, Newtonian and tinely produced by our group (Zein et al., 2002;
non-Newtonian flows, among others. FLUENT also Hutmacher et al., 2004). The poly(␧)caprolactone
provides excellent mesh flexibility as well, allowing (PCL) scaffolds used have been studied extensively for
the grid to be refined or coarsened based on the require- bone engineering applications. The scaffold fibres are
184 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196

Fig. 2.3. MicroCT of a scaffold produced by rapid-prototyping

(Hutmacher et al., 2004).
Fig. 2.2. Profile of novel bioreactor.

Firstly, it must be noted that simulations (without

300 ␮m diameter and form a 90◦ lay-down pattern, to scaffolds) were used to determine appropriate locations
form a scaffold of regular architecture with dimensions for placing the scaffolds. Moreover, flow phenomena
of approximately 5 mm × 5 mm × 5 mm (Fig. 2.3). As were studied, and the locations of turbulent artifacts
with the bioreactor models, the point of origin of the such as eddies and vortices were studied. Consistency
scaffold lies at its very centre. and uniformity of flow throughout the chamber were
also criteria that were particularly sought for. Where
2.4. Scenarios, criteria and guidelines possible, regions of very low or stagnant fluid velocity
were also avoided, as placing scaffolds within these
The following set of simulations were performed, regions would not possibly allow for adequate flow of
and are as follows: culture media within and through the scaffolds, thereby
potentially reducing the flow of nutrients to cells, as
• Bioreactor: rotation about the horizontal X-axis well as the ability of waste to be removed.
(with scaffold); Secondly, only the Z-axis (vertical centre-line pass-
• Bioreactor: rotation about the vertical Z-axis (with ing through the vessel) was utilised as a line of refer-
scaffold); ence, whereby specific positions along this line were
• Bioreactor: bi-axial rotation about the horizontal and identified for placement of scaffolds. The main reason
vertical axes (with scaffold). for this is that shear stresses tend to be minimised at
distances furthest from the walls. In this way, exces-
Table 2.1 sive shear stresses could potentially be avoided. Addi-
Bioreactor dimensions tionally, points along this reference line having higher
Component Dimensions (mm) velocities were selected for scaffold fixation. This was
Vessel diameter 94
done, bearing in mind that there were no eddies or tur-
Vessel height 130 bulent artifacts within close proximity to the selected
Inlet length 6 positions.
Inlet diameter 6
Inlet distance from centre (radially) 23.5
2.5. Conditions, settings and meshing
Outlet tube length 100
Outlet tube inner diameter 6
Outlet tube outer diameter 9 Table 2.2 indicates values and parameters that were
Outlet distance from centre (radially) 23.5 used for the simulations. The contour and velocity plots
Outlet tube through-hole diameter (×6) 3 were captured at equal intervals of 0.0235 m, dividing
Vertical distance between holes 16
the entire vessel into sections of four. On the whole,
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
Table 2.2
Boundary conditions and settings
Culture media
Density (kg/m3 )
Dynamic viscosity (Pa s)
Boundary conditions
Inlet velocity (m/s)
Vessel rotational velocity (rpm)
Gravity (+Z dimension) (m/s2 )
Schemes
Solver: segregated, implicit
Flow: laminar
Pressure: body-force weighted
Pressure–velocity coupling: PISO
Momentum: second order upwind
three sections were at equal intervals along the Y-axis
(Planes 1–3), enabling viewing from the front. Another
three equal sections were taken at the side, enabling
analysis from the transverse perspective point of view.
These transverse sections (Planes 4–6) were conse-
quently “captured” along the X-axis.
All the axial conventions must be appropriately
noted and strictly adhered to. By plotting for the X-,
Y- and Z-directions, the location of the maximum
wall shear stress was determined and pinpointed. The
shear stress at this specific location would therefore
be expected to have adverse effects on cells, and very
possibly resulting in their death. “Counter-measures”
could therefore be devised, so as to prevent damage to
cells at such locations.
The mesh applied (Fig. 2.4) here is that of a tetrahe-
dral mesh. GAMBIT is generally capable of meshing
Fig. 2.4. Sectioned mesh of scaffold via central plane.
185
an entire volume at one go, by applying the same
meshing constraints throughout, such as a constant
interval count. The outlet tube, however, was meshed
1030 individually by meshing its faces due to complex
0.0025
curved surfaces. Meshing was carried out face by face,
with an interval count of 20. Each individual scaffold
0.5895
fibre length was meshed with a 10-interval count. The
35
9.81 remaining volume was the meshed with an interval
count of 50 throughout Mesh convergence analyses
were carried out to prior to the actual simulations
to select a suitable mesh density for this study. Four
sets of 3D meshes of the bioreactor vessel alone were
generated at varying degrees of mesh refinement at
(i) 50606, (ii) 227191, (iii) 282440 and (iv) 321674
tet. elements. All meshes converged successfully with
a convergence error criterion of 0.001. Velocity pro-
files were compared in FLUENT at specific planes and
points. However, only a slight increase in accuracy was
noted (approx. 3%) despite a significant increase in the
number of elements and computational time from (iii)
to (iv). Mesh (iii) was therefore selected based on its
accuracy and its economy.
Fig. 2.5 indicates the sections taken of the bioreactor
for our study. Fig. 2.6 presents a section of the scaffold
as viewed from the front (sectioned vertically), with
Points 1–5 labelled for our analysis. Fig. 2.7 presents
the locations of Points 6–10 within the scaffold, as
viewed from the underside when the scaffold is sec-
tioned horizontally. Fig. 2.8 depicts a single pore within
the scaffold, visually presenting us with flow parame-
ters that will be discussed throughout this article.
Fig. 2.5. Plan view of vessel and corresponding sections along X–Y
planes with planes spaced at equal intervals (quarter-distance) of
0.0235 m.
186 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
Fig. 2.6. Scaffold sectioned along central XZ plane (front view) with
specific points and axis labelled.
Fig. 2.7. Scaffold sectioned along central XY plane (as viewed from
underside of scaffold) with specific points and axis labelled.
Fig. 2.8. Depiction of a scaffold pore with velocity vectors and shear
stresses acting along all surfaces.
Particular attention was paid to the flow environment
within the scaffolds. Prior to this (not shown here), the
most suitable locations were determined for placing the
scaffolds, for each and every rotational scheme. This
was done to “maximise” the performance of each con-
figuration. The configurations were then finally com-
pared.
3. Results and discussion
3.1. Rotation about X-axis
Fig. 3.1.1 reveals the contour plots captured at
Planes 1–3, while Fig. 3.1.2 depicts contour plots at
side Planes 4–6. The scaffold centre was positioned at
the middle of the bioreactor at coordinate (0,0,0).
Fig. 3.1.3 displays the scaffold as sectioned verti-
cally, through its centre. As with the previous scenario,
Points 1–5 were selected for analysis, while Fig. 3.1.4
reveals the presence of an eddy. This eddy can be
observed to exist behind the scaffold. Additionally, it
can also be seen from both figures that the vectors seem
to be of higher “intensity” near the front-right corner
of the scaffold. It is very likely that escalated levels of
wall shear stresses would be expected to occur along
the external surface of the scaffold at these regions.
Table 3.1 displays the velocity magnitudes and u,
v and w components for locations within the scaffold,
as designated by Points 1–5. Points 5 and 4 experience
the highest fluid velocity magnitudes at approximately
3.52 × 10−3 and 3.01 × 10−3 m/s, respectively, while
Points 1 and 2 experience the lowest velocities, at
1.25 × 10−3 and 1.49 × 10−3 m/s. The primary flow
direction is noted to be in the Y-direction. Point 7 can be
seen to experience a relatively high velocity magnitude
with respect to the other points, at 2.81 × 10−3 m/s.
The dominant v velocity component is approximately
−2.6 × 10−3 m/s.
Fig. 3.1.5 depicts the wall shear stresses acting along
the surfaces of the scaffold fibres. The average wall
shear stresses were found to generally lie between 0.8
and 1.2 Pa, as data for a variety of locations within the
scaffold were extracted and counter-checked. However,
specific regions were seen to indicate high peaks in wall
shear stresses.
It can also be noticed that at the very front of the
diagram, a lighter shade of blue can be seen represent-
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196 187

Fig. 3.1.1. Velocity contour plots captured at frontal Planes 1–3 for bioreactor with scaffold rotating about X-axis.

Fig. 3.1.2. Velocity contour plots captured at side Planes 4–6 for bioreactor with scaffold rotating about X-axis.

Table 3.1
Velocities at Points 1–10 within scaffold—bioreactor rotation about X-axis
Points

1 2 3 4 5 6 7 8 9 10
Magnitude (×10−3 m/s) 1.25 1.49 2.12 3.01 3.52 1.77 2.81 1.78 2.04 1.68
x (×10−3 m/s) 0.35 0.50 0.29 0.25 0.31 0.13 0.40 0.30 0.33 0.50
y (×10−3 m/s) −1.20 −1.40 −2.10 −3.00 −3.50 −1.25 −2.60 −1.75 −1.75 −1.00
z (×10−3 m/s) −0.04 −0.12 −0.08 −0.04 −0.14 1.25 1.00 0.05 −1.00 −1.25
188 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196

Fig. 3.1.3. Sectioned scaffold and velocity vector plot at y = 0 m plane within bioreactor rotating about X-axis.

ing approximately 1.2 Pa in shear stresses. This occurs
near the zones of contact between criss-crossing scaf-
fold fibres. It is also noticed that for the external surface,
the colour-coded dark blue colour “softens” to cyan
and light green, from left to right. This implies that
the higher wall shear stresses are located towards the
right side (positive X-direction) of the scaffold. One
very likely reason for this phenomenon relates to the
design of the bioreactor. It is suspected that the pres-
ence of the outlet tube tends to deviate and diverge the
swirling fluid, reducing fluid velocity and shear stresses
especially in areas within close proximity to the outlet
tube. The fluid then accelerates, increasing in veloc-
ity due to the rotating action of the bioreactor, thereby
resulting in higher shear stresses to the right (positive
X-direction) of the scaffold. The location of the peak
wall shear stress can be seen within Fig. 3.1.5, as the
miniscule red spot (indicated by the arrow).

Fig. 3.1.4. Sectioned scaffold and velocity vector plot at z = 0 m plane within bioreactor rotating about X-axis.
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
Fig. 3.1.5. Wall shear stress range of 0–4 Pa for scaffold within biore-
actor rotating about X-axis.
3.2. Rotation of bioreactor with scaffold about
Z-axis
For this scenario, the scaffold was positioned at the
coordinate (0,0,−0.028), 28 mm above the origin rela-
tive to our directional convention. Figs. 3.2.1 and 3.2.2
depict similarities that can be seen to occur with respect
to the Planes 2 and 5, where a central “core” of slow-
moving fluid is noted. This is pronounced in Fig. 3.2.2.
As previously mentioned, the formation of the core
is possibly due to fluid velocity being a function of
Fig. 3.2.1. Velocity contour plots captured at frontal Planes 1–3 for bioreactor with scaffold rotating about Z-axis.
189
the angular velocity and radius of the chamber. As
the angular velocity is held constant, the fluid velocity
increases proportionally as the distance from the rotat-
ing center increases. Furthermore, centrifugal forces
tend to “displace” fluid outwards. In Fig. 3.2.2, an eddy
can be seen to have formed, as shown in Plane 5. The
location of this entity is approximately one third of
the bioreactor height from the base, directly below the
scaffold. This recirculation zone can be related to the
corresponding contour plot in Fig. 3.2.2, within the
dark-blue zone that is in the bottom half of the ves-
sel. Plane 6 also reveals another eddy, corresponding
in location to the contour plot. This phenomenon occurs
near the base of the vessel.
Fig. 3.2.3 displays the sectioned scaffold (as viewed
from the front) where velocity vectors seem to be
deflected by the scaffold. This occurs on the right and
bottom sides of the scaffold. Additionally, a small eddy
can be seen to have formed just slightly to the left of
the scaffold (Fig. 3.2.3). The next frame depicts the
scaffold sectioned along the horizontal plane, where
z = 0 m. Again, the rear (bottom of figure) and right
faces of the scaffold experience heightened levels of
flow (Figs. 3.2.4 and 3.2.5).
Table 3.2 indicates that Points 2 and 5 experi-
ence similarly high fluid velocity magnitudes, being
approximately 3.8 × 10−3 and 3.76 × 10−3 m/s. This
relates well to Fig. 3.2.3, which shows that the vec-
tors approaching from the right are deflected upwards.
190 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196

Fig. 3.2.2. Velocity contour plots captured at side Planes 4–6 for bioreactor with scaffold rotating about Z-axis.

Fig. 3.2.3. Sectioned scaffold and velocity vector plot at y = 0 m plane within bioreactor rotating about Z-axis.

Table 3.2
Velocities at Points 1–10 within scaffold–bioreactor rotation about Z-axis
Points

1 2 3 4 5 6 7 8 9 10
Magnitude (×10−3 m/s) 0.71 3.80 2.02 1.92 3.76 0.94 5.06 2.02 1.04 4.08
x (×10−3 m/s) −0.08 −0.16 −0.22 −0.23 −0.08 −0.75 −0.75 −0.10 0.90 0.75
y (×10−3 m/s) −0.65 3.80 2.00 −1.80 3.75 −0.50 5.00 2.00 0.40 4.00
z (×10−3 m/s) −0.28 −0.05 −0.21 −0.63 −0.20 −0.25 −0.13 −0.23 −0.33 −0.35
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
Fig. 3.2.4. Sectioned scaffold and velocity vector plot at z = −0.028 m plane within bioreactor rotating about Z-axis.
This is further confirmed by the v velocity compo-
nents, which prove that their major velocity contri-
butions are particularly attributed to the fluid flow in
the Y-direction. Point 1, though, experiences a lower
magnitude of velocity, at about 0.71 × 10−3 m/s. This
indicates that cells at this location may be derived of
essential nutrient transport.
The wall shear stress plot for the scaffold again indi-
cates that higher stresses tend to occur at the outer edges
and faces, which are in direct contact with the mov-
ing fluid. It is within the scaffold that these stresses
Fig. 3.2.5. Wall shear stress range of 0–2 Pa for scaffold within biore-
actor rotating about Z-axis.
191
decrease significantly, to a value of 1 Pa and less. Upon
careful examination, it is revealed that higher stresses
here not only occur along the outer surfaces of the scaf-
fold, but especially at the areas where scaffold fibres
intersect along the outer faces of the scaffold. The aver-
age shear stresses along the scaffold fibres within the
scaffold range from 0.2 to 0.6 Pa approximately. In con-
trast, shear stresses at the fibre intersections range from
0.8 to 1.6 Pa. It is clear that the shear stresses acting on
and within the scaffold are not uniform throughout,
but are dependent on the direction of flow, too. This is
despite the relatively small dimensions of the scaffold
of approximately 5 mm length per side.
3.3. Bi-axial rotation of bioreactor with scaffold
This scenario involves positioning the scaffold at
z = −0.02 m, along the vertical centre-line of the ves-
sel. From Fig. 3.3.1, pockets of slow-moving fluid, as
indicated by the dark-blue region captured at Planes 1
and 3. However, for Plane 2, a significant portion of this
dark-blue contour seems to have filled up a large por-
tion of the chamber, primarily at the bottom-half region.
Closer examination of the corresponding velocity vec-
tor plot reveals the presence of an eddy at the bottom
of the vessel, as indicated by the recirculating vectors.
This small recirculation body corresponds in location
to the small “extension, as can be seen in Plane 2 of
192 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196

Fig. 3.3.1. Velocity contour plots captured at frontal Planes 1–3 for bioreactor with scaffold rotating bi-axially.

Fig. 3.3.1, to be extending downwards to the bottom-
right of the chamber (Fig. 3.3.2).
Fig. 3.3.3 depicts velocity vectors being deflected
upwards and to the left by the scaffold. This vector
plot also suggests that the top-right corner would espe-
cially experience relatively higher wall stresses, due to
the “intensity” of the arrows at that location. A small
eddy can also be seen slightly to the left of the scaf-
fold. The next vector plot reveals an eddy to the left
of the scaffold. Here, the faster-moving fluid flowing
along the rear of the scaffold can be seen entering the
scaffold and exiting from the front face of the scaffold,
along the positive Y direction. It is of interest to note
that from Table 3.3, all 10 locations experience sim-
ilar magnitudes of velocity, ranging from 23 × 10−3
to 29 × 10−3 m/s. This indicates that improved flow
and mixing can potentially be achieved by adopting
this scheme. We also see that the magnitudes have
increased by 1 order, indicating that this rotational
scheme improves fluid flow significantly. The data fur-

Fig. 3.3.2. Velocity contour plots captured at side Planes 4–6 for bioreactor with scaffold rotating bi-axially.
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196 193

Fig. 3.3.3. Sectioned scaffold and velocity vector plot at y = 0 m plane within bioreactor rotating bi-axially.

Table 3.3
Velocities at Points 1–10 within scaffold—bioreactor rotation about bi-axial XZ-axes
Points

1 2 3 4 5 6 7 8 9 10
Magnitude (×10−3 m/s) 28.51 28.01 28.31 26.00 24.01 27.72 28.01 27.53 27.55 23.14
x (×10−3 m/s) 0.20 0.05 0.60 0.40 0.05 −0.75 −0.50 1.25 2.50 2.50
y (×10−3 m/s) 28.50 28.00 28.30 26.00 24.00 27.70 28.00 27.50 27.40 23.00
z (×10−3 m/s) 0.50 0.75 0.38 0.05 −0.50 0.55 0.40 0.10 −1.35 −0.10

Fig. 3.3.4. Sectioned scaffold and velocity vector plot at z = −0.02 m plane within bioreactor rotating bi-axially.
194 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
Fig. 3.3.5. Wall shear stress range of 0–8 Pa for scaffold within biore-
actor rotating bi-axially.
ther indicates that the X-velocities generally do not
seem to be major constituents of the respective magni-
tudes, unlike the v or Y-velocities (Fig. 3.3.4).
Fig. 3.3.5 displays the shear stresses along surfaces
as well as within the scaffold. It is noted that shear
stresses seem to increase in the positive X direction,
indicating that the maximum wall shear stresses occurs
at the right side of the scaffold. This is verified by
Fig. 3.3.5, which isolates wall shear stresses within
the range of 0–8 Pa. It can be seen that the intersect-
ing fibres tend to experience higher stresses at points
of intersection, as highlighted by the green-coloured
streaks. The nominal shear stresses within the scaffold,
however, seem to be generally lower, at approximately
1.6–2 Pa, as indicated by the lighter-blue colour. This
is in contrast to the stresses experienced at the fibre
intersections, which fall within the range of 2.4–4.8 Pa
approximately.
In summary, the most outstanding differences relate
to the bi-axial or gyroscopically rotating bioreactor.
The velocity magnitudes at each of the locations from
Points 1 to 10 have increased significantly by manifold,
and up to one order of calculation in some cases. The
data clearly prove that bi-axial rotation of the bioreactor
results in the increase of fluid flow within the scaffold
under the studied conditions. At point 1, a 22.79 ratio
of fluid velocities is noted; by comparing the velocities
generated by the bi-axial and X rotational schemes (see
Table 3.1 and Fig. 3.4). Consequently, a ratio of 40.02
for Point 1 was also noted, by comparing the veloci-
ties due to the bi-axial and Z rotational schemes (see
Fig. 3.4. Fluid velocity magnitudes at Points 1–5 within scaffold
subjected to rotation about respective axes.
Table 3.2 and Fig. 3.5). The above results indicate a
significant enhancement of fluid velocity and mixing,
due to the bi-axial mode of bioreactor rotation. Bi-axial
rotation clearly enhances mixing of the fluid, as fluid
particles rotate under the influence of two axes. This
increases the penetrability of particles into the scaffold,
as they can enter the pores of the scaffold from more
than one direction. It is also noted that fluid mixing has
improved due to the bi-axial rotation, as compared to
uni-axial rotation. This is indicated by “breaking up” of
contours, contrary to the “uniform” contours as noted
for the uni-axial rotation cases. It must be mentioned
that bi-axial rotation, however, may result in a slight
increase of turbulent artifacts such as eddies.
The shear stresses within the scaffolds do not vary
greatly for both uni-axially rotating cases. These gen-
erally values range from 0.4 to 0.6 Pa for the proto-
type model. However, bi-axial rotation of the prototype
model reports an average shear stress value of 1.8 Pa
within the scaffold. This value of shear stress represents
those that occur within the scaffold. Findings also show
that the wall shear stresses acting along the external
Fig. 3.5. Fluid velocity magnitudes at Points 6–10 within scaffold
subjected to rotation about respective axes.
 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
edges of the scaffolds tend to be approximately three
to four times more than the internal shear stresses as
indicated. This is noted to particularly occur at the inter-
sections of scaffold struts/bars and at the circular edges
of the fibre end-faces. One reason for this is that stresses
tend to be concentrated at the edges, which are directly
exposed to the dynamic fluid.
The Reynolds number is a commonly applied non-
dimensional parameter that is applied to assess the state
of a system, where
ρVd Ωr 2
Re = or Re = (3.1)
µ υ
whereby V is the fluid velocity, d the diameter, r the
radius and Ω is the rotational velocity. The Reynolds
number represents a ratio of inertial to viscous effects
and indicates laminar, transient or turbulent flow. For
example, a flow scheme would tend towards the lam-
inar scheme due to a higher fluid viscosity, reducing
the Reynolds number of a fluid. The Reynolds num-
ber is of interest in many bioreactor systems as laminar
flow regimes are often employed. However, our sys-
tem is more complex because bi-axial rotation within
our asymmetric vessel is coupled with inlet and out-
let flows. Hence, we did not attempt to calculate Re.
The gravity and buoyancy effects are of less signifi-
cance with respect to our bioreactor system, and were
therefore neglected in our simulations.
4. Conclusions
1. Three different flow configurations were discussed
with respect to our prototype bioreactor design, pri-
marily rotation about the X (tumble), Z (spin) and
XZ (bi-axial) axes.
2. Fluid velocities and shear stresses within the scaf-
folds were studied and compared, and proved that
bi-axial rotation results in significant improvements
in terms of fluid transport through the scaffolds.
This is due to the combined effects of the rota-
tional velocity “vectors” about both axes, which
when combined, would almost certainly result in a
higher rotational velocity component, as compared
to rotation about a single axis. However, a rise in
shear forces was also reported. Greater transport
within scaffolds for example, would therefore result
195
in most cases (and not just ours) as a result of bi-
axial rotation.
3. These simulations assist in identifying critical
issues and problems, for example, in approximat-
ing the locations of recirculation zones. These
zones may potentially damage cells and inhibit
growth. Furthermore, these recirculating bodies
may impede the flow of fluid into and out of the
scaffold. The choice of flow regime is therefore of
great importance.
References
Bancroft, G., Sikavitsas, V., Mikos, A., 2003. Design of a flow perfu-
sion bioreactor system for bone tissue-engineering applications.
Tissue Eng. 9 (3), 549–554.
Boyd, E.J., Gonda, S., 1990. Mathematical modelling of the flow
field and particle motion in a rotating bioreactor at unit gravity
and microgravity. In: NASA/ASEE Summer Faculty Fellowship
Program, Johnson Space Centre, August 23.
Bursac, B., Papadaki, M., White, J., Eisenberg, S., Vunjak-
Novakovic, G., Freed, L.E., 2003. Cultivation in rotating bioreac-
tors promotes maintenance of cardiac myocyte electrophysiology
and molecular properties. Tissue Eng. 9 (6), 1243–1253.
Cherry, R.S., Papoutsakis, E.T., 1986. Hydrodynamic effects on cells
in agitated tissue culture reactors. Bioprocess Eng. 1, 29–41.
Darling, E., Athanasiou, K., 2003. Articular cartilage bioreactors and
bioprocesses. Tissue Eng. 9 (1), 9–26.
Freed, L.E., Vunjak-Novakovic, G., 2002. Spaceflight bioreactor
studies of cells and tissues. Adv. Space Biol. Med. 8, 177–195.
Freed, L.E., Vunjak-Novakovic, G., Biron, R.J., Eagles, D.B.,
Lesnoy, D.C., Barlow, S.K., Langer, R., 1994. Biodegradable
polymer scaffolds for tissue engineering. Biotechnology 12,
689–693.
Hutmacher, D.W., Risbud, M., Sittinger, M., 2004. Evolution of
computer aided design and advanced manufacturing in scaffold
research. Trends Biotechnol. 22 (7), 354–362.
Kleis, S.J., Pellis, N.R., 1995. Fluid dynamic verification experi-
ments on STS-70. In: NASA/ASEE Summer Faculty Fellowship
Program, Johnson Space Centre, August 9.
Kleis, S.J., Begley, C., Pellis, N.R., 1996. Bioreactor mass transport
studies. In: NASA/ASEE Summer Faculty Fellowship Program,
Johnson Space Centre, August 2.
Lappa, M., 2005. A CFD level-set method for soft tissue growth:
theory and fundamental equations. J. Biomech. 38, 185–190.
Levesque, M.J., Nerem, R.M., 1985. The elongation and orienta-
tion of cultured endothelial cells in response to shear stress. J.
Biomech. Eng. 107, 341–347.
Levesque, M.J., Sprague, E.A., Schwartz, C.J., Nerem, R.M., 1989.
The influence of shear stress on cultured vascular endothelial
cells: the stress response of an anchorage-dependent mammalian
cell. Biotechnol. Progr. 5, 1–8.
Martin, I., Wendt, D., Heberer, M., 2004. The role of bioreactors in
tissue engineering. Trends Biotechnol. 22 (2), 80–86.
196 H. Singh et al. / Journal of Biotechnology 119 (2005) 181–196
Olivier, A.L., Truskey, G.A., 1993. A numerical analysis of
forces exerted by laminar flow on spreading cells in a par-
allel plate flow chamber assay. Biotechnol. Bioeng. 42, 963–
973.
Papoutsakis, E.T., 1991. Fluid-mechanical damage of animal cells in
bioreactors. Trends Biotechnol. 9, 427–437.
Porter, B., Zauel, R., Stockman, H., Guldberg, R., Fyhrie, D., 2005.
3-D computational modeling of media flow through scaffolds in
a perfusion bioreactor. J. Biomech. 38, 543–549.
Pritchard, W.F., Davies, P.F., Derafshi, Z., Polacek, D.C., Tsao, R.,
Dull, R.O., Jones, S.A., Giddens, D.P., 1995. Effects of wall shear
stress and fluid recirculation on the localization of circulating
monocytes in a three-dimensional flow model. J. Biomech. 28,
1459–1469.
Raimondi, M.T., Boschetti, F., Falcone, L., Migliavacca, F., Remuzzi,
A., Dubini, G., 2004. The effect of media perfusion on three-
dimensional cultures of human chondrocytes: integration of
experimental and computational approaches. Biorheology 41,
401–410.
Redaelli, A., Boschetti, f., Inzoli, F., 1997. The assignment of velocity
profiles in finite element solutions of pulsatile flow in arteries.
Comput. Biol. Med. 27 (3), 233–247.
Sodian, R., Lemke, T., Fritsche, C., Hoerstrup, S., Fu, P., Potapov, E.,
Hausmann, H., Hetzer, R., 2002. Tissue-engineering bioreactors:
a new combined cell-seeding and perfusion system for vascular
tissue engineering. Tissue Eng. 8 (5), 863–870.
Stathopoulos, N.A., Hellums, J.D., 1985. Shear stress effects on
human embryonic kidney cells in vitro. Biotechnol. Bioeng. 27,
1021–1028.
Sucosky, P., Osorio, D.F., Brown, J.B., Neitzel, G.P., 2003. Fluid
mechanics of a spinner-flask bioreactor. Biotechnol. Bioeng. 85
(1), 34–46.
Vunjak-Novakovic, G., Freed, L.E., Biron, R.J., Langer, R., 1996.
Effects on mixing on tissue engineered cartilage. AIChE J. 42,
850–860.
Vunjak-Novakovic, G., Obradovic, B., Bursac, P., Martin, I., Langer,
R., Freed, L.E., 1998. Dynamic seeding of polymer scaffolds for
cartilage tissue engineering. Biotechnol. Progr. 14, 193–202.
Yamada, H., Takemasa, T., Yamaguchi, T., 2000. Theoretical study
of intracellular stress fiber orientation under cyclic deformation.
J. Biomech. 33, 1501–1505.
Zein, I., Hutmacher, D.W., Teoh, S.H., Tan, K.C., 2002. Poly(e-
caprolactone) scaffolds designed and fabricated by fused depo-
sition modeling. Biomaterials 23 (4), 1169–1185.