7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
On the flocculation in fiber suspensions
Maria Putkiranta*, Kalle Marjanent, Hannu Eloranta*, Tero Parssinent and Pentti Saarenrinne*
*Tampere University of Technology, Department of Energy and Process Engineering
P.O. Box 589, Tampere, FI33101, Finland
Corresponding author: maria.putkiranta@tut.fi
Tampere University of Technology, Department of Automation Science and Engineering
P.O. Box 692, Tampere, FI33101, Finland
Metso Paper Inc., P.O. Box 587, FI40101 Jyvaskyla, Finland
Keywords: fiber suspension, turbulence, flocculation, image analysis
Abstract
The production of fiber based materials, such as paper, composites and some insulation materials, involves fiber suspension
flows. Physical properties of these products are to a large extent determined by fiberflow and fiberfiber interactions. One of
the key phenomena related to fiber suspension flows is flocculation. As the fiber concentration increases, the fibers start to
form flocs. The inner structure and size of the flocs depend mainly on the fiber characteristics and the flow conditions. In
addition, chemical aids can also be used to control the flocculation. Besides forming the flocs, the flow field can also be used
to break up the fiberfiber connections, i.e. to fluidize the suspension. Thus, fluid mechanics is the key to control the
characteristics of the suspensions.
This paper presents a study on the fiber flocculation dynamics in an accelerating base flow which is directly relevant to
papermaking. Experiments are conducted with 0.5 % suspension of pine fibers in a plane convergent channel. The suspension
is fluidized at the inlet of the channel. As the turbulence decays, the fiber flocs are growing. The floc growth takes place in the
convergent section, which is expected to modify the reflocculation due the normal stress applied on the suspension by
acceleration. Pure normal stress has been suggested to be one of the most effective mechanisms to break up flocs. Experiments
are conducted to measure the streamwise development of floc scale under varying flow conditions. The floc analysis is based
on the computation of 2D spectral estimate for images taken from the flow. Details of this novel approach which is also able to
provide information on the shape of the flocs are explained. The results obtained under controlled flow conditions shed light
on the complicated balance between fluidizing and flocculating effects of the flow.
Introduction
Fiber flocculation and especially the flocculation of fibers
used in papermaking is an extensively studied subject. The
reason for this is that flocculation degrades the formation in
base paper. Poor formation, for one, can lead to impaired
strength properties and printing problems. A natural
consequence of intense flocculation is also holes in the
material, since fibers are concentrated into flocs and thus
other areas have less fibers. Generally, in all fiberbased
products flocculation causes unevenness, which, in turn,
may affect heat conduction, strengths, air permeability,
appearance etc.
The first one to recognize that fibers flocculate due to
hydrodynamic forces was Mason (1950, 1954). He also
stated that electrostatic and chemical forces are negligible in
a flocculation process and that flocculation occurs above a
critical concentration. Later on, experiments have shown
that flocculation tendency of a suspension is dependent on
numerous properties of the suspension, such as fiber length,
concentration, fiber stiffness, coarseness and curls, viscosity
of a suspending medium, interfiber friction and addition of
formation aids. See e.g. Kerekes & Schell (1995), Zhao &
Kerekes (1993) and Schmid et al. (2000). When the flow
conditions are taken into account, it can be concluded, that
fiber flocculation is a complex phenomenon.
The most often used parameter to describe the flocculation
tendency of a fiber suspension is the crowding factor N (Eq.
1) by Kerekes et al. (1985) and Soszynski (1987).
2 5CmL2
N = Cv
3 d co
Despite its generally admitted restrictions (Kerekes &
Schell 1995, Dodson 1996 among others), that it does not
take into account many of those properties listed above,
crowding factor has proved to be a useful parameter in
flocculation experiments and modelling.
By using the crowding factor, suspensions can be divided
into three regimes, which describe the type of fiber contact.
When crowding factor is less than unity the suspension is
dilute and only chance collisions of fibers may occur. As the
crowding factor increases between 1 and 60 the suspension
is semiconcentrated and forced collisions exist. Further
increase in crowding factor leads to continuous contact
between fibers as N exceeds 60. Here, coherent flocs may
appear depending on flow conditions. The value 60 as a
limit for continuous contacts was assigned in the study of
Kerekes and Schell (1992).
Fiber length is a key factor in flocculation, and an increase
in fiber length has a significant effect on flocculation.
Increase in fiber length will increase the crowding factor
rapidly, since the length is squared therein. Floc size is
defined by fiber length up to a certain degree and it is
typically 12 fiber lengths, as postulated by Kerekes and
Schell (1992). Also the lengthtodiameter ratio i.e. the
aspect ratio of fibers, Lid, is of significance. When L/d
exceeds the value 50, suspension's tendency to form
coherent flocs increases within the crowding number range
1
fibers to interlock and thus the suspension remains uniform
even at high concentrations.
It is a wellknown fact that flocculation is a fast
phenomenon. It begins immediately after the fluidizing
turbulence, e.g. after an expansion step, has decayed down
to a certain level. Within milliseconds the floes grow in
number and size. The results of Karema et al. (2001) show
that the higher the turbulence intensity, the smaller the flocs
within the turbulent flow and more effective the fluidization
of the suspension. But, as the turbulence starts to decay, the
growth rate of flocs is following the same curve up to a
saturation size, which is controlled by fiber length, channel
diameter, flow velocity and wall turbulence. It is still
ambiguous, how turbulence scales or intensity correlate
with floc properties such as size and strength in
reflocculation or do they correlate at all. Conclusions both
for and against can be found in the literature. However,
turbulence can have a twoway effect on fiber phase
behavior. Increased turbulence intensity fluidizes the flow
more effectively, but turbulence may cause more collisions
between fibers and thus also amplify flocculation in the
decaying stage. It is also clear that turbulent flocculated
suspensions will need more research, where the dynamics of
both phases is of interest and not just the floc size as in most
of the previous work.
As an alternative to turbulent and shear flows, researchers
have studied the effectiveness of elongational flows in
breaking up flocs. Pure normal stress generated by
converging channel geometries is used to resemble blade
forming conditions after acceleration in a headbox slice
(Yan & Norman 2006), and flow into sudden contractions
which can be found in mixing, screening or in flow into
turbulence generator tubes (Kerekes, 1983). Yan & Norman
studied the behavior of softwood kraft flocs in a 40 mm long
2:1 convergent channel with constant acceleration. They
used power spectrum analysis validated with manual
analysis of floc images. Flocs were stretched in mean flow
direction and compressed to cross direction in the
contraction and one fifth of them were stretched to breakage.
Analysis of power spectral densities in mean flow direction
showed that variations increased at larger wavelengths and
decreased at smaller wavelengths with increasing fiber
concentration, which indicated floc size increase. Also
James et al. (2003) studied floc breakup due to constant
extension rate. They tried to estimate the rupture stress and
found it to be much lower than that found in turbulent shear
flows. Drawbacks of their measurements were very low
fiber concentration and low number of detected flocs.
Simulations of fiber flocculation have been developed to
include fibers physical properties and provide for a study of
the role of interfiber friction, fiber stiffness and curl in
flocculation. Schmid et al. (2000) modeled fibers as chains
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
of rods and they found that in shear flow fibers do not
flocculate without interfiber friction in the absence of
attractive forces between fibers. Interfiber friction has,
besides inhibition of fiber sliding, also other functions in
flocculation of flexible fibers. Interfiber friction contributes
to deforming and storing elastic energy in collisions as
sliding is prevented. Increased fiber stiffness resulted in
simulations to more rapid floc breakup which is consistent
with earlier experiments. In addition, fiber shape had a
significant effect on flocculation tendency.
This paper presents an experimental study of flocculation
dynamics of a softwood pulp in an accelerating base flow
with decaying turbulence. The 2D spectral estimate of
images taken from the flocculated suspension flow shows
the deformation and growth of flocs. It considers the
evolution of floc scale, floc deformation, and dynamics of
fiber phase. Floc scale and deformation obtained in this
study is compared with results found in the literature. The
streamwise velocity gradient of fibers is studied in mean
flow direction as well as velocity fluctuation of the fiber
phase, i.e. fiber turbulence intensity. These flow
characteristics are then compared with the statistics of pure
water in the same geometry obtained in earlier studies.
Nomenclature
concentration (%)
diameter (m)
fiber length (m)
crowding factor ( /m3 kg)
Greek letters
co fiber coarseness (kg m'1)
x floc scale (mm)
Subsripts
m mass
v volume
Abbreviations
AR Aspect ratio
FFT Fast Fourier transform
PIV Particle Image Velocimetry
PSD Power spectral density
Experimental Facility
Experiments are conducted in a plane convergent channel
depicted in Fig. 1. Channel height is 185 mm in the
upstream and 30 mm in the downstream, so the convergence
of 6:1 is achieved. The flow channel is made of 15mmthick
Plexiglas plates, so the channel width is 15 mm. Narrowness
of the channel may effect on the behavior of the flocs by
restricting floc growth and also on the flow field around
them. In pure water channel narrowness does not seem to
have significant effect on the flow along the centerline. This
was found in a previous study (Putkiranta et al. 2009),
where the flow statistics measured in the channel center
plane is compared with results found in the literature for a
wider channel. Channel width is also restricted by the
limitations of backlight imaging through suspension of
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
actual headbox consistency.
A turbulence generator with 9 circular 0 12 mm holes is
located in the channel inlet 150 mm upstream of the
contraction. The turbulence generator is assumed to break
up flocs and fluidize the suspension before it enters the
converging section. A schematic drawing of the channel and
measurement setup is presented in Fig. 1 and a side view of
the channel with dimensions is given in Fig. 2.
affect the spectral estimation.
inlet
E 20
Outlel
Inlet
0 10 20 30
x [mm]
40 50 60
outlet
Camera
Figure 1: Measurement setup.
c P Io
f ~n
185 .1
I _  T
Figure 2: Sideview of the channel with dimensions and
streamwise measurement locations along the center line.
The channel is installed to a closed flow loop. Water
containing fibers is circulating in the loop, in which the flow
rate is controlled by the pump speed. The flow rate is
measured with a magnetic flow meter. Two different flow
rates, 2.25 1/s and 4.0 1/s, are used to study fiber flocculation
at different levels of streamwise velocity gradient and
turbulence intensity. Images are illuminated with a Cavilux
Smart diode laser. A set of 1000 doubleframe images with
PCO Sensicam is acquired in each measurement location.
Size of the imaged area is 40x50 mm2 providing a
resolution of 0.040 mm/pix in doubleframes. Lens aperture
is kept at f#8 to obtain sufficient depthoffield.
The suspension contains 0.5% per mass pine fibers with
length weighted average length 2.13 mm and coarseness
0.18 mg/m given by kajaaniFS300 analysis. Thus the
crowding number of the suspension is slightly over 60,
which is the limit of continuous contact of fibers defined by
Kerekes and Schell (1992). Example images of the
suspension in contraction inlet and outlet are shown in Fig.
3, and as can be seen, the suspension seems to have flocs
already at the inlet (x 0 mm). Results from the last two
positions (x 650mm and x 700mm) are not presented in the
results section, since the imaging area is limited in the
ydirection by the contraction walls. This is expected to
0 10 20 30 40
x [mm]
Figure 3: Images of flocs at the channel
and outlet.
50 80
contraction inlet
Image Analysis
Floc scale and shape
Backilluminated images of the fiber phase are analyzed to
evaluate the streamwise development of the fiber phase
morphology. In each measurement location the dominating
floc scale, mean floc shape and flocculation intensity are
evaluated by computing a 2D Power Spectral Density (2D
PSD) estimate for the collected images. The computation of
the PSD estimate is based on a 2D FastFourier Transform.
The original size of the images is 1280x1024 pix2. In the
preprocessing, the images are cropped to 1024x1024 pix2
and compressed by a factor of 3 utilizing a Gaussian
pyramid (Jahne, 1997). Compression is used to decrease
computational load in the FFT and to smooth small scale
variations. Thus the final size of the images is 128 x128 pix
and the resolution is 0.32 mm/pix. Further preprocessing
steps include background estimation and subtraction. The
background is estimated by compressing the each image by
a factor of 8, smoothing the resulting field with a 3x3
median filter, and interpolating back to the original
resolution. This procedure corresponds to a sliding mean
operator.
I I
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
The compressed and background compensated images are
supplied to the spectral estimation. Each image in the data
set yields a 2D PSD map. In each measurement position a
set of 500 PSD maps (images) is averaged. An example of
the 2D PSD estimate at the position x 0 mm is provided in
Fig. 4. In this plot, the horizontal and vertical axis
correspond to the wave numbers in the x and y directions.
From this map one can evaluate the 1D PSD estimate in any
direction in the image plane by taking a marginal function in
the desired direction. However, this Cartesian representation
is rather difficult to interpret. Thus, the PSD maps are
transformed to the polar coordinates. In Fig. 5 the upper plot
presents the same PSD estimate as in Fig. 4 in polar
coordinates. Below in Fig. 5 the PSD estimate at x 600mm
is given. In these plots, the horizontal axis corresponds to
the wavelength and the vertical axis corresponds to the
angle in respect to the xaxis. This plot portrays the
distribution of variance over the spatial scales as a function
orientation in the xy plane. Since the image data
represents the light transmission through fiber suspension,
the intensity variations are directly related to the fiber
consistency variations. However, this dependency is
nonlinear and in the present experiments the actual
relationship between light transmission and consistency is
not determined.
The dominating length scale of the flocs at each
measurement position is estimated by averaging the 2D PSD
map over the 0axis, i.e. by neglecting the directional
information. From the resulting mean PSD profile, the
wavelength at the location of the peak of the spectrum is
determined. It is obvious from the Fig. 5 that there is only
one peak in the spectrum, which lies in the range between
520mm.
Another way to estimate the flocculation length scale is to
consider the PSD as a probability function. The mean length
scale can be obtained simply by multiplying the normalized
spectral density with the wave number vector. Since this
operation can be performed in any direction, the mean floc
shape can be estimated by computing the mean length scale
as a function of orientation angle in the xy plane. The
results of this analysis are illustrated in Fig 6. The floc
aspect ratio is computed as the ratio between the length
scales in x and y direction. It should be noticed that the
mean length scale computed this way is considerably
smaller than that obtained from the peak of the power
spectrum. Neither of these scales should be considered as
the true floc size. Rather, these scales should be viewed as
characteristic length scales of the suspension.
The PSD is also utilized to assess the flocculation intensity.
This is done by computing the total variance from the PSD
map. In general, high variance in the image data
corresponds to large consistency variations, which, in turn,
indicates intense flocculation.
Cartesian 2D PSD
0.4 0.2 0 0.2 0.4
kx
Figure 4: Mean 2D PSD estimate at the position x 0 mm.
.I.
0225 / x00O
0 10 20 30 40 50
X [mm]
Q225 / x600
J Im ...... .  I        
30
S[mm]
Figure 5: 2D PSD of fiber images in
(x000) and near contraction outlet (x600).
contraction inlet
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Fiber phase velocity
Fiber phase velocity is obtained with a conventional PIV
analysis of 500 doubleframe images of flocculated flow at
each measurement location. Vector fields are calculated with
multipass interrogation area refinement from 128x128 to
64x64 with 50% overlap. The final resolution of the grid is
2.54 mm. The vector fields are validated with local median
criteria and upper and lower limit of pixel shift between the
frames, and the mean velocity fields and RMS fields are
calculated.
Results and Discussion
Floc scale and shape
Floc scale and shape evolution in the channel is studied with
2D PSD analysis of images. The results of the analysis are
shown in Fig. 5 based on 500 images taken at the
contraction inlet and near outlet positions, x 0 mm and x 600
mm respectively. As can be seen, the variance is higher near
the channel outlet particularly in the scales between 5 and
15 mm compared to the inlet.
Based on the 2D PSD the floc mean scale and shape can be
determined as explained earlier. Mean floc shape in
measurement locations from x 0 mm to x 600 mm is
presented in Fig. 6 at both flow rates. At the inlet, the floc
shape is rather spherical. As the flow proceeds towards the
outlet, the flocs are elongated in the flow direction, i.e. flocs
become ellipsoidal. At the higher flow rate the mean floc
scale is clearly smaller than in the lower flow rate.
Fig. 7 illustrates the streamwise development of the
dominating length scale. At both flow rates the floc scale
increases towards the outlet. The dominating floc scale is
larger than 12 fiber lengths, which was suggested by
Kerekes and Schell to be a typical floc size. In this study it
is 34.5 times fiber length, and these mean floc scales are
also in the range obtained by Yan and Norman (2006) for
bleached softwood kraft pulp.
Mean floc shape
Q 2.25
90 5 m [mm]
4 xoo
.... '.'.... xo0
... xoo
S* x505
I 30  xso
A g / x200
.x250
\ 0 x300
x400
rN fx450
S330 x550
_ x600
Mean floc shape
Q 4.00
xOOO
 x050
 x100
xlOO
 xX150
x200
 x250
x250
x300
x350
x400
 x450
 x500
x550
x6oo
Figure 6: Mean floc shape in polar plot from x 0 mm to x
600 mm at flow rates 2.25 and 4.0 1/s.
10.5
10
9.5
9
Peak floc scale
o0 Q225 I, Q
.. 0 Q400 ................................... ............
 00 0 I
..... ...... ... .......... .. .. ..... ...... ............... .................. ................... i ................
o
.................. a IQ.......... ................ ........... ... .................. ................
.0
..... ......... ......... ... .......... ......... ....... ........... ....................... .........
..... .... ...... ...
100 200 300 400 500 600 700
x [mm]
Figure 7: Streamwise development of the dominating floc
scale.
........ ......... .... .  ..... 4 ..............
.......... .......
I
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Based on the data shown in Fig. 6, the evolution of floc
aspect ratio is calculated and plotted against measurement
location in Fig. 8. Even if there is some scatter in the data
close to the outlet, the floc aspect ratio (AR) clearly
increases towards the outlet. All in all, the change in aspect
ratio is rather small compared to elongation of fluid. Based
on earlier studies, the floc breakup due to such a low
elongation is improbable. For example, Kerekes (1983)
found that even 4.5:1 elongation of flocs in sudden
contraction did not lead to breakup. He stated that an
elongation greater than 5:1 is needed to break up a floc of
longfibered pulp at 0.5 %. Aspect ratio obtained in this
study near the end of the contraction, about 1.3, agrees with
the results of Kerekes at contraction ratio 5.5 in the
centerline in the case where channel dimensions exceeded
floc dimensions. The contraction ratio in this case at x 600
mm is 3.5, which is lower, but the contraction time scales
are also different in a sudden and in a planar convergent
channel. It is to note that the elongational strain increases
and decreases very fast in sudden contractions, but in this
study it is continuously increasing, which means that flocs
can have more time to deform. In case one aims to break up
flocs by elongational strain, also the duration of elongation
may be of significance in addition to the level of applied
strain and consistency of the suspension. The importance
of suspension consistency is obvious as the observations in
this study are compared to those made by James et al.
(2003). They studied floc rupture in planar extensional flow
with constant extension rate at a very low, 0.01 %,
consistency. Such a low consistency may result to rather
loose, sparsely spaced and mechanically weak flocs, which
rupture easily even at extension rate 5 s'. The same
streamwise velocity gradient is reached in this study at x
466 mm at 2.25 1/s and at x 342 mm at the flow rate 4.0 1/s
based on potential flow curves presented in next section. In
the earlier study (Putkiranta et al. 2009), where fiber
orientation is studied in the same channel profile as in this
study, no flocs were detected at 0.02 % consistency even for
very long and rather stiff fibers. It is not evident from the
study of James et al. are the flocs formed spontaneously in
the suspension or by an external influence. In the study of
Yan et al. (2006), which is performed in more relevant
consistencies the floc aspect ratio increases from 2 to 3.5
under constant acceleration. Flocs were already elongated in
a section which resembled a headbox nozzle before they
entered the 40 mm long 2:1 contraction. The mean
streamwise velocity gradient in their study, in the range 150
to 200 s', is significantly higher than in this study. The
mean streamwise velocity gradient is discussed more in next
section, where the results of floc phase velocity
measurements are presented.
The flocculation intensity in terms of the total variance of
the 2D PSD map is plotted in Fig. 9. These results indicate
that the flocculation is stronger at the lower flow rate.
Furthermore, the flocculation increases quickly in the
beginning of the channel. After x 200mm the flocculation
intensity remains at a rather constant level.
1.3.
1.3
1.2.
1.2
1.1
1.1
1.0.
1
Floc aspect ratio
o Q225
5     0 Q400
5 ............   ................ .. .. ....... T .......................... t Q 400
10
oo o
5 .. ....... ..... a....
> .... ....... .....^. ...............^ ...... ........... ................. T.... ............. . 
... ....... ... ....... ....... . . .... ........ ..... ........ ..... ,... ......................... ..
t i i
2   i
    
So
2           
    
0
100 200 300 400
x [mm]
500 600 700
Figure 8: Streamwise development of the floc aspect ratio.
1800
1600
14001
Flocculation intensity
o Q225
0 Q400
0o
................ ..... ......... ........................ ............. ................ .............. ... .. ........
Q 0 0 0
0o
      
QO Q V Q Q O U0
0 100 200 300 400 500 600 700
x [mm]
Figure 9: Streamwise development of the flocculation
intensity.
Floc phase velocity
Fiber phase dynamics is characterized with streamwise
mean velocity profiles and turbulence intensity. The profiles
are compared to potential flow curves and PIV results
obtained earlier for pure water (Putkiranta et al. 2009). Flow
Reynolds number for pure water based on local mean
streamwise velocity and channel height is 1.6 x 105 and 2.7
x 105 at flow rates 2.25 1/s and 4.0 1/s respectively. As can
be seen in Fig. 10, velocity of fiber phase agrees well with
the potential flow curve and the PIV results. The flow
velocity is increasing significantly at the end of the channel.
The mean streamwise velocity gradient, dU/dx, is plotted in
Fig. 11 as a function of x location in the channel. Fiber
phase mean streamwise velocity gradient is also agreeing
well with the potential flow and the measured dU/dx for
pure water.
500 600 700
Figure 10: Local mean streamwise velocity as a function of
x position.
Mean streamwise velocity gradient
 40
x
3 30
o
0 100 200 300 400 500 600 700
x [mm]
Figure 11: Local mean streamwise velocity gradient as a
function of x position.
Turbulence intensity of the fiber phase in the streamwise
direction x and in the vertical direction, defined in Eq. 2, is
studied to reflect the floc size growth and turbulence decay.
U, ,, is he local RMS velocity in the x ory direction.
rms,z
Turbulence intensity of the fiber phase, shown in Fig. 12,
can be compared with the turbulence intensity measured
with pure water, shown in Fig. 13. It is to note that the
resolution of fiber phase and pure water is different, since
the imaged area in pure water PIV measurements was only
10 x 8 mm2. Thus fiber phase turbulence data can be
thought as lowpass filtered, though fiber phase turbulence
is not decaying as neatly as that of pure water. Trends and
orders of magnitude are clearly similar: turbulence intensity
is decreasing fast in the beginning of the contraction and it
is in both cases well below 4 % at the end. It seems that in
Mean flow velocity
Turbulence intensity / Fibers
x [mm]
Figure 12: Fiber phase turbulence intensity as a function of
measurement location.
Turbulence intensity/ Water
0 100 200 300 400 500 600 700
x [mm]
Figure 13: Pure water turbulence intensity as a function of
measurement location.
The mutual role of turbulence and streamwise velocity
gradient on controlling floc growth and deformation is
interesting. In this case the channel can be divided into two
sections: upstream and downstream halves. In the upstream
half, the dominant flow characteristic is turbulence. As
dU/dx becomes dominant in the downstream half,
turbulence has already decayed. One can note that the
foremost increase of floc scale and aspect ratio has already
happened in the downstream half. The dU/dx levels of this
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
fiber phase the TIx is higher than TIy at both flow rates,
whereas TIy is higher in pure water. This may result from
the different depthoffield of water PIV and fiber phase
measurements. In water PIV the depthoffield is well
defined by the laser light sheet thickness and the
measurements can be performed in the middle of the
channel. In fiber phase measurement the aim is to cover the
whole channel width by the depthoffield and thus also the
near wall effects are included in the turbulence data.
The significant decay of turbulence intensity and growth of
flocs seem to happen in same locations from x 0 mm to x
200 mm. Similar observations are made in the study of
Karema et al. (2001).
0 I I
0 100 200 300 400
x [mm]
0 Fiber phase, Q2.25
o Purewater Q2.25 ............. ...... .....................
 Potential flow, Q2.25
0 Fiberphase, Q4.00 ............. ... ..............
o Pure water, Q4.00
Potential flow, 04.00 .... ............
.... ... .. .. ...... ............ . .... ........ ..
........ I 
study are not sufficient to breakup flocs smaller, but they
may modify the flocculation by reducing the growth. Based
on this study only, it is not possible to say, whether the flocs
are near their saturation size already in the end of the
upstream half.
Conclusions
The mechanical flocculation of softwood pulp fibers is
studied in a consistency and geometry relevant to
papermaking. The flocculated suspension is subjected to
accelerating flow, which deforms the flocs more ellipsoidal
towards the channel outlet. In this study the flow is not able
to breakup flocs into smaller units.
The twodimensional Power Spectral Density (2D PSD)
estimate is applied in analyzing the images of flocculated
suspension, which enables the study of floc scale and shape.
To authors' knowledge, this is the first time, when 2D PSD
is applied to flocculated pulp suspension.
Many studies of fiber flocs have been published, but most of
them are missing the characterization of fiber phase
dynamics. This may be the first study, where PIV is applied
to measure fiber phase velocity and turbulence to
complement flocculation measurements.
The study of flocs will continue in future. The floc analysis
based on 2D PSD will be developed further and the
measurements will be repeated with shorter fibers, e.g. with
eucalyptus. Turbulence scale and intensity in the channel
can be modified by changing the dimensions of the
turbulence generator.
Acknowledgements
The authors want to acknowledge the Finnish Funding
Agency for Technology and Innovation for the financial
support within the MASIprogram.
References
Dodson C. T. J. Fiber crowding, fiber contacts, and fiber
flocculation. Tappi Journal, 79(9):211215. (1996)
James, D. F., Yogachandran, M. R., Loewen, H., Liu, H. &
Davis A. M. J. Floc Rupture in Extensional Flow. Journal of
Pulp and Paper Science 29(11):377382. (2003)
Jihne, B, Digital image processing, concepts, algorithms
and scientific applications, SpringerVerlag (1997)
Karema, H., Salmela J., Tukiainen, M. & Lepomaki H.
Prediction of paper formation by fluidisation and
reflocculation experiments. 12th Fundamental Research
Symposium, Oxford, pp. 559589. (2001)
Kerekes R. J. Pulp floc behavior in entry flow to
constrictions. Tappi Journal 66(1):8891. (1983)
Kerekes, R. J., Soszynski, R. M. & Tam Doo, P A. The
Flocculation of Pulp Fibres. 8th Fundamental Research
Symposium, Oxford, pp. 265310. (1985)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Kerekes, R. J. & Schell, C. J. Characterization of Fibre
Flocculation Regimes by a Crowding Factor. Journal of
Pulp and Paper Science, 18(1):J3238. (1992)
Kerekes, R. J. & Schell, C. J. Effects of fiber length and
coarseness on pulp flocculation. Tappi Journal,
78(2):133139. (1995)
Mason, S. G. The Flocculation of Pulp Suspensions and the
Formation of Paper. Tappi journal, 33:440444. (1950)
Mason, S. G. Fibre motion and flocculation. Tappi Journal
37(11): 494501. (1954)
Schmid, C. F., Switzer, L. H. and Klingenberg D. J.
Simulations of fiber flocculation: Effects of fiber properties
and interfiber friction. Journal of Rheology 44(4):781809.
(2000)
Soszynski, R. M. The formation and properties of coherent
floes in fibre suspensions. Ph.D. Thesis, Department of
Chemical Engineering, University of British Columbia.
(1987)
Yan, H. & Norman, B. Fibre floc behavior of softwood kraft
pulp in flowing suspensions. Nordic Pulp and Paper
Research Journal 21(1):2429. (2006)
Zhao, R. H. & Kerekes, R. J. The effect of suspending liquid
viscosity on fiber flocculation. Tappi Journal, 76(2):183188.
(1993)
