Sie sind auf Seite 1von 15

th

17 International Symposium on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2014

Analysis of Solid-Liquid Interactions in a Model Bioreactor

Irene Pieralisi1 , Lorenzo Sacchi2 , Giuseppina Montante2 , Veronica Vallini3 ,


Alessandro Paglianti1,*
1 Department of Civil, Chemical, Environmental and Material Engineering, University of
Bologna, via Terracini 28, 40131 Bologna, Italy
2 Department of Industrial Chemistry “Toso Montanari”, via Terracini 28, 40131 Bologna, Italy
3 Eridania Sadam, via della Barchetta, 1, 60035, Jesi Ancona, Italy
* correspondent author: alessandro.paglianti@unibo.it

Abstract   The   aim   of   this   work   is   to   characterize   the   two-­‐phase   fluid   dynamics   occurring   inside   a   model   stirred  
bioreactor  resembling  at  lab  scale  the  typical  design  adopted  industrially  for  the  production  of  biogas  by  fermentation  
of   agricultural   scraps.   As   for   the   solid-­‐liquid   system,   water   and   PMMA   particles   of   average   diameter   equal   to   180   µm,  
resembling  the  unfermented  crushed  scraps  industrially  employed  are  selected.  The  experiments   are  carried  out  by  
two-­‐phase  Particle  Image  Velocimetry,  which  allows  to  simultaneously  and  separately  measure  the  velocity  fields  of  
the   continuous   and   of   the   dispersed   phases.   The   unconventional   geometry   of   the   unbaffled   stirred   tank   makes   the  
measurement  of  the  flow  field  induced  by  the  impellers  and  the  evaluation  of  the  interactions  between  the  solids  and  
the  liquid  phase  particularly  challenging.  The  single-­‐phase  mean  velocity  field  is  preliminary  investigated  under  both  
transient  and  stationary  conditions  at  various  impeller  speeds.  The  two  phase  flow  measurements  are  carried  out  at  
average  solid  mass  concentration  of  0.5  %  and  at  impeller  speeds  lower  than  that  corresponding  to  the  just  suspended  
condition,   therefore  in  all  the  cases  a  deposit  of  particles   accumulated  at  the  bottom   of  the  tank,   modifying  its  shape.  
The   comparison   of   the   results   collected   under   single-­‐   and   two-­‐phase   condition   highlights   that   the   un-­‐suspended   solid  
layer   gives   rise   to   a   significant   variation   of   the   liquid   mean   and   turbulent   velocity   fields,   confirming   the   strong  
dependency   of   the   flow   field   from   the   features   of   the   tank   bottom.   These   results   are   associated   to   the   uneven  
distribution  of  particles  inside  the  reactor  and  to  consequent  very  high  concentration  of  solids  in  the  bottom  region.  
Overall,   the   analysis   of   the   results   allows   to   gain   insight   into   the   fluid   dynamic   behaviour   of   the   model   bioreactor   and  
to  suggest  possible  geometrical  modifications  aimed  at  improving  the  performances  of  the  industrial  apparatuses.  

1. Introduction

Treatment of solid-liquid suspensions in mechanically stirred reactors is one of the most common unit
operations carried out in the chemical and biochemical industry. It is well-known that the performances of
these processes are strongly dependant on the three-dimensional, turbulent, two-phase flow generated by the
mixing system, as it affects heat and mass transfer as well as the chemical reactions involved in the operation
(Derksen, 2003).
In recent years many efforts have been addressed to the exploitation of biological waste materials and to their
transformation through chemical and biochemical processes into high-energy products and biofuels
(Weiland, 2010). Nevertheless, the yields are very low, so the optimization of the fluid dynamics turns out to
be crucial in order to increase the performances of these operations. At this regard, particularly relevant are
the mean and turbulent characteristics of the flow, as they are responsible of the off-bottom suspension,
which is the basic operation mechanism of a solid–liquid mixing equipment (Khazam and Kresta, 2008;
Ayranci et al., 2012).
So far, the effects of solid particles on the continuous phase turbulence have been investigated in baffled
stirred tanks in few conditions (Nouri and Whitelaw, 1992; Guiraud et al., 1997; Micheletti and Yianneskis,
2004; Virdung and Rasmuson, 2007, 2008; Unadkat et al., 2009; Montante et al., 2012). Usually it has been
found that small particles tend to dampen turbulence and large particles tend to augment it and that the extent
of this phenomenon is related to the ratio between the particle size and a characteristic turbulence length
(Crowe et al., 1998). Yet, general predictive equations cannot be applied, as the flow features are strictly
related to the characteristics of the mixing system and of the solid phase. Therefore, further efforts are
demanded to improve the current model capability and the mixing equipment design.
In particular, an interesting category of stirred reactors, which has been investigated at a lesser extent, is
represented by unbaffled tanks. Typically, in these tanks a mainly tangential flow develops, and a definite

-1-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

vortex expands from the free surface, involving a big entrainment of gas that can even damage the process.
On the contrary, in baffled tanks the vortex formation is inhibited and the circulation of the fluid along
tangential patterns is weaker. This leads to a stronger axial flow which improves the mixing efficiency, thus
making these equipments more suitable for many industrial applications. Nevertheless, in some cases the use
of unbaffled tanks may be convenient, as for example they give rise to higher fluid–particle mass transfer
rates for a given power consumption (Grisafi et al., 1994; Yoshida et al., 2008). Moreover, when the process
involves a very viscous fluid or a two-phase mixture at high solid concentration, baffles can actually worsen
the mixer performance, as they give rise to dead zone where the fluid and the solid particles tend to
accumulate (Nagata, 1975; Lamberto et al., 1996). The use of unbaffled tanks is also appropriate in food and
pharmaceutical industries, where vessel cleanness is a fundamental requirement of the processes (Assirelli et
al., 2008) and in biological applications where cell damage is to be avoided (Aloi and Cherry, 1996). The
mean and fluctuating single-phase flow fields generated in an unbaffled reactor, were deeply investigated by
many authors, as Armenante et al. (1996) and Alcamo et al. (1999), using respectively a PBT and a Rushton
turbine. As far as solid-liquid systems are considered, suspension characteristics as the just suspended
impeller speed and the particles distribution inside the tank have been accurately measured and predicted
(Shan et al., 2008; Tamburini et al., 2009; Brucato et al., 2010), whereas detailed experimental
characterizations of the turbulent two-phase flow in unbaffled reactors cannot be found in the literature.
In this work, an unbaffled tank of unconventional geometry is investigated, with the aim of widening the
current set of available data on the fluid dynamic characteristics of this type of industrial digesters and of
improving their energetic efficiency by the optimization of geometrical features and working conditions.

2. Experimental

The investigated stirred tank, shown in Fig. 1, was obtained from the scale-down of a typical biomethanation
digester. It consists of a partially filled, flat bottomed, unbaffled cylindrical vessel of diameter, T, equal to
0.49 m and height, H, equal to T/2. To minimise refraction effects at the curved surface, the vessel was
placed inside a square tank filled with the working liquid. The mechanical agitation was provided by two
identical Lightnin A310 impellers of diameter, D, equal to 0.2 T, the lower of which was placed at an off-
bottom clearance, C1, equal to 0.08 T and at distance from the upper impeller, C2, equal to 0.18 T. The
impellers were mounted on a shaft coaxially with respect to the vessel axis.

Fig. 1 – Geometrical configuration of the stirred reactor.

-2-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

As the continuous phase, demineralised water was adopted, and its level inside the reactor, h, was set at 0.48
T (23.5 cm) for most of the experiments. For the solid phase, a close reproduction of the agricultural scraps
physical properties was hardly achievable, due to their wide size range and uneven shape distribution, which
changes during the fermentation process (O'Neil, 1985); therefore a similar density of the unfermented
crushed scraps and a narrow size range particles were selected as a tentative approximation, resulting in the
choice of Poly Methyl Methacrylate beads of density, ρs, equal to 1200 kg/m3 and average diameter, dp, of
180 µm (size range 150-210 µm). The solid-liquid suspension was analyzed for an average solid mass
fraction, ωs, of 0.5% at various impeller rotational speeds, N, ranging from 100 rpm to a maximum of 300
rpm. This set of values was chosen because it includes the typical working conditions of the model digester,
as evaluated by the scale-down criterion of constant power per unit volume.
The just suspended speed, Njs, calculated from the Zwietering equation (Zwietering, 1958), is 380 rpm, and
was never reached during the experiments. This led to the formation of an uneven deposit of particles at the
tank bottom, and to an actual concentration of suspended solid inside the tank lower than the overall value of
0.5 %. In these conditions, an acceptable optical attenuation of the laser sheet was achieved across the whole
measurement plane up to about N = 230 rpm. The impeller Reynolds number, Re, based on the physical
properties of water, ranged between the values 16000 and 48000, therefore in all the cases investigated the
flow regime, close to the impellers, was fully turbulent.
The origin of the cylindrical reference system was located at the centre of the vessel bottom; all dimensions
and coordinates were normalized with the tank diameter T (r/T, z/T); all mean velocities and turbulence
levels were normalized with the tip blade velocity Vtip = πND. The mean axial velocity, U/Vtip, was positive
if directed upwards and the mean radial velocity, V/Vtip , was positive if directed towards the vessel wall.

2.1 The two-phase PIV technique


The ensemble-averaged turbulent velocity fields of the continuous and the dispersed phases were determined
by a two-phase PIV system (Montante et al., 2012), both in the whole and in selected portions of a
diametrical vertical plane of the vessel, in order to improve the vectors resolution and to account for the
strong variations of the flow moving from the impeller region towards the vessel wall. A sketch of the
different measurement areas analyzed, is shown in Fig. 2.

Fig. 2 – Sketch of the measurement areas investigated in the experiments.

The experimental system included a Litron Nd:YAG laser, emitting light at 532 nm with a maximum
frequency of 15 Hz and energy equal to 65mJ, and two identical HiSense MK II, 1344×1024 pixels CCD
cameras. The continuous phase was seeded by polymeric particles coated with fluorescent Rhodamine B
emitting light at the wavelength of λ=590 nm, while the solid particles had the same emission wavelength as
the laser light. Although the two PIV cameras were provided with appropriate light filters to allow phase
signals separation, their behaviour was not ideal. Thus, the cameras lens aperture was adjusted, to ensure that
each camera could capture the light scattered by only one phase, and at the same time it could detect enough

-3-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

light to provide reliable results. The compromise between the two requirements was achieved by setting the
lens aperture at 8 and at 22 respectively for the liquid and the solid-phase detector camera.
PIV analysis was carried out through the use of a cross-correlation algorithm, applied on interrogation areas
(IA) of size 32 x 32 pixel, with 50% overlap between each others. As a result, the technique produced
equally spaced velocity vectors located in each point of the grid, and allowed the reconstruction of the flow
field with a spatial resolution of 3.5 mm for the area A, and of 1.9 mm for cases B and C. The instantaneous
vectors obtained via cross-correlation underwent validation algorithms based on the evaluation of the peak
heights in the correlation plane and on the velocity magnitude (Montante et al., 2012); the measurements
region was then limited to the zone where at least 90% of the vectors passed the validation step.
The statistical convergence of data was checked on both mean velocities and turbulence levels, and the
number of image pairs necessary to obtain reliable results was set to 1000 for each experiment. Nevertheless,
when transient phenomena occurring inside the flow were investigated, a lower number of image couples
was analyzed.
The laser sheet thickness, zl , exhibited a very important impact on the results, due to the strong three-
dimensionality of the flow, which typically characterizes unbaffled reactors. Usually, when a planar flow is
analyzed, the laser sheet is required to be as narrow as possible in the area of interest, in order to obtain
accurate results (Raffel et al., 1998). Nevertheless, in this work since a significant tangential velocity
component orthogonal to the laser sheet takes place, care had to be devoted in the selection of zl. As shown
by Fig. 3 a-b, the laser thickness modifies significantly the number of validated vectors in the analysed
section. The number of valid velocity vectors experiences a sudden drop precisely where the location of the
minimum laser sheet thickness occurs: (a) in the middle of the measurement plane, (b) close to the impeller.
As the fluid has a strong tangential velocity, the system isn't able to capture the planar displacement of the
seeding particles, and the well known loss-of-pairs phenomenon occurs, resulting in the production of
counterfeit velocity vectors, which are discarded by the validation algorithm subsequently applied. It is
worth noting that these images have been collected at N = 300 rpm and with a pulse delay, ∆t, of 600 µs.

Fig. 3 - Validation maps of the PIV row data collected at N = 300 rpm, number of image pairs equal to 300 . Results
obtained setting the location of the minimum laser sheet thickness: (a) in the middle of the measurement plane, (b) close
to the impeller.

It is well-know that, to prevent signal drop-out when the flow field is strongly three-dimensional, the time
between pulses, Δt, and the thickness of the light-sheet, zl, have to be chosen in relation to the out-of-plane
velocity component, vz, so that the following relationship is satisfied (Gomez et al., 2010):

-4-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

In the present conditions, (N= 300 rpm; assuming Vz, max = 0.5 Vtip), the criterion suggests to use a light
sheet thickness greater than 2 mm, assuming a ∆t equal to 600 µs; or a ∆t smaller than 320 µs, considering a
zl equal to 1 mm. Therefore, to detect the real displacement of the particles within the vertical plane, the light
sheet geometry, and the time delay between the laser pulses (∆t) have been accommodated in turn.
In the first case, to hinder the tangential egress of particles, a traditional value of laser sheet thickness (zl = 1
mm) was set close to the impeller discharge stream, whereas the time delay between the laser pulses, ∆t, was
enormously decreased, down to the value of 30 µs. The vector field and mean velocity magnitude map
obtained at N = 300 rpm, are shown in Fig. 4: good results are displayed in the area close to the impellers,
whereas relevant errors arise moving towards the vessel wall. This behaviour is due to the high velocity
gradients occurring in the present reactors because of the extremely low value of the D/T ratio. As a result,
multiple acquisitions should be performed, increasing the time between pulses when moving the
measurement zone from the impeller toward the vessel wall. As a second possible set-up was tested: the light
sheet was thickened up to about 7 mm in the area close to the impeller, while the ∆t was kept constant to the
initial value of 600 µs: a flow field very similar to the one shown in Fig. 4 was obtained.
Finally, both the solutions were considered inappropriate to describe the fluid dynamics in the whole vertical
plane of the reactor, as they guaranteed good outcomes just in the region close to the impeller.

Fig. 4 - Vector field and mean velocity magnitude map obtained at N=300 rpm, using the following set-up:
∆t = 30 µs, zl= 1 mm in the impeller area.

Therefore, all the following experiments were carried out using the traditional laser configuration, which
assumes that the location of the minimum light-sheet thickness is located at the centre of the reactor; while
for each set of measurement, the time delay is accommodated according to the rotational impeller speed, and
to the extent and location of the area analyzed.
To set the most suitable time delay for each studied area (A, B and C), the method proposed by Gomez et al.
(2010) was applied. Multiple set of PIV measurements were carried out using different values of ∆t, and after
data processing, vertical profiles of the axial component of the velocity were compared. The variation
between each couple of measurements was quantified in terms of the root mean square (rms) deviation.
For a rotational speed of 300 rpm, data were collected in the area A at various ∆t (1500-1000-750-500 µs),
and for a vertical profile located in the impeller jet at r/T = 0.15, the minimum of the rms was found at ∆t =
850 µs. In all the subsequent experiments, this value was used as the starting point to scale the pulse delay
according to the impeller velocity investigated.

-5-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

3. Results and discussion


3.1. The single-phase transient and steady flow

A preliminary investigation on the single-phase flow conditions has shown that the fluid dynamic behaviour
of the model reactor is characterized by very long initial transient and that significant flow field variations
take place during the time interval between the agitation onset and the steady state achievement (Montante
and Paglianti, 2014). This reveals the strong complexity of the flow generated by the interaction of the axial
thrust of the turbine, with the overall tangential movement of the fluid due to the absence of baffles inside
the reactor.
Fig. 5 a-c shows the velocity fields obtained at N=300 rpm after 1, 10 and 60 minutes from the agitation
onset; in the last image the flow is considered stationary, because no more appreciable changes were detected
after that time.

Fig. 5 – Liquid velocity vector plots and colour maps at N=300 rpm and single-phase condition, after (a) 1 minute, (b)
10 minutes, (c) 60 minutes from the agitation onset.

As soon as the mechanical stirring starts (Figure 5 a), the circulation of the fluid spreads throughout the
whole section of the reactor, while the flow field close to the two turbines is strongly axial. As time increases
(Figure 5 b), the impellers discharge streams start to tilt radially, until after approximately one hour from the
agitation onset (Figure 5 c), the lower jet is completely lifted from the bottom. In this condition, the motion
of the fluid within the reactor is confined just in the region close to the impellers, whereas moving away
from this central area, the velocities tend to gradually decrease, to the point that near the tank wall the flow

-6-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

seems to be stagnant. Hence at this speed (N=300 rpm), the mixing system seems to be more efficient during
the transient regime rather than at steady state, as in the latter case the fluid circulates only within a little
portion of the tank. To improve the performances of the reactor an intermittent use of the impellers might be
considered. In addition, the present experimental observations on the time span of the process confirm that in
order to collect meaningful data, particular care must be devoted to identify the characteristic time in
unbaffled stirred tanks, since significant deviations with respect to standard geometry can be found, as
observed also in previous investigations of tall unbaffled stirred tanks (Pinelli et al., 2001; Derksen, 2006).
The stationary flow dynamics occurring inside the reactor were investigated at other rotational speeds. As
can be observed in Fig. 6 a-b, an interesting result is obtained for N = 100 rpm: at this condition the single-
phase system can assume alternatively two different configurations. In the first image, the mean velocity
field reveals the same features as the one found at steady state at 300 rpm (Fig, 5c), as the lifted impeller jet
and the recirculation zone at the bottom of the tank can be clearly identified. However, Fig. 6b shows a
rather different flow field, as the lower impeller discharge stream reaches the bottom, and here the fluid does
not experience any suction towards the centre, but simply heads for the wall.

Fig. 6 – Different layouts of the single-phase flow field and colour map obtained at N=100 rpm.

Both the results obtained at N=100 rpm were stable for the time necessary to complete the data acquisition,
and no intermittent mode-switching was observed, as the distributions of the instantaneous velocities were
not bimodal for any set of measurement.

-7-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

This analysis suggests that in unbaffled stirred reactors the stationary flow field results from the complex
interaction of multiple fluid-structures, which assume in turn different relevance and strength depending on
the impeller speed and on the geometrical characteristics of the system.
Therefore at N=100 rpm, two stable working conditions are possible, besides the action of the impeller, some
other effects may become predominant, determining the double-mode behaviour just reported. Ayranci et al.
(2012), who studied the two-phase flow generated in a reactor stirred by A310 impellers, assessed that a
strong rotational flow appears below the turbine when the impeller clearance is very low, causing particles to
drop out in the centre of the vessel; whereas if there is sufficient distance between the impeller and the tank
bottom, the tangential velocity decreases and the swirl stops to interfere with the main flow pattern. In the
present study, a very low value of the clearance is considered, therefore a similar swirling mechanism may
occur at the bottom, driving the fluid towards the centre of the reactor. The phenomenon starts to be relevant
at 100 rpm, giving rise to the double-mode behaviour reported in Fig. 6 a-b, whereas at 300 rpm it is fully
established, generating a stationary flow that heads always for the centre at the bottom of the tank (Fig. 5c).
Further analysis on the tangential velocities of the fluid should be carried out to confirm this hypothesis.

3.2. The two-phase flow: effect of particles on the liquid flow field

The goal of the two-phase analysis was to identify the effect of the suspended and the unsuspended solids on
the overall mixing characteristics of the model bioreactor and to describe the interactions between the two
phases.
In order to identify the PIV measurements conditions and extent, a preliminary visual analysis of the tank
was carried out at different rotational speeds. In Fig. 7 a picture of the bottom of the reactor operating after
the steady state attainment at N =100 rpm is shown: the formation of a solid annulus located far from the
impeller discharge stream can be observed, as the system was far away from its just suspended condition (Njs
= 380 rpm). Based on these preliminary observations, particular care was devoted to the analysis of two
zones by the PIV measurements (Fig. 2), the former located close to the lower impeller, while the latter at the
same height but near the vessel wall.

Fig. 7 – Picture of the bottom of the reactor at N = 100 rpm after the steady state attainment.

Fig. 8 a-b shows the liquid phase mean-averaged velocity fields generated in the two selected portions of the
diametrical vertical plane at N=100 rpm. As can be observed, the fluid is axially pumped by the turbines
(Fig. 8a), reaches the bottom of the vessel, slides on the surface towards the wall, and at r/T = 0.35, where
the solid particles annulus is located, rises up and runs toward the upper part of the vessel (Fig. 8b). The flow
field of the solid phase resulted to be almost identical to the liquid one, hence is not reported here.

-8-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

Comparing these experimental outcomes with those shown in Fig. 6 a-b, and especially observing the region
close to the bottom of the tank, some observations can be risen about the influence of the solid on the liquid
fluid dynamics. The results indicate that the dispersed phase seems to hinder the suction of the fluid towards
the centre, and to enhance the attitude of the impeller jet to adhere and stick to the bottom of the tank. This
means that in this condition the effects of the rotational flow occurring below the turbine are overcome by
some other mechanism, and that the presence of the solid phase plays a stabilizing role in the system fluid
dynamics, as the double-mode behaviour detected in the single-phase condition at the same speed, here is no
longer reported.
In particular, the observed qualitative mismatch between the single and the two-phase conditions may be a
consequence of the attraction of the impeller discharge stream exerted by the vessel bottom, which geometry
is modified by the presence of the unsuspended solid particles.
The 300 rpm flow field couldn't be investigated, because at that speed the relevant amount of solid particles
suspended in the flow determined a critical attenuation of the laser light-sheet, and the PIV measurements are
not viable.

Fig. 8 – Vector fields and colour maps of the liquid-phase at N=100 rpm: (a) impeller region; (b) solid deposit region.

In order to perform a more detailed assessment of the influence of the particles on the liquid flow field, mean
radial and axial velocity profiles obtained in both the single and the two-phase systems (labelled as

-9-
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

continuous-phase), are plotted in Fig. 9 a-b. The curves were obtained for two different agitation conditions
(N=100 rpm, and N= 170 rpm), and at a fixed radial location, r/T, equal to 0.086, therefore they well
represent the behaviour of the fluid in the turbine region. The second layout of the 100 rpm two-phase flow
(Fig. 6b) field was chosen to perform the comparison. As can be observed, differences between the axial and
radial profiles are apparent: at both the investigated rotational speeds. These differences are clearly
associated to the variation of the flow pattern of the liquid phase when the dispersed phase is added in the
vessel.

Fig. 9– Mean radial and axial velocity profiles at r/T = 0.086, obtained at: a) N=100 rpm; b) N=170 rpm.

To better appreciate the extent of the changes produced in the flow by the particles, velocity fluctuation
profiles at the fixed radial locations of r/T = 0.086 are shown in Figs. 10-11.

Fig. 10 – Profiles of liquid radial r.m.s. velocity at r/T = 0.086, obtained at: a) N=100 rpm; b) N=170 rpm.

Overall, the radial and axial velocity fluctuations of the single-phase flow appear of the same order of
magnitude, although the radial r.m.s. component experiences a higher peak around z/T=0.08, where the fluid
meets the edge of the blade.

- 10 -
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

At any axial location, the single-phase velocity fluctuations assume values lower than those collected in the
two-phase system, meaning that the particles significantly affect also the turbulent velocity field, which is
uncommon for dilute systems. This might be due to the fact that, although the overall content of solid inside
the reactor is very low, its distribution is not spatially uniform, and in the impeller jet area a very high
concentration of particles accumulates and interacts with the liquid flow. Further investigations on the
concentration levels inside the tank should be carried out to validate this hypothesis.

Fig. 11 – Profiles of liquid axial r.m.s. velocity at r/T = 0.086, obtained at: a) N=100 rpm; b) N=170 rpm.

Similar conclusions can be derived from the maps of the axial r.m.s. velocities shown in Fig. 12 a-b, and
obtained at 170 rpm in the single- and the two-phase conditions. In the whole region displayed, and in
particular in the impeller discharge stream, the turbulence levels of the continuous-phase are the highest,
underlying again the strong influence exerted by the particles on the fluctuating liquid flow.

Fig. 12 – Axial liquid r.m.s. velocity maps obtained at N=170 rpm: (a) single-phase flow; (b) two-phase system.

- 11 -
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

3.3. Effect of the shape of the tank bottom on the liquid flow field
The mismatch reported between the main flow patterns of the single- and the continuous-phase in two-phase
system, was ascribed so far to the interaction between the lower impeller jet and the tank bottom which shape
is modified by the presence of particles. To validate the hypothesis, a further analysis was carried out. A
Perspex solid annulus, with geometrical features similar to the deposit of unsuspended particles occurring in
the two-phase system, was built and set to the bottom of the reactor (Fig. 13). In this conditions, the vessel
was filled with liquid and the single phase vector field was investigated at 100 and 300 rpm.

Fig. 13 – Layout of the reactor equipped with the Perspex solid annulus at the bottom.

As displayed by Fig. 14 a-b, a similar mean flow field is obtained for both the rotational speeds: the impeller
discharge stream cannot lift, and the fluid runs towards the wall sticking to the bottom without any backward
suction. These results clearly differ from those attained with the flat-bottomed reactor (Fig. 5-6a), suggesting
that the shape of the tank basis affects remarkably the fluid dynamics of the system, and confirming that in
the two-phase condition, the attitude of the impeller jet to adhere to the bottom, is likely to be associated with
the deposit of unsuspended particles.
This phenomenon, which implies the deflection of a turbulent jet towards a solid boundary, is well-known as
the Coanda effect, and has been seldom observed in stirred tanks (e.g. Montante et al., 2001).
The double-mode behaviour of the single-phase flow, which was detected in the flat-bottomed reactor at 100
rpm ,here is suppressed by the presence of the solid annulus, which therefore plays a stabilizing role, as it
happened with the deposit of particles in the two-phase system.

- 12 -
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

Fig. 14 – Liquid vector fields and velocity magnitude colour maps obtained in the single-phase condition after the
insertion of the solid annuls at the bottom of the tank, at: a) N=100 rpm, b) N=300 rpm.

4. Conclusions

The results collected in this work highlight that the fluid dynamic behaviour of stirred digesters typically
adopted for the biomethanation of organic wastes is significantly different from that of the common stirred
tanks employed in the process and chemical industries, due to their specific geometrical features. The smaller
liquid height with respect to the tank diameter and the absence of baffles make the fluid mixing
characteristics significantly different with respect to the case of conventional geometries.
Even the PIV measurements resulted particularly challenging in these conditions, therefore particular care
for the set-up was required, involving the adjustment of parameters like: the time between pulses, the camera
lens aperture and the laser sheet thickness.

- 13 -
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

A deep investigation on the single-phase flow conditions of the reactor has shown that the time necessary for
the steady state attainment can increase up to one order of magnitude with respect to the standard stirred
tanks, and that the fluid dynamic characteristics change dramatically during the time span of the process. In
addition, the analysis of the single-phase flow has outlined another interesting aspect: at 100 rpm the system
can assume alternatively two different configurations, which are both stable, as no intermittent mode
switching is observed between them.
Under the solid-liquid mixing conditions typically adopted at industrial scale, the unsuspended solid on the
bottom of the tank changes the vessel shape, thus giving rise to a significant modification of the continuous-
phase mean velocities and turbulent velocity fluctuations.
Additional experiments have been carried out in the single-phase condition with the insertion of a solid
annulus at the bottom of the tank, and the outcomes confirm the hypothesis about the influence exerted on
the flow by the shape of the reactor bottom.
The PIV results discussed in this work can be usefully adopted for improving the energetic efficiency of the
industrial digesters by the optimization of the geometrical features and of the working conditions.

References

Alcamo R , Micale G, Grisafi F, Brucato A, Ciofalo M (2005) Large-ddy simulation of turbulent flow in an
unbaffled stirred tank driven by a Rushton turbine. Chemical Engineering Science 60: 2303–2316.
Aloi LE, Cherry RS (1996) Cellular response to agitation characterized by energy dissipation at the impeller
tip. Chem. Eng. Sci. 51 (9): 1523–1529.
Armenante PM and Chou CC (1996) Velocity profiles in a baffled vessel with single or double pitched-blade
turbines. A.I.Ch.E.J. 42: 42-54.
Assirelli M, Bujalski W, Eaglesham A, Nienow AW (2008) Macro- and micromixing studies in an unbaffled
vessel agitated by a Rushton turbine. Chem. Eng. Sci. 63: 35-46.
Ayranci I, Machado MB, Madej AM, Derksen JJ, Nobes DS, Kresta SM (2012) Effect of geometry on the
mechanisms for off-bottom solids suspension in a stirred tank. Chemical Engineering Science 79: 163-
176.
Brucato A, Cipollina A, Micale G, Scargiali F, Tamburini A (2010) Particle suspension in top-covered
unbaffled tanks. Chemical Engineering Science 65: 3001–3008.
Crowe C, Sommerfeld M, Tsuji Y (1998) Multiphase Flows with Droplets and Particles. CRC Press.
Dantec Dynamics (2011) DynamicStudio v3.20 User’s Guide.
Derksen JJ (2003) Numerical simulation of solids suspension in a stirred tank. AIChE Journal 49: 2700-
2714.
Derksen JJ (2006) Long-time solids suspension simulations by means of a large-eddy approach. Chem. Eng.
Res. Des. 84: 38-46.
Gómez C, Bennington CP J, Taghipour F (2010) Investigation of the Flow Field in a Rectangular Vessel
Equipped With a Side-Entering Agitator. Journal of Fluids Engineering 132.
Grisafi F, Brucato A, Rizzuti L (1994) Solid–liquid mass transfer coefficients in mixing tanks: influence of
side wall roughness. IChemE Symposium Series 136: 571-578.
Guiraud P, Costes J, Bertrand J (1997) Local measurements of fluid and particle velocities in a stirred
suspension. Chemical Engineering Journal 68: 75-86.
Khazam O, Kresta SM (2008) Mechanisms of solids drawdown in stirred tanks. Canadian Journal of
Chemical Engineering 86: 622-634.
Kowalczyk A, Harnisch E, Schwede S, Gerber M, Span R (2013) Different mixing modes for biogas plants
using energy crops. Appl. Ener. 112, 465-472.
Lamberto DJ, Muzzio FJ, Swanson PD, Tonkovich AL (1996) Using time dependent RPM to enhance
mixing in stirred vessels. Chem. Eng. Sci. 51: 733-741.
Micheletti M, Yianneskis M (2004) Study of fluid velocity characteristics in stirred solid–liquid suspensions
with a refractive index matching technique. Proceedings of the Institution of Mechanical Engineers 214:
191–204.
Montante G, Lee KC, Brucato A, Yianneskis M (2001) Experiments and predictions of the transition of the
flow pattern with impeller clearance in stirred tanks. Comp. Chem. Eng. 25: 729-735.

- 14 -
 
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014

Montante G, Paglianti A, Magelli F (2012) Analysis of dilute solid-liquid suspensions in turbulent stirred
tanks. Chem. Eng. Res. Des. 90: 1448-1456.
Montante. G., Paglianti, A., 2014. Fluid dynamics characterization of a stirred model bio-methanation
digester, Chem. Eng. Res. Des. DOI: 10.1016/j.cherd.2014.05.003.
Nagata S (1975) Mixing: Principle and Applications. Wiley, New York.
Nouri JM, Whitelaw JH (1992) Particle velocity characteristics of dilute to moderately dense suspension
flows in stirred reactors. International Journal of Multiphase Flow 18: 21-33.
O'Neil DJ (1985) Rheology and mass/heat transfer aspects of anaerobic reactor design. Biomass 8: 205-216.
Pinelli D, Nocentini M, Magelli F (2001) Solids distribution in stirred slurry reactors: influence of some
mixer configurations and limits of the applicability of a simple model for predictions. Chem. Eng. Comm.
188: 91-107.
Raffel M, Willert C, Kompenhans J (1998) Particle Image Velocimetry: A Practical Guide. Springer.
Shan X, Yu G, Yang C, Mao ZS, Zhang W (2008) Numerical Simulation of Liquid-Solid Flow in an
unbaffled Stirred Tank with a Pitched-Blade Turbine Downflow. Ind. Eng. Chem. Res. 47: 2926-2940.
Tamburini A, Gentile L, Cipollina A, Micale G, Brucato A (2009) Experimental investigations of dilute
solid-liquid suspension in an unbaffled stirred vessel by a novel pulsed laser based image analysis
technique. Chemical Engineering Transactions, 17: 531-536.
Virdung T, Rasmuson A (2007) Measurements of continuous phase velocities in solid–liquid flow at
elevated
concentrationsi n a stirred vessel using LDV. Transactions of IChemE Part A: Chemical Engineering
Research and Design 85: 193–200.
Virdung T, Rasmuson A (2008) Solid–liquid flow at dilute concentrations in an axially stirred vessel
investigated using particle image velocimetry. Chemical Engineering Communications 195: 18-34.
Weiland P (2010) Biogas production: current state and perspectives. Appl. Microbiol. Biotechnol. 85: 849-
860.
Yoshida M, Kimura A, Yamagiwa K, Ohkawa A, Tezura S (2008) Movement of solid particles on and off
bottom of an unbaffled vessel agitated by unsteadily forward–reverse rotating impeller. J. Fluid Sci.
Technol. 3 (2): 282–291.
Zwietering TN (1958) Suspending of solid particles in liquid by agitators. Chemical Engineering Science 8:
244-253.

- 15 -
 

Das könnte Ihnen auch gefallen