Sie sind auf Seite 1von 9

Chemical Engineering Journal 231 (2013) 163171

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal


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

The inuence of porous structure and biolm on the hydrodynamics


of two types of trickle lters
Pang Haoran, Letisse Valrie, Paul Etienne, Hbrard Gilles
Universit de Toulouse, INSA, UPS, INP, Laboratoire dIngnierie des Systmes Biologiques et des Procds (LISBP), 135 avenue de Rangueil, 31077 Toulouse, France
INRA, UMRA792, LISBP, 31400 Toulouse, France
CNRS, UMR5504, LISBP, 31400 Toulouse, France

h i g h l i g h t s
 Concrete-Block media with/without biolm in MSB.
 Majority of liquid held by particles with biolm.
 Biolm reduces dynamic fractions and delays liquid release.
 Biolm lengthens liquid residence time and increases active liquid volume in lter.
 Biolm improves liquid mixing in lter and reduces liquid lm thickness.

a r t i c l e

i n f o

Article history:
Received 10 March 2013
Received in revised form 25 June 2013
Accepted 28 June 2013
Available online 5 July 2013
Keywords:
Hydrodynamic
Liquid holdup
Porous structure
Liquid lm, RTD

a b s t r a c t
The hydrodynamic behaviors of a Multi-Section Bioreactor (MSB) and a Trickle Fixed-bed Column (TFC)
packed with porous concrete block medium were investigated in terms of liquid holdup and RTD modeling in regimes with and without biolm. It was found liquid static holdup was always higher than
the dynamic holdup due to the porous structure of medium, rather than the conguration of bioreactor.
Dynamic holdup increased when the hydraulic load increased. The presence of biolm retarded the liquid
release through the bioreactor; however, it enhanced the mass diffusion between static and dynamic
fractions. The RTD measurements and modeling showed that the presence of biolm lengthened the
liquid residence time, and it promoted the mass diffusion between the liquid and biolm. Liquid lm
thickness estimation implied that the biolm reduced the liquid lm thickness resulting both from the
increase of surface area and reduction of dynamic holdup volume.
2013 Elsevier B.V. All rights reserved.

1. Introduction
Trickle Filter (TF) technology is widely used in small wastewater treatment plants because it is robust and relatively uncomplicated with low specic energy consumption. In a TF, complex
mechanisms take place which are highly dependent on hydraulics.
To improve and operate such TF technology, it is indispensible to
understand the physical parameters which affect the hydraulics
in order to take accurately these effects into account through software modeling such as Hydromantis GPS-X. It is important for
example, to assess the contact time between the polluted stream
and the biolm responsible for pollutant biodegradation; this

Corresponding author at: Universit de Toulouse, INSA, UPS, INP, Laboratoire


dIngnierie des Systmes Biologiques et des Procds (LISBP), 135 avenue de
Rangueil, 31077 Toulouse, France. Tel: +33 (0)5 61 55 97 89; fax: +33 (0)5 61 55 97
60.
E-mail address: Gilles.Hebrard@insa-toulouse.fr (H. Gilles).
1385-8947/$ - see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cej.2013.06.115

contact time is directly dependent on liquid holdup and liquid lm


thickness.
With conventional TF, the design and reliable scale-up are
highly dependent on hydrodynamic performance and on the transport phenomena imposed by the type of internal packing and its
geometry [1]. The hydrodynamic behavior of certain types of medium has been widely investigated by other researchers [27]. However, a number of natural and man-made non-spherical porous
medium with biolm are too complicated to be well investigated
even appropriate boundary conditions were settled in some studies [8]. The effect of biolm on hydrodynamics in a TF has seldom
been reported [5,9]; most reports have assumed the biolm to be
homogeneous along the packing bed, but even so, they have not
quantied the biolm.
Usually, total liquid holdup can be divided into two aspects:
static and dynamic [10,11]. Static holdup means the fraction of liquid that is retained within and around the particles by capillary
forces after complete draining. It depends on the ow rate and also

164

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

Nomenclature
hLt
hLS
hLd

ucb
e
e0
dp

r2
qL
qparticle
dL
bd
bs
bt

s
h
hcb
Fad
TFC
MSB

total liquid holdup (m3)


liquid static holdup (m3)
liquid dynamic holdup (m3)
apparent packing-bed void fraction ()
total packing-bed void fraction ()
particle porosity ()
equivalent sphere diameter (cm)
sphericity of particle ()
variance of calculated RTD from experimental RTD ()
liquid density (kg m3)
particle density (kg m3)
liquid lm thickness (mm)
liquid dynamic retention (dynamic volume/pure solid
volume) ()
liquid static retention (static volume/pure solid volume)
()
total liquid retention (total liquid volume/pure solid
volume) ()
theoretical liquid residence time (s)
dimensionless time ()
height of the packing bed (m)
liquid adsorption fraction of the medium ()
trickling xed-bed column
Multi-section bioreactor

Q
mDP
mLS
mfd
msd
VLt
VLS
VLd
Vsolid
Vp,L
Nparticles
SA
SAeff
S.H.L
m
fW
fLSE
PF
CSTR
LRT

volumetric ow rate (m3 h1)


total dry packing mass (kg)
liquid static holdup mass (kg)
fast dynamic holdup mass (m3)
slow dynamic holdup mass (m3)
total liquid holdup volume (m3)
liquid static holdup volume (m3)
liquid dynamic holdup volume (m3)
pure solid volume (m3)
liquid volume around single particle (m3)
number of particles ()
total surface area of packing bed (m2)
effective surface area of packing bed (m2)
surface hydraulic loads (m h1)
fraction of active zone in packing bed ()
wetting fraction of the packing bed ()
fraction of partial static holdup volume of tracer exchange (%)
plug ow
continuous stirred-tank reactor
liquid residence time (s)

on the type of medium. Dynamic holdup is formed by the owing


liquid and is highly dependent on the liquid load. For some types of
medium, e.g. structured and non-porous medium [1], dynamic
holdup can be considered as the predominant part of the total liquid holdup [2,12]. However, due to the physical characteristics
of the porous packing employed in our study, the static holdup
fraction may have contributed more to the total liquid holdup than
the dynamic.
With the ultimate aim of optimizing TF design and operation,
the main objective of this study was to characterize the hydrodynamic behavior of two types of TF lled with the same porous
medium in order to assess the impact of the properties of the medium on the overall hydrodynamics. Such impact was measured in
term of liquid holdup fractions, liquid lm thickness under the regimes with and without biolm all along the column after two organic loading rates cultivation. Further objective was using RTD
experiments and modeling to investigate the changes in liquid ow
pattern and residence time due to the presence of biolm in the
MSB. Additionally, the study attempted to verify whether the conguration of the bioreactor (TFC or MSB) would affect its hydrodynamic characteristics.

consecutive sections. The geometrical characteristics of the two


bioreactors are shown in Table 1.
Non-spherical pieces of Concrete Block were used for the solid
phase. They were made up of y ash, gypsum, and cement. A typical Concrete Block particle without biolm is shown in the middle
of Fig. 1 and to the right; two colonised particles obtained from
two organic regimes are shown. The physical characteristics of
the particles and the packing bed, determined in earlier studies,
are presented in Table 2.
The MSB, lled with particles of the Concrete Block medium,
was fed with the primary sludge from a WWTP-GINETOUS, after
sifting through a 1 mm-mesh sieve and mixing with a certain volume of tap water to full the inuent requirements for two organic
loads. The compositions of these two feed wastewaters are shown
in Table 3.
The concentration values of constituent pollutants were ascertained from the mean measured values. The pilot-scale MSB was
fed with two organic loads for 120 days for each, at the same ow
rate of 0.08 m3/d. (In the following contents, the regimes with biolm 1 and biolm 2 represent the regime with lower and high cultivated organic loads, respectively).

2. Material and methods

2.2. Methods

2.1. Experimental system and medium

2.2.1. Static holdup measurements


Liquid static holdup without biolm was determined experimentally by a weighing method as reported in other study [13].
The difference between the mass of the moist packing after drainage and the mass of the dry packing is the static holdup mass
which represents the liquid captured in the particles.
The method for characterizing the static holdup when biolm
was present in the MSB was slightly different from the method described; to avoid the destruction of the biolm, the coated particles
cannot be moved from their original positions and the biolm
should not be completely dehydrated; therefore a constant feed
ow rate of 0.0046 m3/d was applied to maintain bioactivity.
Thanks to the easy sampling of each basket containing the biolm

The experimental system was set up with two types of Trickle


Filter, a Trickle Fixed-bed Column (TFC) (Fig. 1a) and a Multi-section Bioreactor (MSB) (Fig. 1b) linked with a balance or a conductometer connected with a data acquisition system.
The TFC had a diameter of 0.4 m and a height of 2 m and was
enclosed in a structure made of PVC. It was packed with Concrete
Block particles to a height of 1.9 m. The MSB was composed of 5
uniform open-structured frames (like baskets) stacked vertically
whose individual height and diameter were 0.4 m and 0.2 m,
respectively. The total pilot height was 2.4 m while the packing
bed was 2 m high with an interval of 0.1 m between any two

165

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

Fig. 1. Diagram of the TFC (a) and the MSB (b).

Table 1
Geometrical characteristics of the TFC and the MSB.
Bioreactor

Column height (m)

Diameter (m)

Packing depth (m)

Distributor height (m)

Single section height (m)

Injection spot height (m)

TFC
MSB

4
2.4

0.4
0.2

1.9
2

2
2.5

0.4

2.5
2.5

Table 2
Physical properties of Concrete Block medium and packing bed.
Particle porosity
(%)

Material density
(kg/m3)

Specic surface area


(m2/m3)

Particle density
(kg/m3)

Particle equivalent
diameter (cm)

Particle
sphericity

Total packing bed void


fraction

e0

qpowder

586.4

dp
2.17 0.4

1337

SSA
350 70

qparticle

61

0.74 0.07

79.1/77.4a

For the MSB packing bed void fraction calculation, the interval void spaces between two adjacent sections were included.

support, the mass of each section packed with particles coated by


moist biolm at steady-state (no free water draining out but still
moist) was measured. The pilot was then fed at a dened higher
ow rates (0.0091, 0.0182, 0.0137 or 0.0228 m3/h) for 1hr, and
then stopped; the mass of each section packed with moist particles
and biolm was weighed again when no further draining water
was observed. The difference between the two masses obtained
gave the liquid static holdup mass captured in both the particles
and biolm.
The liquid static holdup volume VLS, solid volume Vsolid, total
packing volume Vpacking, the porosity of particle e, the total void
fraction of packing bed e, and the volume of column Vcolumn were
then taken into account to calculate the liquid static retention bS,
in accordance with [9]:

bS

V LS
V LS
V LS

V solid V packing 1  e0 V column 1  e

The water adsorption fraction of the medium, Fad, can then be


calculated as the ratio of liquid static holdup mass mLS and dry
packing mass mDP in the following equation:

Fad

mLS
qL
V LS
q 1  e0


L
 bS
mDP qparticle V packing
qparticle

2.2.2. Dynamic holdup measurements


The drainage method reported by Brunazzi and Viva [13] was
adopted in our study for measuring the liquid dynamic holdup.
The procedure was as follows: the column was rst stabilized by
feeding with a high hydraulic load for several minutes; after stabilization at the required ow rate (constant ow measured at the
outlet), the inlet was shut and the liquid held by the packing began
to drain down into the collector placed on the scale. The real-time
drained mass of liquid collected was measured and recorded by the
acquisition system. When steady state drainage was reached (constant outlet values observed), measurement was stopped.
The dynamic retention bd was calculated from the following
equation:

bd

V Ld
V Ld

V solid V column 1  e

166

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

where VLd represents the liquid dynamic holdup volume and Vsolid
the pure solid volume.
2.2.3. Residence time distribution (RTD)
Residence time distribution experiments were carried out by
the tracer method reported in [14]. After stabilization of the ow
in the column, a salt tracer (30 ml  300 g/L aqueous solution of
sodium chloride) was injected using a syringe from the injection
spot. The conductivity of the liquid with tracer leaving the column
was measured via a ow-through probe (XE100, RADIOMETER
ANALYTICAL S.A.) connected to a conductometer (CDM210 Conductivity Meter, RADIOMETER ANALYTICAL) and the real-time conductivity was monitored by the data acquisition system in the
computer. After each experiment, rinsing was done to remove
any residual salt absorbed of the particles.
The dimensionless E (h) as the function of the dimensionless
time h derives from Eq. (4) and the dimensionless time h in Eq.
(5). The mean liquid residence time l derives from the following
equation:

Ct
C0

Eh

s
P

tCt
l P

Ct

where C0 = mtracer/Vliquid gives the tracer concentration obtained if


all the injected mass was diluted into the liquid volume; C (t) is
the tracer mass held by liquid at measuring time t; l is the mean
Liquid Residence Time (LRT) and the theoretical liquid residence
time s = Vliquid/Q; Q is the ow rate of liquid;
2.2.4. Liquid lm thickness determination
The equivalent total solid volume calculated from the equivalent diameter of spherical particles (2.17 cm) and their number
of particles (8350 pieces) is about 0.045 m3, which was close to
that of real packing bed volume (0.044 m3) estimated from solid
mass (25.63 kg) introduced into the column and the solid density
(586.4 kg/m3) reported in Table 2. So, we assume the particles were
spherical, the thickness of the liquid layer was calculated by the
volume of dynamic liquid owing over the wetted surface around
the particles (with or without biolm). Assumptions for the calculation are schematized in Fig. 2.
Assuming a homogeneous distribution of biolm in the packing
bed and a uniform coating on all particle surfaces, particles coated
with biolm were also assumed spherical. The volume of liquid Vp,L
around a single particle can be calculated from the following
equation:

V p;L

pdp 2Lf 2dL 3  pdp 2Lf 3


6

 4pdp =2 Lf dL

where dL is the liquid lm thickness; Lf, the mean biolm thickness


(equal to 0 when no biolm is present).
For the entire packing bed, the liquid dynamic volume derives
from Eq. (8). In such a case, the effective wetting surface area of
the packing bed SAeff was taken into account:

V Ld Nparticles  V p:L Nparticles dL  SAeff Nparticles dL  SA  fw

where Nparticles is the estimated number of particles in the packing


bed; SA represents the total surface area of packing bed; fw denotes
the wetting factor.
The liquid lm thickness was thereby calculated with the following equation:

dL

V Ld
V Ld

Nparticles SA  fw Nparticles 4pdp =2 Lf 2 fw

3. Results and discussion


In order to study the hydrodynamic behavior of a pilot-scale
MSB packed with a Concrete Block medium and also the inuence
of biolm on changes in such hydrodynamic behavior, the static
and dynamic holdup were investigated by conventional methods
while RTD experiments were carried out to determine liquid lm
thickness, the liquid prole and liquid residence time. A comparison was made with the TFC, mainly using regimes without biolm,
to investigate the effect of lter conguration on the hydrodynamic behavior.
3.1. Static holdup
3.1.1. Without biolm
During the static holdup experiments without biolm, it was
found that a constant water adsorption fraction Fad approximate
56% 6% at steady-state, which implies this medium, has a high
water adsorption capacity. The static retention of the two bioreactors, based on the water adsorption fraction Fad and experimental
measurements, are shown in Table 4.
Whatever the reactor type, it was observed that no signicant
differences existed in static retention values at steady-state, showing that the impact of the interval and the open structure of MSB
on the static holdup can be disregarded.
Additionally, the static holdup volume VLS was found to increase
linearly when the number of particles Nparticles increased. This linear relation between static holdup volume and number of particles
implies a possible extrapolation of the number of particles to liquid
lm thickness calculations based on Eq. (9).
The results found accord well with what is generally reported,
i.e. the liquid static holdup depends much more on the properties
of a medium and on the porosity created than on a reactors
conguration.
3.1.2. With biolm
The liquid static retention of both the entire pilot set-up and of
each individual section (each basket) was calculated after longterm biolm development. During the biolm development, biolm thickness varied along the packing bed depth and along each
section. Even on the same horizontal stratum, the distribution of
biolm thickness was not homogeneous. Owing to the lack of enough accurate information from biolm thickness distribution,
the values of mean biolm thickness adopted in our study were
calculated for each section as the ratio of the volume of total dry
biomass per corresponding particle surface area. The assumption
underlying this calculation is of a homogeneous biolm thickness
and a uniform coating of the particles by the biolm. The liquid

Table 3
Composition of feed wastewater for two cultivated organic loads.
Composition

Concentration values

Unit

Lower load

High load

TSS
VSS
CODt
TKN
TN
Ammonia
Nitrite & nitrate

318 95
170 64
298 71
46 5
46 5
30 3
0

1006 115
606 77
1016 164
161 14
161 14
100 11
0

g/m3
g/m3
gCOD/m3
gN/m3
gN/m3
gN/m3
gN/m3

Organic loads

0.76

2.55

kgCOD/m2d

167

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

Liquid static retention (%)

120%
s of entire pilot Lf=144 mm
s of section 1 Lf=181 mm
s of section 2 Lf=143 mm
s of section 3 Lf=135 mm
s of section 4 Lf=124 mm
s of section 5 Lf=120 mm

100%
80%
60%
40%
20%

(a) low OLR cultivation


0%

0.2

0.4

0.6

0.8

0.6

0.8

Surface hydraulic load (m/h)


Fig. 2. Diagram of liquid layer and contact surface.

static retention fractions in each section and in the entire pilot setup are plotted in Fig. 3 against the surface hydraulic loads.
Fig. 3 shows that the liquid static retention increased when the
hydraulic loads rose (0.29 to 0.73 m/h) in each section and in the
entire pilot set-up (from 34% to 83% in regime 1, from 24% to
79% in regime 2). The liquid static retention observed in each section of regime 1 increased when biolm thickness increased (from
120 lm to 181 lm) which could result from lower packing-bed
voids (from 62% to 43%, a calculation based on the study by Robin
and Alfred, [15]). Throughout the whole depth of the pilot set-up,
from the top down to the lower sections, the biolm thickness
gradually decreased. However, static retention did not decrease
in exact correspondence with the decline in thickness, i.e. when
under greater loads: the static holdup of 2nd section with a biolm
thickness of 282 lm was lower than that of Sections 3 and 4 whose
biolm thicknesses were 243 and 219 lm, respectively. This implies that biolm porosity and variations in thickness should be taken into account simultaneously. The trend as described for regime
1 is the same as that for regime 2. The decline of static retention
when biolm thickness increased may be due to the reduction of
biolm porosity. Where the biolm is thick (e.g. 294 lm in the
1st section under greater loads), it acts like a barrier to the internal
pores of the particle, limiting the outside/inside water exchange. At
a lower hydraulic load (0.29 m/h), liquid was adsorbed mainly into
the pores of the thick biolm and thus captured; in contrast, at a
higher load (0.73 m/h), the liquid could penetrate the biolm and
be partially adsorbed by the particles, thus leading to greater static
retention. Thanks to lower biolm density and to a higher ow
rate, thinner biolm (e.g. 135 lm in 3rd section) could also produce greater static retention (70%) than thick biolm in the same
section (58% with biolm thickness of 243 lm in 3rd section).
3.2. Dynamic holdup
The dynamic holdup experiments in two bioreactors without
biomass and under different hydraulic loads were rst carried
out using the draining method. On the basis of the real-time liquid
draining mass curve during the dynamic experiments, two portions of dynamic holdup were distinguished: a linear segment, representing the fast dynamic fraction, and the coherent curve
segment, representing the slow dynamic fraction.

Liquid static retention (%)

120%
s of entire pilot Lf=248 mm
s of section 1 Lf=294 mm
s of section 2 Lf=282 mm
s of section 3 Lf=243 mm
s of section 4 Lf=219 mm
s of section 5 Lf=202 mm

100%
80%
60%
40%
20%

(b) high OLR cultivation


0%

0.2

0.4

Surface hydraulic load (m/h)


Fig. 3. Liquid static retention of each section (Section 1 was located at the top) and
the entire pilot set-up under different hydraulic conditions.

3.2.1. Without biolm


A real-time drainage process in the TFC is plotted in Fig. 4 for a
ow rate of 0.3 m3/h.
From Fig. 4, three standard stages of drainage can be identied:
Stage 1: stabilization of the ow; Stage 2: unbound liquid around
the particles surface leaves the packing bed; Stage 3: liquid
embedded inside the particles pores begins to leave the packing
bed. Because of the small pore size of the solid, it was assumed that
the majority of the fast dynamic holdup corresponded to the external liquid lm surrounding the surface of the particles (fast liquid
dynamic mass = 10.5 kg). The slow dynamic fraction corresponded
to a retained water fraction (slow liquid dynamic mass = 4.6 kg).
Repeated measurements showed the ratio between fast and slow
dynamic mass as approximately 2 at the volumetric ow rate of
0.3 m3/h. Similarly, these calculations of fast and slow dynamic
holdup mass were thereafter carried out for MSB. However, the ratio between the fast and slow dynamic holdup in the MSB was
found to have increased (from 0.6 to 1.8) when the ow rate rose,
because of the relatively stable water adsorption capacity of the
medium. The fast dynamic increased gradually when the ow rate
rose, though the slow dynamic holdup remains constant.
The dynamic and static retention performances in both bioreactors without biolm are plotted in Fig. 5 against the surface
hydraulic load.
Fig. 5 shows that the static retention values for both bioreactors
without biolm were close at approximately 84%, even though the

Table 4
Static retention measurements and Fad of two reactors.
Reactor

mDP (kg)

mLS calculated (kg)

mLS measured (kg)

VLS (m3)

e (%)

bS calculated (%)

bS measured (%)

TFC
MSB

75
25

42
14

41.842.2
13.814.1

0.042
0.014

79.1
77.4

84
84

8485
8384

168

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

15%

stage stage

Stage 3

End of drainage

15
12

Slow dynamic

9
Fast dynamic

6
3

flowrate=0.3m3/h in TFC

Stop the influent

0
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Liquid dynamic retention (%)

Draining liquid mass (kg)

18

d of entire pilot
d of section 1
d of section 2
d of section 3
d of section 4
d of section 5

12%
9%
6%
3%

(a) low OLR cultivation


0%
0

Time (h)

0.2

0.4

0.6

0.8

Surface hydraulic load (m/h)

Fig. 4. Dynamic drainage process in the TFC at a ow rate of 0.3 m /h.

static retention in MSB


dynamic retention in MSB

Liquid retention (%)

100%

Liquid dynamic retention (%)

25%
static retention in TFC
dynamic retention in TFC

80%
60%
40%

20%
15%
10%
5%

(b) high OLR cultivation


0%

20%
0%

d of entire pilot
d of section 1
d of section 2
d of section 3
d of section 4
d of section 5

0.2

0.4

0.6

0.8

Surface hydraulic load (m/h)


0

0.5

1.5

2.5

Surface hydraulic load (m/h)


Fig. 5. Liquid retention in relation to surface hydraulic load in two bioreactors
without biolm.

hydraulic loads used for the TFC was higher. Just as the discussion
from Table 4, it implies this type of media has constant adsorptive
capacity of liquid when no biolm was present. In contrast, in the
case of MSB the dynamic retention increased (from 9% to 25%)
when the hydraulic loads rose (from 0.14 to 0.87 m/h). It indicates
that the dynamic holdup will increase when increasing the ow
rate. Even if the experiments in the TFC were carried out at a greater hydraulic load (2.39 m/h), its dynamic retention (25.6%) was no
much higher than the best recorded for the MSB (25%). This resulted from the different mass of packing applied into the column;
however, the net volume of liquid dynamic holdup was higher than
that in the MSB.
3.2.2. With biolm
In the MSB, the liquid retention experiments were carried out
after biolm development on the medium at the same ow rates
as those applied when no biolm was present. The liquid retention
at regimes with biolm in each section and in the entire pilot setup is presented in Fig. 6 in relation to the surface hydraulic loads.
From Fig. 6, it can be seen that the dynamic retention varied little (from 5% to 10% for the entire pilot set-up) when the ow rates
rose compared to the static retention which signicantly increased
(from 24% to 83%) with higher hydraulic loads (see Fig. 3). Those
results mean that the majority of the liquid was held by the particles coated by biolm. It was also observed that under various
operating conditions in each section, the liquid dynamic retention
fractions stayed lower, ranging between 5% and 15%. The maxima
on liquid dynamic retention observable in Fig. 6, at hydraulic loads
slightly higher than 0.40 m/h, is difcult to justify by an error measurement as mass balance and timer accuracies are very good. So
we believe that it could be due to an accumulation of water in
the 2nd section, due to a thicker biolm presence compare to other

Fig. 6. Liquid dynamic retention in relation to the surface hydraulic load with
biolm.

sections where the biolm thickness was lower. This accumulation


could justify for this condition the large dynamic liquid volume
collected on the balance after the stop time.
We also observed biolm detachments that occurred in the 1st
and 2nd section at higher than 0.40 m/h that could sustain this
assumption.
In the regimes without biolm, in contrast (see Fig. 5), the dynamic retention increased slightly (from 9% to 25%) when the ow
rate rose and the static retention kept a constant at about 84%.
It was found that the majority of the liquid was held inside the
pores of the mediums particles coated by biolm. The dynamic
retention in regimes with biolm was lower than in the regimes
without biolm. The high water absorption efciency shown by
our porous packing was reduced by the biolm layer that retarded
water movement from the inner to the outer parts of the particles.
However, at steady state, it was considered that the external liquid
volume (dynamic holdup) would be sufcient to ensure the requisite nutriment mass transport from the waste liquid to the biolm.
Liquid lm thickness with and without biolm will be studied further later in this paper.
3.3. Residence time distribution (RTD)
Analyzing the real RTD curves enables us to identify the variation of ow pattern in the column and also the liquid fraction involved in the tracing. Furthermore, the shape of the RTD curves
and, consequently, the distribution of liquid in the packing are related to the conguration of the packing. In order to acquire the
dimensionless function E (h), the trends of the conductivity versus
time were rst established. The experimental real-time conductivity versus measuring time in the MSB without biolm at 4 different
surface hydraulic loads is plotted in Fig. 7.
Fig. 7 shows that increasing the hydraulic load led to taller, narrower curves and thus smaller distribution of liquid residence

169

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

1.6

SHL=0.29 m/h tpeak=1260 s;


SHL=0.44 m/h tpeak=960 s;

SHL=0.58 m/h tpeak=780 s;

1.2

0.8

SHL=0.73 m/h tpeak=640 s;

1.0

E ()

conductivity (ms/cm)

1.4

Flow rate=0.0091 m 3/h


Surface hydraulic load=0.29 m/h

1.2

0.8

0.6
0.4

0.6

MSB without biofilm


MSB with biofilm 2

0.2

0.4

0.2

2000

4000

6000

8000

10000

12000

9 10 11 12 13 14 15 16

0.0
14000

Time (s)

Fig. 9. Experimental RTD curve at the ow rate of 0.0091 m3/h in the MSB with/
without biolm.

Fig. 7. Conductivity versus time under different surface hydraulic loads in MSB.

Table 5
Liquid holdup volume calculation with tpeak and ow rate Q.

1.2

Flow rate=0.0182 m 3/h


with biofilm 2

1
Q
(m3/h)

TFC

0.3

MSB without biolm

MSB with biolm 1

MSB with biolm 2

tpeak
(s)

Vliquid
(m3)

VLd
(m3)

VLt
(m3)

fLSE
(%)

129

0.01075

0.01172

0.05610

0.0091
0.0137
0.0182
0.0228

1260
960
780
640

0.0032
0.0037
0.0039
0.0041

0.00290
0.00370
0.00390
0.00430

0.01710
0.01790
0.01810
0.01850

2
0
0
0

0.0091
0.0137
0.0182
0.0228

1200
960
720
590

0.0030
0.0037
0.0036
0.0037

0.00092
0.00104
0.00132
0.00178

0.00672
0.00964
0.01300
0.01598

36
30
20
14

0.0091
0.0137
0.0182
0.0228

1080
960
960
480

0.0027
0.0037
0.0049
0.0030

0.00083
0.00098
0.00112
0.00114

0.00493
0.00868
0.01262
0.01654

46
35
32
12

0.2
0

Surface hydraulic load


TFC 2.39m/h
MSB 0.29m/h
MSB 0.44m/h
MSB 0.58m/h
MSB 0.73m/h

E ()

2
1.5

Fig. 10. Experimental and calculated RTD curves in the MSB with biolm 2.

obtained from the drainage method, in order to determine which


volume represents the liquid holdup volume in the RTD curves.
Additionally, the fractions of static holdup volume fLSE that exchanges the tracer with dynamic holdup volume were determined
by the partial static holdup volume involved in Vliquid over total static holdup volume VLS. The results are shown in Table 5.
In Table 5, calculated values of Vliquid are basically close to the
values of VLd except when biolm is present, the measured values
of VLd being thus much smaller than the calculated Vliquid. Thus the
biolm acts as a sponge that sustains and partially releases static
holdup liquid. The fraction of exchange tracing static holdup volume fLSE decreased (e.g. 36% to 14% of regime 1) when the ow rate
rose; however, these fractions increased with biolm development
as from regime 1 to regime 2 (e.g. from 36% to 46% at a ow rate of
0.0091 m3/h and similarly under other ow rates). The decrease of

with biofilm 2

2.5

Surface hydraulic load


MSB 0.29m/h
MSB 0.44m/h
MSB 0.58m/h
MSB 0.73m/h

1
0.5

9 10 11 12 13 14 15 16

E ()

without biofilm

0.6
0.4

time, the higher peaks implying less tracer adsorption by the solids. However, the asymmetric behavior of the tail in the curves
may be due to the presence of stagnant zones. If the time tpeak
(when the conductivity is at the maximum) is assumed as the theoretical liquid residence time stheo, the liquid holdup volumes Vliquid
can be calculated as the tpeak multiplied by the ow rate Q. The liquid holdup represented by the RTD curve should generally correspond to the dynamic holdup [16]. However, Viva and Brunazzi, [1]
reported that the liquid holdup in the RTD curves of catalytic structured packing corresponded to the total liquid holdup (the dynamic holdup plus the static holdup) determined by the drainage
method. In our study, the results of Vliquid can be compared with
the dynamic holdup volume VLd and the total holdup volume VLt

2.5

Calculated curve

0.8

E ()

Types

experimental curve

1.5
1
0.5

0
0

10

(a)

(b)

9 10 11 12 13 14 15 16

Fig. 8. Dimensionless RTD curves at different ow rate in two reactors without and with biolm.

170

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

Table 6
Analysis from the model of n CSTR with dead zone in series.
Flow rate (m3/h)
Q

S.H.L (m/h)

TFC

0.3

MSB no biolm

0.0091
0.0137
0.0182
0.0228

MSB with biolm 2

0.0091
0.0137
0.0182
0.0228

Type

Active zone fraction


m

No. of CSTR
n

Variance

Mean LRT (h)

r2

2.39

0.88

23

0.21

0.02

16

0.29
0.44
0.58
0.73

0.61
0.46
0.4
0.49

7
8
9
13

0.51
0.51
0.54
0.45

0.83
0.64
0.58
0.45

5
7
13
14

0.29
0.44
0.58
0.73

0.72
0.63
0.6
0.55

2
4
5
6

0.61
0.83
0.64
0.52

2.71
1.93
1.82
0.65

7
6
7
15

these fractions implies that increasing the hydraulic loads will reduce the diffusion of tracer between the static liquid phase and the
dynamic liquid phase; whereas the diffusion will increase during
the biolm development process under the same hydraulic conditions. These changes may result from the reduction of contact time
between the liquid and solid phase when the ow rate rises.
The curves of dimensionless E (h) as a function of the dimensionless time h at different hydraulic loads for the bioreactors both
without and with biolm are plotted in Fig. 8a and b. The dimensionless time h was calculated as the ratio of the time t to tpeak.
The liquid holdup volume Vliquid taken from Table 5 was also used
for the C0 calculation.
In Fig. 8, comparing the RTD curves acquired at a greater ow
rate in the TFC, such RTD curves at lower ow rates in the MSB display lower peaks but a greater distribution area (h P 12) which
was assumed to be due to the sustained release of liquid in the column. The tails of the curves in the MSB may be due to the presence
of stagnant zones in the column.
Furthermore, the regimes with and without biolm were compared at each volumetric ow rate (see Fig. 8). An example for a
ow rate of 0.0091 m3/h is shown in Fig. 9.
From Fig. 9, it is clear that the presence of biolm lengthened
the liquid residence time in the lter (h ? 15) compared to times
without biolm (h ? 11). This could imply two time scales phenomena: one was related to the dynamic ow; the other was related to the diffusive process that was promoted by the presence
of biolm which decreased the packing bed porosity and increased
the contact time between liquid and biolm. A sustained-release
process is also signied by the longer tail in the RTD curve. In addition, lower peak when biolm was present implies more tracer
adsorption in the biolm compare to the higher peak when biolm
was absent. Furthermore, the same behavior was also observed at
other ow rates. In a bioreactor with biomass, the tracer can be
transported within the biolm by diffusion. The biolm slowly ex-

Liquid film thickness (mm)

0.4

Nparticles =23920

Nparticles =8350

0.3

0.2

LinTFCwithoutbiofilm
LinMSBwithoutbiofilm
LinMSBwithbiofilm1
LinMSBwithbiofilm2

0.1

0.0
0.0

0.5

1.0

1.5

2.0

2.5

3.0

Surface hydraulic load (m/h)


Fig. 11. Liquid lm thickness estimation versus surface hydraulic loads for in two
bioreactors with/without biolm.

Error (%)

changes the tracer with the ow-through zones; this can signicantly increase the tailing potential and lead to a sustained
release process which was also observed and investigated by Riemer et al. [17] who proposed a biodiffusion model to t the liquid residence time distribution in their study.
For the liquid distribution in a trickle bed, it is widely accepted
in many studies that a Gaussian Plug Flow (PF) model with axial
dispersion may t the liquid prole when longer tails are present;
on the other hand, the model of n Continuous Stirred Tank Reactor
(CSTR) with stagnant/dead zone reported by Levich et al.; SantAnna et al. [18,19] has also been widely employed for investigating
the liquid prole in a Trickle Filter. Moreover, a Plug Flow reactor
can be approximated by a large number of CSTR in series. Hence
the model of n CSTR with stagnant/dead zones in series in the form
of Eq. (10) was applied to investigate the RTD in both bioreactors
with the variance derived in the following equation:

1
nn
nh
hn1 e m
n
m n  1!

Eh

r2

t 2 Ct
P
1
Ct

10

11

where m is the fraction of active zone, n is the number of CSTR.


When m equals 1, it means no stagnant/dead zone is present in
the packing bed.
With a variance r2  0, the liquid distribution in the column is
close to plug ow pattern, whereas when r2  1, the liquid distribution is close to completely stirred pattern.
RTD curves were calculated on the basis of Eq. (10) and compared with the experimental curves of all regimes. Fig. 10 presents
an example of the calculated RTD curve and the experimental RTD
curve in the MSB with biolm.
In Fig. 10, the RTD curve calculated on the basis of Eq. (10) ts well
with the experimental RTD curve expect for the tail segment which
is more gradual than the model. General results from the models for
both bioreactors are shown in Table 6. A certain discrepancy exists
between our experimental and modeling curves (about 10%); these
errors have no consequences on the following discussion, where the
ow pattern was discussed in relation to operating conditions.
In Table 6, in the TFC without biolm at a higher ow rate of
0.3 m3/h, the estimated number n of CSTR is 23 which is greater than
that of the MSB at lower ow rates (n ranging from 7 to 13), which
implies the liquid prole in the TFC at a higher ow rate, close to
the plug ow pattern. The variance value of 0.21 in the TFC,
approaching 0, also suggests the liquid prole close to the plug ow.
However, the variance values (0.610.83) for MSB with biolm 2 at
lower ow rates, approaching 1, imply better tracer diffusion between liquid and biolm. The fraction of the active zone increased
(0.550.72) when the biolm was present and the number of CSTR
decrease (26) compared to the cases without biolm (713). This
increase of m corresponds well to the trend to the increased fLSE

P. Haoran et al. / Chemical Engineering Journal 231 (2013) 163171

where the exchange between the dynamic and static phases increased when the biolm was present. It indicates that the presence
of biolm will result in enhanced mass diffusion in the lter.
Comparing the mean Liquid Residence Times l obtained for the
regime without biolm and for the biolm 2 regime, the presence
of biolm indeed lengthened the residence time in the bioreactor(e.g. from 0.83 to 2.71 h at a ow rate of 0.0091 m3/h).
The presence of biolm effectively lengthened the liquid residence time, improving efciency in pollutants treatment.

3.4. Liquid lm thickness


The liquid lm thickness in the different regimes was calculated
by multiplying the liquid dynamic holdup volume VLd divided by
the effective surface area, assuming that stagnant zones existed
in the pilot set-up and the tracer was only exchanged in the active
fraction. When no biolm was present, Vliquid equaled VLd. In contrast, when biolm was present, the values of VLd from Table 5
were introduced to Eq. (9). Additionally, during the hydrodynamic
experiments, it was observed that even the inuent of liquid was
not well distributed; particles that were not initially wetted were
eventually wetted thanks to their diffusion among adjacent particles after sufcient time. Hence it was assumed for the sake of simplicity that the wetting factor fw in Eq. (9) equals 1. Applying the
equivalent particle diameter dp of 0.0217 m, the total surface area
of the medium was calculated by total number of particles Nparticles
multiplied by the surface area of a single particle when no biolm
was present. When calculating these values for regimes with biolm, the mean biolm thickness Lf was introduced into Eq. (9).
Given these assumptions, the liquid lm thickness for different
surface hydraulic loads of regimes with and without biolm in
both bioreactors is plotted in Fig. 11.
As shown in Fig. 11, the liquid lm thickness acquired in the TFC
without biolm at a hydraulic load of 2.39 m/h was 0.33 mm. Under lower hydraulic loads (0.290.73 m/h) in the MSB, the calculated liquid lm thicknesses were less than those obtained in the
TFC: when biolm was absent, it was ranged from 0.23 to
0.35 mm; for biolm 1 regime, it was ranged from (0.07 to
0.14 mm) and for biolm 2 regime, from (0.06 to 0.09 mm).
The regimes without biolm had obvious thicker liquid lm
than the regimes with biolm, which may be due to higher liquid
dynamic holdup volume. Furthermore, as the biolm thickness increased (from regime 1 to regime 2), liquid lm thickness values
were less which resulted from the increased surface area of the
packing bed.
It was found that in regimes without biolm, the thickness of the liquid layer was signicantly greater than in regimes with biolm, which
may be due to the greater volume of dynamic holdup. However, in the
same bioreactor but at a lower ow rate, the smaller dynamic holdup
volume resulted in the reduction of liquid lm thickness.
The presence of biolm led to a thinner liquid lm compared to
that obtained without biolm, a result attributable to the increase
of surface area with biolm development and the decrease of dynamic draining volume (see Table 5). Such behavior will promote
the oxygen transfer rate by decreasing the limited liquid layer
and lengthening the contact time between the liquid and biolm.

4. Conclusion
Different hydraulic experiments were carried out, essentially in
a new type of TF made up of a Multi-Section Bioreactor packed
with concrete block medium particles. The purpose of the study
was to investigate the hydrodynamic characteristics particularly
in the light of the inuence of biolm on hydrodynamic behavior.

171

The static experiments with biolm indicate that most of the liquid will be retained by the particles coated with biolm, increasing
static retention and, consequently, reducing dynamic retention. It
was also found that the liquid static holdup makes a greater contribution to total liquid holdup than the dynamic holdup on account of the
high adsorption potential of the Concrete Block medium resulting
from its porous structure. However the static holdup is not in correlation with the conguration of TF comparing with the results of TFC.
RTD experiments show that at lower ow rates, the mass diffusion between the liquid and biolm was better than that at higher
ow rates. Increasing the hydraulic load will result in the ow pattern approaching to the plug ow pattern in the bioreactor. The liquid volumes represent in the RTD agreed with the dynamic
holdup volumes obtained by the drainage method with no biolm
is present. However, with biolm, these liquid volumes represent
not only the dynamic holdup but also partial static holdup volume
resulting from the sustained release of partial static holdup in the
biolm. The LRT calculated from RTD models shows that the presence of biolm will lead to a longer liquid residence time in the lter and thus promote the distribution of liquid in the bioreactor. So,
comparing the drainage and RTD methods allow us to show that
diffusion phenomenon can occur in the biolm, increasing the contact time between liquid and biolm.
Furthermore, the presence of biolm will decrease the thickness
of the liquid lm compared with the regimes without biolm under the same hydraulic conditions on account of the greater surface
area and a smaller dynamic holdup volume.
References
[1] A. Viva, E. Brunazzi, Experimental analysis of the liquid hold-up contributions
in catalytic packing Katapak-SP, Proc. Euro. Congress Chem. Eng. (2007) (ECCE6), Copenhagen.
[2] P. Suess, L. Spiegel, Hold-up of Mellapack structured packings, Chem. Eng.
Process. 31 (1992) 119124.
[3] R.D. Darakchiev, N.N. Kolev, Packed bioreactor: hydrodynamic behavior and
mass transfer, Theor. Found. Chem. Eng. 30 (1996) 144147.
[4] F.M. Samb, M. Deront, N. Adler, P. Peringer, Dynamic liquid holdup and oxygen
mass transfer in a cocurrent upow bioreactor with small packing at low
Reynolds numbers, Chem. Biochem. Eng. J. 62 (1996) 237240.
[5] F. Seguret, Y. Racault, M. Sardin, Hydrodynamic behaviour of full scale trickling
lters, Water Res. 34 (2000) 15511558.
[6] E. Brunazzi, U. Di Festa, C. Galletti, C. Merello, A. Paglianti, S. Pintus, Measuring
volumetric phase fractions in a gas-solid-liquid stirred tank reactor using an
impedance probe, Can. J. Chem. Eng. 80 (2002) 688694.
[7] G. Urrutia, P. Bonelli, M.C. Cassanello, A.L. Cukierman, On dynamic liquid
holdup determination by the drainage method, Chem. Eng. Sci. 51 (1996)
37213726.
[8] M. Tuller, D. Or, Hydraulic functions for swelling soils: pore scale
considerations, J. Hydrology 272 (2003) 5071.
[9] F. Mounir, Epuration deau use urbaine par lit bactrien et biodisques, in:
Genie des procedes et des reactions chimiques, Ph.D. thesis. Universite
Montpellier II, 1991, p. 140.
[10] J.R.F. Johann, G. Stichlmair, Distillation: Principles and Practices, 1 ed., WileyVCH, USA, 1998.
[11] Behrens, Hydrodynamics and mass transfer of modular catalytic structured
packing, Ph.D thesis. Delft University of Technology, 2006.
[12] S. Nakov, D. Djonova-Atanassova, N. Kolev, Dynamic hold-up of packings with
vertical walls, Chem. Eng. Process. 39 (2000) 391397.
[13] Brunazzi, Viva, Experimental investigation of reactive distillation packing
Katapak-SP11:Hydrodynamic aspects and size effects, IChemE (2006) 554
562.
[14] A. Viva, E. Brunazzi, The inuence of modular structure on the hydrodynamics
of catalytic structured packings for reactive separation processes, Icheap-9:
9th Int. Con. Chem. Process Eng., Pts 13, 17 (2009) 15191524.
[15] G. Robin, C. Alfred, Inuence of Biolms on Porous Media Hydrodynamics,
Porous Media, CRC Press, 2010. pp. 173230.
[16] S. Sharvelle, E. McLamore, M.K. Banks, Hydrodynamic characteristics in
biotrickling lters as affected by packing material and hydraulic loading
rate, J. Environ. Eng. Asce. 134 (2008) 346352.
[17] M. Riemer, G. Holm Kristensen, P. Harremos, Residence time distribution in
submerged biolters, Water Res. 14 (1980) 949958.
[18] V.G. Levich, V.S. Markin, Y.A. Chismadzhev, On hydrodynamic mixing in a
model of a porous medium with stagnant zones, Chem. Eng. Sci. 22 (1967)
13571367.
[19] G. SantAnna, M. Roustan, H. Roques, K.T. Nyadziehe, R.B. Aim, Hydrodynamics
of plastic media trickling lters, Environ. Technol. 3 (1982) 395404.

Das könnte Ihnen auch gefallen