Simulation of Growth and Detachment in

Biofilm Systems Under Defined

Hydrodynamic Conditions

Harald Horn,1,* Helmut Reiff,2 Eberhard Morgenroth1,3

1
Department of Civil and Environmental Engineering, University of Illinois
at Urbana-Champaign, Urbana, Illinois 61801
2
Ingenieurgesellschaft für Bauwesen und Umwelttechnik mbH, D-76137
Karlsruhe, Germany
3
Department of Animal Sciences, University of Illinois at
Urbana-Champaign, Urbana, Illinois 61801
Received 14 March 2002; accepted 22 July 2002

DOI: 10.1002/bit.10503

Abstract: Detachment from biofilms was evaluated using
a mixed culture biofilm grown on primary wastewater in
a tube reactor. The growth of biofilms and the detach-
ment of biomass from biofilms are strongly influenced
by hydrodynamic conditions. In a long-term study, three
biofilms were cultivated in a biofilm tube reactor. The
conducted experiments of biofilm growth and detach-
ment can be divided into three phases: 1) an exponential
phase with a rapid increase of the biofilm thickness, 2) a
quasi-steady-state with spontaneous fluctuation of the
biofilm thickness between 500 and 1,200 µm in the in-
vestigated biofilm systems, and 3) a washout experiment
with increased shear stress in three to four steps after
several weeks of quasi-steady-state. Whereas the biofilm
thickness during the homogeneous growth phase can be
regarded constant throughout the reactor, it was found
to be very heterogeneous during the quasi-steady-state
and the washout experiments. Growth and detachment
during all three phases was simulated with the same
one-dimensional biofilm model. For each of the three
phases, a different detachment rate model was used.
During the homogeneous growth phase, detachment
was modeled proportional to the biofilm growth rate.
During the quasi-steady-state phase, detachment was
described by random detachment events assuming a
base biofilm thickness. Finally, the washout experiment
was simulated with detachment being a function of the
biofilm thickness before the increase of the shear stress.
© 2003 Wiley Periodicals, Inc. Biotechnol Bioeng 81: 607–617,
2003.
Keywords: biofilm; tube reactor; detachment; model
INTRODUCTION
Detachment from biofilms is caused by a combination of
processes, including 1) abrasion, 2) erosion, 3) sloughing,
and 4) predator grazing (Bryers, 1988). Detachment occurs
when external forces (e.g., through shear) are larger than the
*Current address and correspondence to: Harald Horn, Hydrochemis-
try, Hochschule Magdeburg-Stendal (FH), Breitscheidstr. 2, D-39114
Magdeburg, Germany; telephone: +49391 8864238; fax: +49391 8864234;
e-mail: harald.horn@chemie.hs-magdeburg.de
internal strength of the matrix that is holding the biofilm
together. Thus, in principle there are two mechanisms that
can lead to detachment: i) increase of the external shear
forces (e.g., during backwashing), or ii) decrease of the
internal strength (e.g., through hydrolysis of the polymeric
biofilm matrix). The discussion of what biological, chemi-
cal, or physical mechanism is the dominating factor in de-
tachment is ongoing (Bakke, 1986; Peyton and Characklis,
1993; Bryers, 2000; Stewart et al., 2000).
Understanding mechanisms and rates of detachment is
necessary to evaluate biofilm development and persistence.
Biofilm formation and the microbial ecology within the bio-
film are determined by the balance of growth and detach-
ment processes. Efficient detachment plays a central role for
the removal of unwanted biofilms in systems such as heat
exchangers (Flemming et al., 1996), dental hygiene (Wilson
et al., 1998), biomaterial implants (Dankert et al., 1986),
and on ship hulls (Cooksey and Wigglesworth-Cooksey,
1995). In drinking water distribution systems, removal of
biofilms is desired; however, detachment of biofilm can
result in a secondary contamination of the drinking water.
For beneficial biofilms in ethanol production (Tzeng et al.,
1991), water purification (Lee and Rittmann, 2000), waste-
water treatment (Morgenroth and Wilderer, 1999), and soil
remediation (Bouwer and Zehnder, 1993) a balance be-
tween accumulation of sufficient active biomass, on the one
hand, and physical clogging of the system, on the other, has
to be achieved. As shown by Morgenroth and Wilderer
(2000), in mixed culture biofilms the rate and dynamics of
detachment also has a significant impact on bacterial com-
petition within the biofilm.
Despite the central importance of detachment in biofilm
development, very little is known about the biological,
chemical, and physical mechanisms of detachment (Stewart
et al., 2000). As a result, most mathematical models to
describe biofilm development are based on simplified as-
sumptions by either completely neglecting detachment (Kis-

© 2003 Wiley Periodicals, Inc.

sel et al., 1984; Fruhen et al., 1991; Kreft et al., 2001;
Picioreanu et al., 1998) or by assuming a constant biofilm
thickness (Wanner and Gujer, 1984; Eberl et al., 2000).
Experimental approaches to quantify overall detachment
rates are based on macroscopic mass balances for biofilm
reactors, relating the mass of effluent suspended solids re-
moved per time to the total mass of biofilm in the system
(Rittmann, 1989). To evaluate local detachment, microscale
flow channels have been used in combination with image
analysis (Murga et al., 1995; Sawyer and Hermanowicz,
1998; Stoodley et al., 1999). However, for these microscale
experiments it is generally not possible to relate local phe-
nomena with an overall mass balance for suspended solids.
Resulting detachment rates have been introduced into math-
ematical biofilm models based on biofilm thickness (Bryers,
1984; Chang and Rittmann, 1987; Kreikenbohm and
Stephan, 1985; Rittmann, 1989; Tijhuis et al., 1995; Trulear
and Characklis, 1982; Wanner and Gujer, 1986), shear
stress (Bakke et al., 1984; Rittmann, 1982; Wang and Bry-
ers, 1997), change of shear stress (Peyton and Characklis,
1993), growth rate (Peyton and Characklis, 1992, 1993;
Robinson et al., 1984; Speitel and DiGiano, 1987; Stewart,
1993), time-dependent detachment events (Benefield and
Molz, 1985; Morgenroth and Wilderer, 1998, 1999, 2000),
or using dimensionless analysis of parameter combinations
(Chang et al., 1991; Nicolella et al., 1996). Recently, a
mathematical model was presented that combines modeling
of heterogeneous biofilms with a mechanistic model of bio-
film detachment assuming that the biofilm can be approxi-
mated as a homogeneous elastic material and that biofilm
detachment results from the combined effect of liquid shear
and biofilm strength (Picioreanu et al., 2001). The abun-
dance of available detachment rate equations partially re-
flects the failure of any one expression to model the detach-
Figure 1. Experimental setup with biofilm tube reactor, mixing tank, and recirculation pump.
608 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 81, NO. 5, MARCH 5, 2003
ment rate over a wide range of conditions (Peyton and Char-
acklis, 1993).
Many laboratory studies under controlled hydrodynamic
conditions (e.g., in flow channels or annular reactors) have
been restricted to measurements lasting less than a week
(Trulear and Characklis, 1982). Within this restricted time
frame it cannot be expected that the biofilm has reached
steady-state conditions and these young and thin biofilms
(<50 ␮m) may not be representative for biofilms observed
in engineering applications. Most of these studies evaluate
biofilm detachment only under constant shear conditions.
However, in natural and engineered biofilm systems, bio-
film properties change over time and external forces can
also change significantly. Only limited information on de-
tachment under changing conditions are available (Bakke,
1986). The purpose of this study was to evaluate biofilm
development and detachment over extended periods
(weeks) and with dynamic variations of shear stress. Then it
was evaluated whether measured biofilm growth and de-
tachment during different stages of the experiments could
be represented using a mathematical model.
MATERIALS AND METHODS
Reactor
The investigated heterotrophic biofilms were grown in a
biofilm tube reactor shown in Figure 1. The experimental
setup consisted of the tube reactor, a mixing tank, a flow
meter, and an eccentric screw recirculation pump (Reiff,
1992). The pump capacity allowed for flow velocities up to
4.8 m/s in the tube reactor. The tube reactor was built out of
Plexiglas (PMMA) and was divided into five segments
which could be weighed separately. The tube was fixed at
an angle of 30° to avoid the collection of gas bubbles in the
reactor. The technical specifications of the biofilm tube re-
actor are listed in Figure 1. Biofilms were grown using
primary settled wastewater from a wastewater treatment
plant (Kassel, Germany). For practical reasons the system
was operated in a sequencing batch mode where the entire
wastewater was exchanged daily causing some fluctuations
of organic substrate concentration over a day (Table I).
Aeration was achieved through recirculation and gas trans-
fer within the mixing tank, resulting in a mean oxygen
concentration >0.5 g/m3 in all experiments. Due to the high
concentration of microorganisms in the wastewater, a sepa-
rate inoculation was not necessary.
Growth Conditions
Three experiments were carried out to investigate the
growth and detachment in the biofilm system under defined
hydrodynamic conditions. Organic substrate concentrations
in the primary wastewater were high and, as a result, oxygen
was the limiting component in all three experiments. The
COD and oxygen concentration in the mixing tank were
measured 2–3 times per week. The samples were taken from
the mixing tank before the wastewater was exchanged. In
addition, pH, total nitrogen concentration, and phosphorus
concentrations were determined according to the DEV
(1981). The biofilms were grown under three different shear
conditions where wall shear stress (␶) was calculated for
turbulent flow using the Blasius equation (Chadwick and
Morfett, 1993):
␶ = 0.0395 Re−0.25 ␳Waterw2
where Re is the Reynolds numbers, w is the mean flow
velocity in the tube, and ␳Water is the density of the water.
The principal growth conditions, including Re, ␶, COD, and
oxygen are shown in Table II.
Biofilm Thickness and Washout Experiments
The biofilm thickness was estimated using the method of
Horn and Hempel (1997). After draining the five segments
Table I. Hydraulic conditions and substrate loading during the experiments.
Procedure
Previous growth conditions
15-min 2-fold flow velocity compared
to growth condition
→ Determination of biofilm thickness
15-min 4-fold flow velocity
→ Determination of biofilm thickness
15-min 8-fold flow velocity
→ Determination of biofilm thickness
15-min 16-fold flow velocity
→ Determination of biofilm thickness
HORN ET AL.: SIMULATION OF GROWTH AND DETACHMENT IN BIOFILM SYSTEMS
vertically for 20 min, the wet biofilm mass was determined
by weighing each drained segment individually. The mean
biofilm thickness was then calculated from the mass of the
wet biofilm assuming a density of the wet biofilm of ␳W ⳱
1 g/cm3. The mean biofilm density ␳F (g DM/m3) was de-
termined by collecting samples of biofilm from the tubes
after the draining procedure. The wet mass of the samples
(mWM) was measured to calculate the biofilm volume VF (⳱
mWM / ␳W). Afterwards, the biofilm samples were dried at
105°C and weighed again (mDM). The mean biofilm density
of the cultivated biofilm was then estimated from:
mDM
␳F = (2)
mWM
␳W
With the exception of Experiment I (Table II), biofilms
were cultivated until they reached a quasi-steady-state with
respect to the biofilm thickness. This quasi-steady-state
phase or plateau phase can be described by a fluctuation of
the thickness in a certain range. After several weeks of
operation, washout experiments were performed. The flow
velocity in the tube reactor during the washout experiment
depended on the flow velocity during the growth of the
biofilm. The washout experiments were conducted in the
following manner: Water from bank filtration was used for
the washout experiments. The bank filtration water obtained
was pumped next to the river Fulda. This water was used
because it contained only negligible amounts of COD (<10
g/m3) and suspended solids (<1 g/m3), while maintaining an
ionic composition close to the growth conditions of the
biofilm. Compared to the batch mode during biofilm culti-
(1) vation, the washout experiments were carried out in a con-
tinuous mode. COD and suspended solids were measured in
the effluent during the washout experiments. Each washout
event was monitored using multiple samples. During Ex-
periment I, five samples were taken during the following
time intervals: 0–0.5 min, 0.5–1.5 min, 1.5–3.5 min, 3.5–6.5
min, and 6.5–15 min. During Experiments II and III, nine
samples were taken in the time intervals: 0–0.17 min, 0.17–
.33 min, 0.33–0.5 min, 0.5–0.83 min, 0.83–1.17 min, 1.17–
1.5 min, 1.5–3.5 min, 3.5–6.5 min, and 6.5–15 min.
Between each increase of the flow velocity, the biofilm
Experiment I Experiment II Experiment III
w [m/s] w [m/s] W [m/s]
0.15 0.3 0.6
0.3 0.6 1.2
0.6 1.2 2.4
1.2 2.4 4.8
2.4 4.8 —
609
Table II. Hydraulic conditions and substrate loading during the experiments.

Duration of Mean flow Reynolds Shear stress Mean COD Mean oxygen Biofilm density
experiment velocity w number ␶ concentration concentration ␳
Experiment days [m/s] Re [N/m2] [g/m3] [g/m3] [kg DM/m3]

I 30 0.15 6,600 0.10 250 ± 103 0.7 55 ± 15

II 45 0.3 13,200 0.33 146 ± 68 1.3 51 ± 15
III 81 0.6 26,400 1.12 268 ± 128 1.5 64 ± 24

thickness was measured as described above. Due to biomass
that had accumulated in the tubes and pump, a mass balance
of the detached biomass could not be done. After an in-
crease of the flow velocity during a washout experiment,
effluent suspended solids were derived from detachment in
the tubular reactor but also from the tubes and pump. Thus,
the mass of effluent suspended solids could not be directly
correlated with the change in biofilm thickness in the reac-
tor. However, normalized effluent suspended solids concen-
trations could be used to calibrate the dynamics of detach-
ment in the mathematical model.
Model Considerations
Kinetic and Stoichiometry
Biofilm growth was described using a simple model includ-
ing growth, maintenance, and inactivation of heterotrophic
bacteria (Table III). The model includes the two dissolved
components for organic matter cS and oxygen cO2 and the
two particulate components for heterotrophic biomass XH
and inactive biomass XI. Growth and maintenance describe
the substrate uptake and growth of the heterotrophic bio-
mass. Inactivation describes the formation of inactive or-
ganic material, for example, EPS, out of heterotrophic bio-
mass. Because of the low oxygen concentrations, it was
assumed that maintenance was a main process in the de-
scribed biofilm systems, which translated into lower KS-
values in the Monod terms of the process maintenance com-
pared to the KS-values for the process growth (Horn and
Hempel, 1997). The kinetic parameters are shown in Table
IV. The presented mass transfer coefficients were estimated
on the basis of Wäsche et al. (2000). Bulk phase concen-
trations for organic substrate and oxygen were taken from
the experiments (Table II). Both the kinetic model and the
detachment model were integrated into a biofilm compart-
ment in the software package AQUASIM (Reichert, 1994).
Detachment
Three different approaches to model detachment were re-
quired to describe distinctly different phases of the experi-
ments (see, for example, Fig. 4): 1) homogeneous growth
with a homogeneous biofilm morphology and steadily in-
creasing biofilm thickness, 2) quasi-steady-state with spon-
taneous detachment events and an approximately constant
mean biofilm thickness, and 3) washout conditions in an
experiment with an incremental increase of shear forces.
During the homogeneous growth phase the detachment
rate coefficient kd of biomass was considered to be depen-
dent on the advective velocity uF by which the biofilm
surface moves perpendicular to the substratum (Wanner and
Reichert, 1996). The detachment was described as follows:
dLF
= uF * kd,hom (3)
dt
where kd,hom can range between 0 and 1 and couples the
detachment velocity to the current growth rate. This proce-
dure allows the description of a continuous detachment rate
during the homogeneous growth phase of the biofilm.
Following the homogeneous growth phase, the biofilm
thickness LF in Experiments II and III reached quasi-steady-

Table III. Process matrix for the heterotrophic biofilm (for kinetic parameters, see Table IV).

Dissolved Solid
components components

Process cS cO2 XH XI Process rate

Growth of heterotrophic bacteria −1 1−YH 1 — cs c02

− ␮H X
YH YH Ks + cs KO2 + cO2 H

Maintenance of heterotrophic −1 −1 — — cs c02

bacteria km X
KSmain + cs KO2main + cO2 H

Inactivation of heterotrophic — — −1 1 Ks K02

bacteria k1 X
Ks + cs KO2 + cO2 H

610 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 81, NO. 5, MARCH 5, 2003

 Table IV. Model parameters for the heterotrophic biofilm.

Parameter Model Values

1
Kinetic
Mean maxium growth rate:
heterotroph ␮H 5 d−1
Yield coefficient:
heterotroph YH 0.5 gX g−1COD
Monod constant:
COD Ks 10 g m−3
Oxygen KO2 0.5 g m−3
Maintenance coefficient km 1.95 d−1
Monod constant:
COD KSmain 1 g m−3
Oxygen KO2main 0.2 g m−3
Reaction rate constant for the process inactivation
heterotroph bacteria kI 0.028 d−1
Diffusion coefficient:
oxygen DO2 2.1 cm2 d−1
glucose Ds 0.58 cm2 d−1
Mass transfer coefficients2:
experiment I ␤ 3 m/d
experiment II ␤ 6 m/d
experiment III ␤ 10 m/d
Density of solid phase:
heterotroph ␳H 110 kg m−3
inert ␳I 110 kg m−3
Initial condition for volume fraction of the
heterotroph bacteria ⑀H ini 0.01 —
inert material ⑀I ini 0.49
Initial biofilm thickness Lfini 1 ␮m
Detachment3
Coefficient for the detachment velocity kd,hom 0.1
Mean base biofilm thickness LF,buse 600 ␮m
Detachment rate constant for forced
detachment kd,detach 800000 d−1
1
Kinetic data were taken from Horn and Hempel (1997).
2
Mass transfer cefficients were estimated using Wäsche et al. (2000).
3
Data fitted using the experimental results.

state conditions with fluctuations of the biofilm thickness
between 500 and 1,200 ␮m. Frequent sloughing events re-
sulted in a very heterogeneous biofilm thickness and the
reported biofilm thickness describes an overall mean bio-
film thickness. Fluctuations of the biofilm thickness were
usually spontaneous and started after approximately 20 days
of growth. In the mathematical model, these spontaneous
events were modeled using a fixed base biofilm thickness
(Morgenroth and Wilderer, 2000) combined with a random
detachment process:
dLF
= kd,random 共LF − LF,base兲
dt
where kd,random is a time-dependent detachment rate that
was modeled in AQUASIM by using a real list variable. The
dynamics of detachment during the quasi-steady-state phase
resulted on average in one detachment event every 6 days.
LF,base is the mean base biofilm thickness. LF,base was esti-
mated based on measured biofilm thickness in Experiments
II and III.
At the end of each growth experiment increased detach-
ment was initiated in washout experiments by incrementally
increasing the shear forces acting on the biofilm surface.
Different modeling approaches had to be evaluated to de-
scribe the change in measured mean biofilm thickness dur-
ing washout. First, the amount of detached biomass (⌬LF)
was correlated with the increased flow velocity correspond-
ing to increased shear stress during washout conditions (Fig.
2a). Second, ⌬LF was correlated with the ratio of the shear
stress during the washout experiment to the shear stress
during the growth phase (Fig. 2b). This approach was based
on the idea that biofilm strength will be a function of shear
(4) conditions during biofilm development. Third, the biofilm
thickness after detachment (LF,detach) was correlated with
the biofilm thickness before each washout event (Fig. 2c).
Surprisingly, both the flow velocity during detachment
and the shear stress ratios did not correlate well with the
detached biomass. Correlating the mean biofilm thickness
LF,detach after the washout experiment with the mean biofilm
thickness LF before the detachment resulted in an acceptable
linear fit and this approach was subsequently used in the
described mathematical model:

HORN ET AL.: SIMULATION OF GROWTH AND DETACHMENT IN BIOFILM SYSTEMS 611

 where kd,detach is the detachment rate constant for the wash-
out experiments. The detachment rate was estimated based
on dynamics of effluent suspended solids during the wash-
out experiments. This approach of using a base biofilm
thickness after detachment is based on Morgenroth and
Wilderer (2000). A similar approach to model periodic de-
tachment where biofilm is removed only in the outer layers
but bacteria below a base thickness were protected from
detachment was used by Rittmann et al. (2002).

RESULTS AND DISCUSSION

The results of the biofilm growth and washout experiments

are shown in Figures 3–5. All three biofilms grew within
2–3 weeks to a maximum biofilm thickness ranging from
500–1,200 ␮m. In Experiment I, a washout experiment was
performed at the end of the homogeneous growth phase. In
Experiments II and III the biofilm was maintained for sev-
eral weeks at quasi-steady-state conditions where biofilm
thickness varied in the range of 500–1,200 ␮m. It should be
noted that detachment during this quasi-steady-state oc-
curred both spontaneously during normal operation and also
during the determination of the biofilm thickness. The pro-
cedure for determining the biofilm thickness required drain-
ing of all water from the tube reactor. Draining was done in

Figure 2. Dependence of the detached biomass ⌬LF (a) on flow velocity

during the washout experiment and (b) on the shear stress ratio. (c) De-
pendence of the mean biofilm thickness after detachment LF,detach on the
biofilm thickness LF before the washout experiment.

LF,det ach = 0.61 LF − 30 (5)

It should be noted that each incremental increase of the
water velocity was exactly doubling the previous flow rate.
LF,detach was then used as the new base biofilm thickness in
the kinetic expression to simulate the next subsequent wash-
out experiment:
dLF
= kd,det ach 共LF − LF,det ach兲 (6) Figure 3. Experiments and simulation of the mean biofilm thickness LF
dt (a) during growth phase and (b) the washout experiment in Experiment I.

612 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 81, NO. 5, MARCH 5, 2003

Figure 4. Experiments and simulation of the mean biofilm thickness LF
(a) during growth phase, steady-state phase, and (b) the washout experi-
ment in Experiment II.
reverse flow direction to normal operating conditions and
resulted in an increased stress for the biofilm.
Biofilm densities were measured to characterize biofilm
structure. Biofilm densities in all three experiments were in
the same range but had a high standard deviation that can be
explained by the method used to determinate the dry bio-
mass (Table II). Low bulk phase oxygen concentrations in
Experiment I resulted in slower biofilm growth compared to
Experiments II and III. Thus, it could have been expected
that the densities in Experiment I are higher compared to
Experiments II and III. Overall the densities are high, which
can be explained by high shear stress during the growth
phase (Table II). Biofilm densities are within the range of
densities reported in the literature. Zhang and Bishop (1994)
reported average biofilm densities of 40 kg/m3 that was
shown to vary over the thickness of the biofilm. Tanyolac
and Beyenal (1997) reported higher biofilm densities for a
fluidized bed reactor ranging from 34–76 kg/m3. In tube
reactors operated at a wall shear stress below 0.05 N/m2
biofilm densities ranged between 5–15 kg/m3 (Wäsche et
al., 2000).
A single set of kinetic parameters (Table IV) was used to
simulate all experiments. Growth kinetics and stoichiometry
were mainly based on default values from the literature. The
HORN ET AL.: SIMULATION OF GROWTH AND DETACHMENT IN BIOFILM SYSTEMS
Figure 5. Experiments and simulation of the mean biofilm thickness LF
(a) during growth phase, steady-state phase, and (b) the washout experi-
ment in Experiment III.
following detachment parameters were estimated based on
experimental results—but all three experiments were de-
scribed using the same set of parameters:
● kd,hom: During the homogeneous growth phase
● kd,random, LF,base: During the quasi-steady-state phase
● kd,detach, LF,detach: During the washout experiment
Detachment during the homogeneous growth phase was
based on Eq. [3]. The simulated biofilm thickness in Figures
3–5 fits best with a value of 0.1 for kd,hom. Thus, 90% of the
net biomass production accumulated during the homoge-
neous growth phase in the biofilm, while only 10% were
removed through detachment.
The detachment during the quasi-steady-state phase of
Experiments II and III was described based on Eq. [4]. The
mean base biofilm thickness, LF,base, was set to 600 ␮m for
the presented simulations. The mean base biofilm thickness
is not constant and is dependent on the growth conditions of
the biofilm. Under very high shear stress and low substrate
supply, the biofilm might never reach a thickness of 600
␮m. So the mean base biofilm thickness is a function of:
● hydrodynamic conditions and reactor design
● substrate load
● involved bacteria
● grazing organisms
613
The random detachment coefficient, kd,random, is a time-
dependent constant and was provided as a real list variable
in AQUASIM. The real list consisted of two values, which
were randomly distributed using a frequency of 0.16 d-1 for
detachment events:
kd,random 再 0 d−1 without detachment
4 d−1 during detachment event
Obviously, such a statistical detachment approach will not
be able to describe individual detachment events correctly.
However, the general dynamics of randomly distributed de-
tachment events are well represented (Figs. 4, 5). In general,
the biofilm thickness during the homogenous growth and
quasi-steady-state phase in Experiments I and III was over-
estimated by the model. On the other hand, the biofilm
thickness in Experiment II was underestimated.
The washout experiments are presented in the expanded
Figures 3b–5b. The mean biofilm thickness LF,detach after a
washout experiment was calculated using Eq. [5]. The de-
tachment coefficient kd,detach describing the dynamics of de-
tachment in Eq. [6] was fitted with the measured response
curves of suspended solids at the outlet of the tube reactor
during the washout experiment. These response curves
were, independent from the biofilm, very similar throughout
all experiments. Fifty to 70% of the detached biomass was
found in the first sample. Figure 6 shows the results of
Experiment III for the first 4 min of the washout experi-
ments. Due to the fixed biomass inside the pump and on the
tube walls the total suspended solids in the effluent were
approximately twice what could be balanced by the mass
calculated using measured ⌬LF and biofilm densities in the
reactor.
In Figure 7, modeled and measured biomass concentra-
tions are shown, where effluent suspended solids are nor-
malized with the maximum effluent concentration. The de-
tachment coefficient kd,detach for these washout experiments
was 800,000 d-1 to achieve rapid detachment of biomass, as
Figure 6. Measurement of suspended solids at the outlet of the tube
reactor during washout in Experiment III.
614 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 81, NO. 5, MARCH 5, 2003
Figure 7. Simulation of the suspended solids at the outlet of the tube
reactor during the first stage of the detachment in Experiment I.
had been observed in reported experiments. All used de-
tachment coefficients are summarized in Table IV.
The driving force for detachment in the washout experi-
ments was the increase in shear stress during the experi-
ments. After the first detachment of biomass, no significant
further detachment was observed during the ongoing ex-
periment. However, it cannot be predicted how the biofilm
might have behaved if the period with higher shear stress
had been extended to more than 15 min. Washout was per-
formed using river bank filtrate with a low COD. It has been
shown that reduced substrate concentrations resulting in
lower growth rates can lead to lower detachment rates
(Speitel and DiGiano, 1987; Peyton and Characklis, 1993;
Tijhuis et al., 1995). However, the response time of detach-
ment rates to changes in growth rates is on the order of
hours. For the washout experiment described in this study,
the influence of a reduced COD is assumed to be negligible.
Another factor that can lead to detachment is a change of
ionic strength and the ionic composition (Turakhia et al.,
1983; Keiding and Nielsen, 1997). However, the ionic com-
position of the river bank filtrate was similar to the primary
wastewater used during biofilm growth. As shown in Figure
6, detachment occurred within seconds after the increase of
shear stress and most of the detached biomass was removed
within the first minute of the washout experiment. It is
concluded that the main mechanism leading to the rapid
detachment during the washout experiments is the increase
of the shear stress. Our findings are in line with Bakke
(1986), who reported a significant increase of detachment
rates during the transient increase of shear stress, while
shear stress did not have a significant influence on steady-
state detachment rates.
The simulated detachment curves in Figures 4b and 5b fit
well with the measured mean biofilm thickness after the
washout experiment. On the other hand, the simulated re-
sults of washout in Experiment I do not fit very well with
the measured data. The simulated decrease of the mean
biofilm thickness is much higher than the measured value.
Two explanations are possible: 1) Significant detachment
had already occurred during measurements of the biofilm
thickness. The measured value was lower than the precedent
value. 2) The doubling of the flow velocity was not the only
parameter which influenced the detachment rate, but also
the value of the shear stress during the growth phase, which
was the lowest during Experiment I.
While the simulation results are not fully satisfying, the
applied detachment model is able to describe the two main
detachment processes investigated: random detachment in a
fully developed biofilm; and forced detachment by in-
creased shear stress. The presented model is based on a
discussion of the mean biofilm thickness, LF. As was shown
by Bakke (1986), the heterogeneity of a biofilm increases
over cultivation time. In the current study the biofilm het-
erogeneity was evaluated by comparing the average biofilm
thickness in different sections of the reactor and by visual
observations (data not shown). During the initial growth
phase the biofilm was homogeneously distributed through-
out the reactor. Random sloughing during quasi-steady-state
operation and the washout experiments resulted in a patchy
biofilm. Thus, detachment during the quasi-steady-state
phase and the washout experiments were not homoge-
neously distributed over the substratum area and the re-
ported values of ⌬LF represent detachment averaged over
the substratum area. Only during the homogeneous growth
phase can the biofilm thickness be regarded as nearly con-
stant along the reactor axis. In Figure 8, a conceptual model
is shown of the biofilm thickness development during the
three periods of Experiments II and III.
Figure 8. Detachment model for three phases of biofilm development.
Biofilm thickness is in the range of micrometers, the distance along the
substratum represents millimeters. LF,base and LF,detach are average thick-
nesses used in the presented mathematical model.
HORN ET AL.: SIMULATION OF GROWTH AND DETACHMENT IN BIOFILM SYSTEMS
Different approaches can be used to include these hetero-
geneous biofilm thickness distributions in mathematical
models. Multidimensional models could be used to explic-
itly describe these heterogeneous biofilm morphologies
(Noguera et al., 1999; Hermanowicz, 1999; Eberl et al.,
2000; Picioreanu et al., 2001). However, studies that are
based on macroscopic mass balances usually provide suffi-
ciently detailed experimental data to justify the use of these
complex multidimensional models. Based on one-
dimensional modeling some effects related to the heteroge-
neous biofilm thickness distribution can be predicted. Mor-
genroth and Wilderer (2000) evaluated the influence of lo-
cal dynamic detachment on competition within a biofilm by
integrating over time results from a one-dimensional model
with dynamic detachment in different intervals. Morgenroth
et al. (2000) used multiple one-dimensional biofilm com-
partments where mass transport in the heterogeneous bio-
film structure was modeled using convective links between
the one-dimensional biofilm compartments. Results of this
modified one-dimensional model were compared with
three-dimensional modeling results. The main question will
be to evaluate how far biofilm heterogeneity will determine
competition between different groups of microorganisms
within a biofilm (Morgenroth and Wilderer, 2000). In the
present study, the biofilm was based mainly on heterotro-
phic growth, and competition between different groups of
organisms was beyond the scope of this study.
CONCLUSIONS
The investigation and simulation of growth and detachment
in biofilm tube reactors show four main results: 1) At a
certain time during biofilm development the biofilm thick-
ness switched from a homogeneous growth phase to a quasi-
steady-state phase with a base biofilm thickness. 2) When
an increase in shear stress was applied to the biofilm, de-
tachment occurred in the first few seconds of the washout
experiment. The amount of detached biomass was found to
depend mainly on the biofilm thickness before the detach-
ment event. 3) During the quasi-steady-state and the wash-
out experiments, detachment resulted in a heterogeneous
patchy biofilm. 4) Due to similar biofilm densities and
growth rates in the conducted experiments, correlation with
these parameters are not detectable.
References
Bakke R. 1986. Biofilm detachment. Ph.D. thesis. Missoula: Montana State
University.
Bakke R, Trulear MG, Robinson JA, Characklis WG. 1984. Activity of
Pseudomonas aeruginosa in biofilms: steady state. Biotechnol Bioeng
26:1418–1424.
Benefield L, Molz F. 1985. Mathematical simulation of a biofilm process.
Biotechnol Bioeng 27:921–931.
Bouwer EJ, Zehnder AJB. 1993. Bioremediation of organic-compounds —
putting microbial-metabolism to work. Trends Biotechnol 11:
360–367.
615
Bryers JD. 1984. Biofilm formation and chemostat dynamics: pure and
mixed culture considerations. Biotechnol Bioeng 26:948.
Bryers JD. 1988. Modeling biofilm accumulation. In: Bazin M, Prosser JI,
editors. Physiological models in microbiology. Boca Raton, FL: CRC
Press. p 109–144.
Bryers JD. 2000. Biofilm formation and persistence. In: Bryers JD, editor.
Biofilms II: process analysis and application. New York: Wiley-Liss.
p 45–88.
Chadwick AJ, Morfett JC. 1993. Hydraulics in civil and environmental
engineering, 2nd ed. London: E & FN Spon.
Chang HT, Rittmann BE. 1987. Mathematical modeling of biofilm on
activated carbon. Environ Sci Technol 21:273–280.
Chang HT, Rittmann BE, Amar D, Heim R, Ehlinger O, Lesty Y. 1991.
Biofilm detachment mechanisms in a liquid-fluidized bed. Biotechnol
Bioeng 38:499–506.
Cooksey KE, Wigglesworth-Cooksey B. 1995. Adhesion of bacteria
and diatoms surfaces in the sea: a review. Aquat Microb Ecol 9:
87–96.
Dankert J, Hogt AH, Feijen J. 1986. Biomedical polymers, bacterial ad-
hesion, colonization and infection. CRC Crit Rev Biocompat 2:
219–301.
DEV. 1981. Deutsche Einheitsverfahren zur Wasser-, Abwasser- und
Schlammuntersuchung, Wasserchemische Gesellschaft. Fachgruppe in
der GDCh / in Gemeinschaft mit dem Normenausschu␤ Wasserwesen
(NAW) im DIN Deutsches Institut für Normung (Eds.). Weinheim:
Wiley-VCH. (German standard methods for examination of water,
wastewater and sludge.)
Eberl HJ, Picioreanu C, Heijnen JJ, van Loosdrecht MCM. 2000. A three-
dimensional numerical study on the correlation of spatial structure,
hydrodynamic conditions, and mass transfer and conversion in bio-
films. Chem Eng Sci 55:6209–6222.
Flemming H-C, Griebe T, Schaule G. 1996. Antifouling strategies in tech-
nical systems — a short review. Water Sci Technol 34:517–524.
Fruhen M, Christan E, Gujer W, Wanner O. 1991. Significance of spatial
distribution of microbial species in mixed culture biofilms. Water Sci
Technol 23:1365–1374.
Hermanowicz SW. 1999. Two-dimensional simulations of biofilm devel-
opment: effects of external environmental conditions. Water Sci Tech-
nol 39:107–114.
Horn H, Hempel DC. 1997. Growth and decay in an auto-/heterotrophic
biofilm. Water Res 31:2243–2252.
Keiding K, Nielsen PH. 1997. Desorption of organic macromolecules from
activated sludge: effect of ionic composition. Water Res 31:
1665–1672.
Kissel JC, McCarty PL, Street RL. 1984. Numerical simulation of mixed-
culture biofilm. J Environ Eng ASCE 110:393–411.
Kreft JU, Picioreanu C, Wimpenny JWT, van Loosdrecht MCM. 2001.
Individual-based modelling of biofilms. Microbiol Sgm 147:
2897–2912.
Kreikenbohm R, Stephan W. 1985. Application of a two-compartment
model to the wall growth of Pelobacter acidigallici under continuous
culture conditions. Biotechnol Bioeng 27:296–301.
Lee KC, Rittmann BE. 2000. A novel hollow-fibre membrane biofilm
reactor for autohydrogenotrophic denitrification of drinking water.
Water Sci Technol 41:219–226.
Morgenroth E, Wilderer PA. 1998. Modeling of enhanced biological phos-
phorus removal in a sequencing batch biofilm reactor. Water Sci Tech-
nol 37:583–587.
Morgenroth E, Wilderer PA. 1999. Controlled biomass removal — the key
parameter to achieve enhanced biological phosphorus removal in bio-
film systems. Water Sci Technol 39:33–40.
Morgenroth E, Wilderer PA. 2000. Influence of detachment mechanisms
on competition in biofilms. Water Res 34:417–426.
616 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 81, NO. 5, MARCH 5, 2003
Morgenroth E, Eberl HJ, van Loosdrecht MCM. 2000. Evaluating 3-D and
1-D mathematical models for mass transport in heterogeneous bio-
films. Water Sci Technol 41:347–356.
Murga R, Stewart PS, Daly D. 1995. Quantitative analysis of biofilm
thickness variability. Biotechnol Bioeng 45:503–510.
Nicolella C, Felice RD, Rovatti M. 1996. An experimental model of bio-
film detachment in liquid fluidized bed biological reactors. Biotechnol
Bioeng 51:713–719.
Noguera DR, Pizarro G, Stahl DA, Rittmann BE. 1999. Simulation of
multispecies biofilm development in three dimensions. Water Sci
Technol 39:123–130.
Peyton BM, Characklis WG. 1992. Kinetics of biofilm detachment. Water
Sci Technol 26:1995–1998.
Peyton BM, Characklis WG. 1993. A statistical-analysis of the effect of
substrate utilization and shear-stress on the kinetics of biofilm detach-
ment. Biotechnol Bioeng 41:728–735.
Picioreanu C, van Loosdrecht MCM, Heijnen JJ. 1998. A new combined
differential-discrete cellular automaton approach for biofilm model-
ing: application for growth in gel beads. Biotechnol Bioeng 57:
718–731.
Picioreanu C, van Loosdrecht MCM, Heijnen JJ. 2001. Two-dimensional
model of biofilm detachment caused by internal stress from liquid
flow. Biotechnol Bioeng 72:205–218.
Reichert P. 1994. Concepts underlying a computer program for the iden-
tification and simulation of aquatic systems (AQUASIM 1.0). Düben-
dorf: Schriftenreihe der EAWAG.
Reiff H. 1992. Wachstum und Abtrag der Sielhaut in Mischwasserkanälen
[Growth and detachment of biofilms in sewers]. Ph.D. thesis. Univer-
sity of Kassel.
Rittmann B. 1982. The effect of shear stress on biofilm loss rate. Biotech-
nol Bioeng 24:501–506.
Rittmann BE. 1989. Detachment from biofilms. In: Characklis WG,
Wilderer PA, editors. Structure and function of biofilms. New York:
Wiley. p 49–58.
Rittmann BE, Stilwell D, Ohashi A. 2002. The transient-state, multiple-
species biofilm model for biofitration processes. Water Res 36:
2342–2356.
Robinson JA, Trulear MG, Characklis WG. 1984. Cellular reproduction
and extracellular polymer formation by Pseudomonas aeruginosa in
continuous culture. Biotechnol Bioeng 26:1409–1417.
Sawyer LK, Hermanowicz SW. 1998. Detachment of biofilm bacteria due
to variations in nutrient supply. Water Sci Technol 37:211–214.
Speitel GE, DiGiano FA. 1987. Biofilm shearing under dynamic condi-
tions. J Environ Eng ASCE 113:464–475.
Stewart PS. 1993. A model of biofilm detachment. Biotechnol Bioeng
41:111–117.
Stewart PS, McFeters GA, Huang CT. 2000. Biofilm formation and per-
sistence. In: Bryers JD. editor. Biofilms II: process analysis and ap-
plication. New York: Wiley-Liss. p 373–405.
Stoodley P, Dodds I, Boyle JD, Lappin-Scott HM. 1999. Influence of
hydrodynamics and nutrients on biofilm structure. J Appl Microbiol
85:19S–28S.
Tanyolac A, Beyenal H. 1997. Prediction of average biofilm density and
performance of a spherical bioparticle under substrate inhibition. Bio-
technol Bioeng 56:319–329.
Tijhuis L, van Loosdrecht MCM, Heijnen JJ. 1995. Dynamics of biofilm
detachment in biofilm airlift suspension reactors. Biotechnol Bioeng
45:481–487.
Trulear MG, Characklis WG. 1982. Dynamics of biofilm processes. J
Water Pollut Control Fed 54:1288–1301.
Turakhia MH, Cooksey KE, Characklis WG. 1983: Influence of a calcium-
 specific chelant on biofilm removal. Appl Environ Microbiol 46:
1236–1238.
Tzeng JW, Fan LS, Gan YR, Hu TT. 1991. Ethanol fermentation using
immobilized cells in a multistage fluidized-bed bioreactor. Biotechnol
Bioeng 38:1253–1258.
Wang GTY, Bryers JD. 1997. A dynamic model for receptor-mediated
specific adhesion of bacteria under uniform shear flow. Biofouling
11:227–252.
Wanner O, Gujer W. 1984. Competition in biofilms. Water Sci Technol
17:27–44.
HORN ET AL.: SIMULATION OF GROWTH AND DETACHMENT IN BIOFILM SYSTEMS
Wanner O, Gujer W. 1986. A multispecies biofilm model. Biotechnol
Bioeng 28:314–328.
Wanner O, Reichert P. 1996: Mathematical-modeling of mixed-culture
biofilms. Biotechnol Bioeng 49:172–184.
Wäsche S, Horn H, Hempel DC. 2000. Mass transfer phenomena in Bio-
film Systems. Water Sci Technol 41:57–360.
Wilson M, Patel H, Noar JH. 1998. Effect of chlorhexidine on multi-
species biofilms. Curr Microbiol 36:13–18.
Zhang TC, Bishop PL. 1994. Density, porosity, and pore structure of
biofilms. Water Res 28:2267–2277.
617