You are on page 1of 33

Journal of Biotechnology 80 (2000) 1 – 33

www.elsevier.com/locate/jbiotec

Review article

Wastewater treatment with particulate biofilm reactors


C. Nicolella a,1, M.C.M. van Loosdrecht b,*, J.J. Heijnen b
a
Department of Food Science and Technology, The Uni6ersity of Reading, P.O. Box 226, Reading RG6 6AP, UK
b
Department of Biochemical Engineering, Kluy6er Institute for Biotechnology, Delft Uni6ersity of Technology, Julianalaan 67,
2628 BC Delft, The Netherlands
Received 22 November 1999; received in revised form 21 February 2000; accepted 25 February 2000

Abstract

The review presented in this paper focuses on applications of particulate biofilm reactors (e.g. Upflow Sludge
Blanket, Biofilm Fluidized Bed, Expanded Granular Sludge Blanket, Biofilm Airlift Suspension, Internal Circulation
reactors). Several full-scale applications for municipal and industrial wastewater treatment are presented and
illustrated, and their most important design and operation aspects (e.g. biofilm formation, hydrodynamics, mass
transfer, mixing) are analysed and discussed. It is clear from the review that this technology can be considered a
grown up technology for which good design and scale-up guidelines are available. © 2000 Elsevier Science B.V. All
rights reserved.

Keywords: Wastewater treatment; Biofilm reactor; Internal circulation; USB; EGSB; Fluidized bed; Airlift

1. Nomenclature De diffusion coefficient (m s−2)


DH dilution rate (day−1)
a specific surface area (m2 m−3) FC organic loading rate (kg day−1)
A area (m2) F/M food to microorganism ratio
C substrate concentration (kg (kgCOD kg−1SS )
m−3) Fr Froude number
CD drag coefficient g acceleration of gravity (m s−2)
d diameter (m) h height (m)
k0 zeroth order kinetic rate con-
* Corresponding author. Tel.: +31-152-781618; fax: + 31- stant (kg m−3 s−1)
152-782355.
kLa volumetric gas–liquid mass
E-mail addresses: c.nicolella@afnovell.reading.ac.uk (C.
Nicolella), mark.vanloosdrecht@stm.tudelft.nl (M.C.M. van transfer coefficient (h−1)
Loosdrecht) kLS liquid–solid mass transfer coeffi-
1
Fax: +44-118-9310080. cient (m h−1)

0168-1656/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 6 5 6 ( 0 0 ) 0 0 2 2 9 - 7
2 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

n single cells, in downstream processing, by facili-


N substrate flux (kg m−2 h−1) tating cell–liquid separation by sedimentation or
OTR oxygen transfer rate (kg m−3 filtration. The term floc is used to refer to an
day−1) assemblage of individual cells and micro-colonies
q overflow rate (m3 m−2 h−1) occurring under specific reactor conditions or af-
Q flow rate (m3 day−1) ter addition of various agents to the medium
Re Reynolds number (Boonaert et al., 1999). A biofilm can be defined
Sc Schmidt number as a complex coherent structure of cells and cellu-
Sh Sherwood number lar products, like extra-cellular polymers
u velocity (m s−1) (Characklis, 1990), which either form sponta-
V volume (m3) neously as large, dense granules (Lettinga et al.,
X biomass concentration (kg m−3) 1980), or grow attached on a static solid surface
Y biomass yield coefficient (static biofilms) or on suspended carriers (particle-
YO oxygen yield coefficient supported biofilms) (Heijnen, 1984).
Greek letters Microbial aggregates (either in the form of
a height to diameter ratio biofilms, granules or flocs) and the bulk culture
b reaction order transition charac- medium constitutes two distinct phases. This key
teristic [(Eqs. (12) and (13)] feature has three major consequences.
d biofilm thickness (mm) 1. Biomass retention can be used to improve
g settler to reactor volume ratio reactor volumetric conversion capacity when
o hold-up the conversion is limited by the amount of
m viscosity (kg m−1 s−1) biomass present. If no biomass retention is
mmax maximum specific growth rate applied (e.g. in the standard chemostat), the
(day−1) biomass concentration depends only on the
n viscosity (m s−2) substrate concentration in the feed, and conse-
r density (kg m−3) quently large retention times are required in
rS particulate biofilm density (kg the presence of diluted feeds. Depending on
m−3) the settling characteristics of the aggregates,
tH retention time (days) biomass can be readily separated (e.g. by sedi-
mentation) from the bulk liquid and retained
Subscripts
in the bioreactor. In this respect, granules and
c reactor
particle-supported biofilms have an extra ad-
d downcomer
vantage in that they can be more easily sepa-
f biofilm
rated than flocs (i.e. higher biomass
G gas
concentration possible) and have a high reac-
L liquid
r riser tor specific surface area (i.e. a large mass
S solid transfer area than static biofilms).
t terminal 2. The substrates (e.g. oxygen, carbon and nitro-
gen sources) have to cross the aggregate–liq-
uid interface and be transported through the
aggregate to reach the microbial cells and be
consumed. This transport is in general by dif-
2. Introduction fusion and results in a concentration gradient
within the aggregate. The penetration depth of
Microbial cell aggregates, such as flocs and substrates in biofilms mainly depends on the
biofilms, are of great interest in biotechnology. porosity of the biofilm, substrate concentra-
They offer advantages, with respect to suspended tion in the bulk liquid, mass transfer at the
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 3

biofilm–liquid interface and reaction rate in organisms (growth rate\ 0.1 h − 1) are used,
the biofilm. For poorly soluble substrate (e.g. there is no advantage in biomass retention.
oxygen) the penetration depth is shallow Sufficient biomass will be formed to metabolize
(typically 100–150 mm for oxygen) (Denac et the substrate within relatively short residence
al., 1983). times. For the majority of industrial fermenta-
3. Due to diffusional substrate concentration tion processes, where high substrate concentra-
gradients, a growth rate gradient also exists tions are used, biofilm formation is either
within the aggregate. In multi-species biofilm unnecessary or even disadvantageous, and the
systems this will lead to a biofilm with a range of applications of immobilized-cell systems
layered structure, where the organisms with in industry is limited. Several applications of
the highest growth rate will be found at the biofilm reactors in various biotechnological pro-
outside of the biofilm, whereas slower grow- cesses (e.g. fermentation, production of enzymes,
ing organisms will be found inside (Heijnen production of primary and secondary metabo-
et al., 1989). As a result of this organization, lites, production of antibiotics and bioconver-
slower-growing organisms will be protected sions), have been reviewed by Furusaki (1988),
from external shear forces, and are less likely Schugerl (1989, 1997), and Godia and Sola’
to be lost due to detachment and wash-out. (1995). However, with the exception of wastewa-
In this case not the absolute maximum ter treatment processes, the application of bio-
growth rate of the organisms should be con- film reactors at full industrial scale is scarce
sidered but the maximum growth rate under (Godia and Sola’, 1995).
conditions in the reactor (e.g. in the presence
An extensive use of biofilms is made within
of an inhibitor).
the field of environmental biotechnology for
The extent to which these features are rele-
three main reasons:
vant for a specific system depends, among other
1. compared with most other industrial biopro-
factors, on the physical and structural properties
cesses, large volumes of dilute aqueous solu-
(e.g. size, density, porosity, settling velocity, etc.)
tions have to be treated;
of the aggregates. Due to a smaller size (typi-
2. natural, mixed populations of microorgan-
cally between 10 and 150 mm) and higher poros-
ity, diffusional transport is generally faster in isms, which readily form biofilms, are used;
flocs than in granules or biofilm (whose size is 3. the process can be operated at high biomass
usually in the range 0.5 – 3 mm). The substrate concentration in the reactor, without the
concentration and biomass distribution gradients need for settlers for biomass retention and
are, therefore, less important in flocs than in recirculation. A polishing step of the effluent
biofilms. On the other hand, biofilms and gran- is usually needed to remove remaining sus-
ules present better settling properties than flocs pended (detached) biomass.
(terminal settling velocity of about 40 and 5 m Trickling filters have been in use in wastewa-
h − 1 for particle-supported biofilms and flocs, re- ter treatment processes for over a century, and
spectively), and can be more easily retained in their design and operation are well established
the bioreactors. The physical and structural in the practice of wastewater engineering (Met-
properties of particle-supported biofilms and calf and Eddy, 1991). The application of partic-
granules are similar, and so are their hydrody- ulate biofilm reactors is relatively recent, and,
namic, mass transfer and reaction characteris- though in most cases the principles of their de-
tics. For this reason, in the following sign and operation are known or can be derived
particle-supported biofilms and granules will be from other disciplines, there is a lack of system-
considered as a single category, the particulate atic information. This article covers relevant en-
biofilms (Fig. 1). gineering aspects of particulate biofilm reactors
When the substrate concentration in the feed and reviews several examples of full-scale appli-
is high ( \10 gCOD l − 1) and rapidly growing cations.
4 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

2.1. Application range static biofilms (e.g. in trickling filters), particulate


biofilms (e.g. in biofilm fluidized bed reactors,
Wastewater treatment processes are based on upflow anaerobic sludge blanket reactors and bio-
the use of three types of microbial aggregates: film airlift suspension reactors), and flocs (in acti-

Fig. 1. Particulate biofilms. (a) Aerobic granules (average diameter: 1.2 mm); (b) particle supported biofilms (average diameter: 1.7
mm).
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 5

Fig. 2. Concentration-flow rate phase diagram for application of floc and biofilm reactors.

vated sludge processes). Basic design criteria can The sludge is retained in the reactor without
be used to identify the range of operating condi- need for external separation and recycle.
tions suitable for the application of these aggre-
gates in wastewater treatment processes. Fig. 2
shows, in the concentration-flow rate plane, the 3. Applications of particulate biofilms in
lines corresponding to design criteria applicable environmental biotechnology
to different reactor configurations for the case
of industrial aerobic processes (details are given The main reactor types applicable for the sus-
in Appendix A). These lines define different re- pension of particulate biofilms in wastewater
gions of applicability in the C – Q phase dia- treatment processes are Biofilm Upflow Sludge
gram: Blanket (USB), Fluidized Bed (BFB), Expanded
A. Retention time is so long that microorgan- Granular Sludge Blanket (EGSB), Biofilm Airlift
isms grow in suspension. Suspension (BAS), and Internal Circulation (IC)
B. At high flow rates, particulate biofilms and reactors (Fig. 4). In USB, BFB and EGSB reac-
flocs are washed-out and only static biofilms tors (Fig. 4a,b,c, respectively), particles are kept
can be retained in the reactors, or the reac- fluidized by the up-flowing influent. In BAS re-
tors get a very flat and extended shape. actors (Fig. 4d) an airlift suspension is obtained
C. Flow and loading conditions are suitable for by pumping air into the system, whilst in IC
application of particulate biofilm reactors. reactors (Fig. 4e) the gas produced in the system
This range of applications is covered in the drives the circulation and mixing of liquid and
present review. solids in the reactor. The advantages and disad-
D. Flow and loading conditions are suitable for vantages of particulate biofilm reactors are pre-
applications of flocs, provided that separa- sented in Table 1 (Heijnen et al., 1989; Bryers
tion and biomass recycle are used (e.g. acti- and Characklis, 1990; Harremoes and Henze,
vated sludge processes). This region partially 1995; Wright and Raper, 1996). The basic de-
overlaps the particulate biofilm region. sign characteristics and some examples of indus-
E. For high strength and low flow wastewater, trial applications of particulate biofilm reactors
upflow sludge blanket reactors can be used. are given in Tables 2 and 3, respectively.
6 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

3.1. Historical o6er6iew Perhaps the most significant recent develop-


ment at the commercial-scale level is the use of
Although the development of water and upflow fluidized bed reactors containing granules
wastewater biological reaction systems using a and no bed media. The USB reactor was devel-
fluidized bed of biomass can be traced back to oped in The Netherlands in the late 1970s for
1940 in the UK (Mishra and Sutton, 1991), devel- anaerobic treatment of low-strength wastes, and
opment of particle-supported biofilm reactors did the first full-scale plant (200 m3) was installed in
not occur until the early 1970s. Researchers at 1978 at a sugar-beet factory in Halfweg (The
Manhattan College in New York were granted a Netherlands) (Lettinga et al., 1980). The EGSB
US patent in 1976 (Jeris et al., 1976) for the and the IC reactors are the most recent evolutions
application of Biofilm Fluidized Bed (BFB) pro- of the USB concept. Biothane Biobed EGSB
cess configuration to denitrifying wastewater. The (Lourens and Zoetemeyer, 1992) and Paques IC
first commercial-scale application for the BFB (Vellinga, 1986) systems operate at much higher
process configuration occurred at Pensacola liquid upflow velocities than the conventional
(USA) in the mid-1970s, with the installation of a USB.
system for the denitrification of nitrified munici- Fundamental and applied research on Biofilm
pal wastewater (Mishra and Sutton, 1991). The Airlift Suspension (BAS) reactors in the late 1980s
first commercial-scale application of the BFB (Heijnen, 1985; Heijnen et al., 1991, 1993), includ-
technology to industrial treatment occurred in the ing research at Gist-Brocades, TNO (Dutch Or-
late 1970s when an Ecolotrol HY-FLO system ganisation for Applied Scientific Research) and
was installed at a soft drink bottling plant in Delft University of Technology (The Nether-
Birmingham (USA) (Jeris, 1983). Since the early lands), led to the concept of CIRCOX airlift
1980s, the BFB process has been used on an reactor, developed and patented by Gist-Brocades
industrial scale for most conventional-sewage and commercialized by Paques.
treatment processes including carbonaceous oxi- High rate anaerobic pre-treatment followed by
dation (Jeris et al., 1977; Cooper and Wheeldon, aerobic post-treatment using biofilm reactors
1982), nitrification (Dunn et al., 1983; Nutt et al., (Heijnen et al., 1991) has been shown on numer-
1984), denitrification (Melcer et al., 1984; Nutt et ous occasions to be a very cost effective way to
al., 1984), and anaerobic treatment (Hickey and remove organic compounds from wastewaters
Owens, 1981; Jewell et al., 1981; Jeris, 1983). (Fig. 3). The IC anaerobic reactor followed by the

Table 1
Advantages and disadvantages of particulate biofilm reactors

Advantages Disadvantages

High terminal settling velocity of solids (50 m h−1; for Biofilm formation on carriers poses problems leading to long
flocculated sludge: 5 m h−1), leading to possible start-up times
elimination of external clarification/separation stages
High reactor concentration (30 kg m−3; for flocculated sludge Control of biofilm thickness is difficult
systems: 3 kg m−3)
High biofilm surface area (3000 m2 m−3; in trickling filters: Overgrowth of biofilms leads to elutriation of particles
300 m2 m−3)
High biomass concentration and mass transfer area result in Liquid distributors for fluidized systems are costly for
high conversion capacities (for oxygen, 20 kg m−3 day−1; large-scale reactors and pose problems with respect to clogging
in activated sludge and trickling filter processes: 3 kg m−3 and uniform fluidization
day−1)
Compact reactor with small area requirements
High biomass age (several weeks); and minimization of excess
sludge production
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33
Table 2
Some design characteristics of particle biofilm reactors

Reactor type Commercial names Flow pattern Liquid velocity H/D Mixing References

BFB HY-FLO, Ecolotrol (USA); Up-flow 10–30 m h−1 (upward) 2–5 Liquid Heijnen et al. (1989)
ANAFLUX, Degremont
(France); OXYTRON, ANY-
TRON, Dorr-Oliver (USA)
UASB BIOPAQ, Paques (The Nether- Up-flow 0.5–1 m h−1 (upward) 0.2–0.5 Gas produced Lettinga et al. (1980),
lands) Pereboom and Vereijken
(1994), Austermann et al.
(1999)
EGSB BIOBED, Biothane (USA) Up-flow 10–15 m h−1 (upward) 4–5 Liquid, gas pro- Zoutberg and Frankin (1996)
duced
IC IC, Paques (The Netherlands) Mixed Bottom: 10–30 m h−1; top: 4–8 3–6 Gas produced Pereboom and Vereijken
m h−1 (circulation) (1994), Gorur et al. (1995)
BAS CIRCOX, Paques (The Mixed 0.4–0.8 m s−1 (circulation) 4–5 Gas introduced Heijnen et al. (1993), Gorur et
Netherlands) al. (1995)

7
8
Table 3
Examples of full-scale applications of particle biofilm reactors for environmental control

System Development topics Examples of full-scale Design parametersa References


applications

BFB ANAFLUX, Degremont, Upflow anaerobic FBF using a Starch factory, Habourdin, N: 1; D: 6 m; OLR: 12 000 Holst et al. (1997)
France mineral support (BIOLITE France (1993) kgCOD day−1; RC: 60 kgCOD
R280) as fluid bed media to m−3 day−1; hCOD: 86%
treat a variety of brewery,
food-processing and paper

C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33


industry wastewater
BFB OXYTRON, ANYTRON, Carbonaceous oxidation, By-product coke plant, N: 2; D: 9 m; Hc 8.5 m; Sutton et al. (1999)
Dorr-Oliver, USA nitrification, denitrification and Ont., Canada (1996) OLR: 6690 kgCOD day−1; RC:
anaerobic reduction of 10.5 kgCOD m−3 day−1; Qw:
municipal and industrial 1104 m3 day−1
(automotive industry,
coke-making operations)
wastewater using sand and
activated carbon as fluid bed
media
UASB BIOPAQ, Paques, The Processes incorporating the Distillery, Wellington, N: 1; V: 450 m3; OLR: 6750 Wolmarans and Driessen
Netherlands sludge granulation concept for South Africa (1996) kgCOD day−1; RC: 15 kgCOD (1996), Wolmarans and Nell
the anaerobic treatment of high m−3 day−1; hCOD: \90% (1996)
strength wastewaters
EGSB BIOBED, Biothane, Processes combining the best Chemical factory N: 1; V: 220 m3; H: 17.3 m; Zoutberg and Frankin (1996)
USA features of the UASB and BFB (production of OLR: 2880 kgCOD day−1; RC:
technologies for the anaerobic formaldehyde from 13.1 kgCOD m−3 day−1; Qw:
treatment of high strength methanol), Rotterdam, The 144 m3 day−1; Qr: 2400 m3
wastewaters (wastewater from Netherlands (1993) day−1; uL: 6.3 m h−1; uG:
food, brewery, yeast, 0.65 m h−1; HRT: 2.7 h
pharmaceuticals, paper, chemical
industry)
IC Paques, The Netherlands System consisting of two UASB Inuline and fructose N: 1; V: 1100 m3; H: 22 m; Habets et al. (1997a,b)
reactor compartments on top of production from chicory OLR: 35 000 kgCOD day−1;
each other, one high loaded and beet, Rosendaal, The RC: 31 kgCOD m−3 day−1; X:
one low loaded. The gas Netherlands (1995) 31 kg m−3; hCOD: 70–80%
collected in the first stage drives
a gas lift and an internal
circulation. The system is used
for the anaerobic treatment of
high strength wastewaters
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33
Table 3 (Continued)

System Development topics Examples of full-scale Design parametersa References


applications

BAS CIRCOX, Paques, The Airlift technology with biomass Brewery, Enschede, The N: 1; V: 230 m3; H: 19 m; Gorur et al. (1995)
Netherlands on carrier (sand or basalt) app- Netherlands (1994) HRT: 1.3 h; Qw: 2245 m3
lied to pharmaceutical and day−1; F/M: 0.2 kgBOD kg−1
MLSS
brewery wastewaters which are day−1; RC: 4–14 kgCOD m−3
anaerobically pre-treated, and to day−1
municipal wastewater

a
N, number of units; D, column diameter; H, column height; V, reactor volume; OLR, organic loading rate; RC, reactor capacity; h, removal efficiency; HRT,
hydraulic retention time; uL, liquid velocity; uG, gas velocity; F/M, food to microorganism ratio; X, volatile suspended solid concentration; Qw, wastewater flow rate;
Qr, recirculation flow rate.

9
10 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

and passes upward through the dense anaerobic


sludge bed (biomass concentration: 60–70 kg
m − 3). Soluble COD is readily converted to
biogas, which is rich in methane, and an upward
circulation of water and gasborne sludge is estab-
lished. A settler section allows effective degassifi-
cation to occur. The dense, granular sludge
particles now devoid of attached gas bubbles, sink
back to the bottom, establishing a return down-
ward circulation. The upward flow of gasborne
sludge through the blanket combines with the
return downward flow of degassed sludge and
creates continuous convection. This insures effec-
tive sludge to wastewater contact. The design of
the reactor allows a highly active biomass concen-
tration and thereby high loading rate (10–15 kg
COD m − 3 day − 1), i.e. short hydraulic retention
time (less than 48 h for most applications).
Suspended solids in the influent, which accumu-
late in the reactors, pose a major problem to the
operation of USB and reduce the reactor capacity.
To overcome these limitations, the BFB and,
later, the EGSB and the IC concepts were
Fig. 3. Full scale Paques CIRCOX (front, 140 m3) and IC developed.
(back, 385 m3) reactors at a brewery in Brazil.
3.3. Biofilm Fluidized Bed (BFB) reactor
CIRCOX airlift aerobic technology has been re-
cently selected to treat the effluent from a large In Biofilm Fluidized Bed (BFB) reactors (Fig.
brewery in The Netherlands (Gorur et al., 1995). 4b), the liquid to be treated is pumped through a
After a year of operation, this innovative combi- bed of small media (typically sand with a particle
nation was reported to have overall total and size range of 0.2–0.8 mm) at a sufficient velocity
soluble COD removals of 80 and 93.5%, respec- to cause fluidization. In the fluidized state the
tively. Current volumetric loading rates have aver- media provide a large specific surface for attached
aged 14 kg m − 3 day − 1 in the IC and 10 kg m − 3 biological growth and allows biomass concentra-
day − 1 in the airlift system. tions in the range 10–40 kg m − 3 to develop
(Cooper and Sutton, 1983). For aerobic treatment
processes the reactor is aerated. This is done by
3.2. Upflow Sludge Blanket (USB) reactor recirculating the liquid from the reactor to an
oxygenator where air, or possibly oxygen, is bub-
The principle of operation of the Upflow bled (Cooper, 1981). To overcome problems re-
Anaerobic Sludge Blanket (USB) reactor (Fig. 4a) lated to high recirculation rates, needed when
(Lettinga et al., 1980) is similar to that of biofilm there is an high oxygen demand in the reactor, the
fluidized bed reactors, but hydrodynamic condi- reactor might be aerated directly (three-phase
tions are more quiescent due to lower superficial BFB reactor) (Fan et al., 1986; Tang and Fan,
liquid velocities. USB processes are based on the 1987; Trinet et al., 1991).
development of dense granules (1 – 4 mm) in the The BFB technology is typically most useful for
reactor. Wastewater enters the bottom of the reac- treatment of streams contaminated with organic
tor vessel through the inlet distribution system or inorganic compounds (e.g. ammonia) requiring
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 11

long solids residence time conditions (longer than OXYTRON (Sutton et al., 1980; Hoyland and
15 days) for biological oxidation or reduction and Robinson, 1983) and ANYTRON (Sutton and
low (less than 100 mg l − 1) concentrations of Mishra, 1990, 1994) are the commercial embodi-
suspended solids (Sutton and Mishra, 1990). ments of, respectively, the aerobic and anaerobic

Fig. 4. Biofilm reactor configurations. (a) USB; (b) BFB; (c) EGSB; (d) BAS; (e) IC.
12 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

BFB process configuration developed by Dorr- the upflow velocities for liquid (10 m h − 1) and
Oliver, Inc. Twelve aerobic BFB reactors were gas (7 m h − 1) approach those of the BFB system.
installed at General Motors manufacturing facili- EGSB reactors can be operated as ultra high
ties in the 1980s (Mishra and Sutton, 1991), and loaded anaerobic reactors (up to 30 kg COD m − 3
four anaerobic BFB reactors were started-up in day − 1) to treat effluents from chemical, biochemi-
1982 at a soy protein factory in Muscatine, IA cal and biotechnological industries (Zoutberg and
(Heijnen et al., 1989). Most recently, the BFB de Been, 1997). Biothane B.V. have built dozens
technology provided by Dorr Oliver was selected of Biobed EGSB units for various types of
for the treatment of the wastewater from by- wastewater (food, chemical, pharmaceutical in-
product coking operations with a phenolic load of dustry) in all parts of the world.
1120 kg day − 1 (Sutton et al., 1999). This BFB
system consists of two towers of 9 m diameter and 3.5. Biofilm Airlift Suspension (BAS) reactor
8.5 m height. The system includes provisions for
controlled pre-dissolution of high purity oxygen Airlift reactors (Chisti, 1989) consist of two
to the influent using an oxygen contactor and a connected sections, a riser and a downcomer.
means for monitoring and controlling expansions Different configurations are possible, including
of the fluidized bed because of biofilm growth on internal loop and external loop reactors. The prin-
the sand media. ciple of operation is the same for both configura-
Biothane B.V. has built several anaerobic two- tions. A gas is sparged at the bottom, moves
stage BFB plants for purification of their fermen- upward and exits at the top of the riser section. In
tation process wastewater (Heijnen et al., 1989) internal-loop airlift reactors, air may recirculate
treating wastes at COD volumetric loading rates through the downcomer section and provide aera-
in excess of 20 kg m − 3 day − 1. The first two tion throughout the reactor. The difference in
plants were built in the mid-1980s at Delft in The density between riser and downcomer, due to the
Netherlands and Prouvy in France. difference in gas hold-up, drives the liquid to
Degremont S.A. developed the ANAFLUX circulate between the two sections. When the liq-
process (Durot and Prevot, 1987; Bernard and uid velocity is sufficiently high, small particles will
Rovel, 1989; Oliva et al., 1990), an upflow anaero- be suspended and recirculated with the liquid.
bic reactor using a mineral bacterial support, in This results in a thorough mixing of both particles
the mid-1980s. The first industrial reactor was and liquid throughout the reactor. The airlift
built in 1986 for the treatment of brewery waste- technique has found two major applications in
water. Since then, Degremont S.A. has con- wastewater treatment processes, the Biofilm Air-
structed over 25 full-scale ANAFLUX reactors lift Suspension (BAS) reactor (Fig. 4d) for aerobic
treating, world-wide, a variety of industrial waste- treatment and the gas-lift reactor for anaerobic
waters (Holst et al., 1997). The system is highly treatment. The BAS technology was originally
efficient due to the large reactor biomass concen- developed for aerobic purification of anaerobi-
tration (30–90 kg m − 3) and to the intimate con- cally treated industrial wastewaters (Heijnen,
tact between biomass and substrate created by 1984; Heijnen et al., 1990, 1993). In anaerobic
high upflow liquid velocities (5 – 10 m h − 1). processes, the circulation in the reactor is induced
by recirculating the gas produced by the biochem-
3.4. Expanded Granular Sludge Bed (EGSB) ical reactions (gas-lift reactors; van Houten et al.,
reactor 1994, 1997).
CIRCOX, a commercial embodiment of the
The Expanded Granular Sludge Bed (EGSB) BAS reactor, is an aerobic system which ensures
reactor (Fig. 4c) (Frankin et al., 1992; Zoutberg high biological load (4–10 kg m − 3), short liquid
and Frankin, 1996) combines both characteristics residence time (0.5–4 h), high biomass settling
of USB and BFB processes. Biomass is present in velocities (50 m h − 1) and high biomass concentra-
a granular form, but conditions with respect to tion (15–30 kg m − 3). The CIRCOX airlift tech-
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 13

nology has been applied to treat municipal biodegradable COD is removed. The liquid veloc-
wastewater at a sewage plant in The Netherlands ity in this compartment normally varies from 2 to
(Frijters et al., 1997). Two CIRCOX reactors, an 10 m h − 1. The IC system consists of a slender
airlift reactor and an airlift reactor with integrated vertical reactor with a height of up to 25 m and a
denitrification, were used to treat pre-settled influ- relative small surface area.
ent. High BOD and nitrogen removal rates could Paques has several full-scale IC systems in oper-
be obtained resulting in a good effluent quality. ation in Europe. One of the largest systems is
Approximately 15 such systems are presently in located at a brewery in The Netherlands
operation. (Pereboom and Vereijken, 1994) and consists of
TURBOFLO is a biofilm reactor for secondary six IC reactors each 162 m3 in volume. The reac-
and tertiary wastewater treatment developed by tors are designed at a COD loading rate of 24 kg
Cie Lyonnaise des Eaux (France) using the con- m − 3 day − 1. The largest reactors (22 m high, 9.5
cept of an internal circulating airlift reactor m in diameter and approximately 1500 m3 in
(Lazarova and Manem, 1996; Lazarova et al., volume) are presently at ADM in Ireland and at
1997; Mousseau et al., 1998). The reactor com- Zhongya Chemicals in China.
prises a rectangular column filled with high den-
sity polyethylene granules (size: 0.5 – 2.5 mm;
density 860 kg m − 3). An industrial-scale reactor 4. Engineering aspects
was operated at a wastewater treatment plant at
Evry (France). 4.1. De6elopment of particulate biofilms
An industrial prototype of an external loop
airlift reactor, the BIOLIFT process, was devel- For a reliable operation of biofilm reactors,
oped by OTV S.A. (France) and installed at the high biomass concentrations, present in stable
Maxeville (France) wastewater treatment plant granular biofilms, are essential. For this reason,
(Badot et al., 1994). the carriers should be completely covered with
biofilms. However, these biofilms should be
3.6. Internal Circulation (IC) reactor smooth, otherwise the biofilm particles will be
easily washed out of the reactor.
The IC reactor (Fig. 4e) (Vellinga, 1986) con- Biofilm development is the difference between
sists of two USB-like reactor compartments on biofilm growth and attachment on one hand, and
top of each other, one highly loaded and one low detachment processes on the other. Biofilm devel-
loaded. The first reactor compartment contains an opment is influenced by various processes, includ-
expanded granular sludge bed, where most of the ing adsorption and desorption of microorganisms
COD is converted to biogas. The biogas produced to and from the solid surface, attachment of
in this compartment contains an expanded bed, is microorganisms to the surface, biofilm growth
collected by the lower level phase separator and is and biofilm detachment (Characklis, 1990). In
used to generate a gas lift by which water and steady state, the balance between biofilm growth
sludge are carried upward via the riser pipe to the and detachment determines the physical structure
gas/liquid separator on the top of the reactor. of the biofilm (van Loosdrecht et al., 1995), and
Here the biogas is separated from the water/ thereby the settling and fluidization characteris-
sludge mixture and leaves the system. The water/ tics.
sludge mixture is directed downwards to the
bottom of the reactor via the concentric downer 4.1.1. Biofilm formation
pipe, resulting in the internal circulation flow. In Heijnen (1984) postulated the hypothesis that,
the lower section, the upward liquid velocity in ideally mixed reactors, formation of biofilm
varies from 10 to 20 m h − 1. The effluent from the covered carriers only takes place if the hydraulic
first compartment is post-treated in the second, retention time is less than the inverse of the
low loaded compartment, where remaining maximum growth rate (the dilution rate is larger
14 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

than the maximum growth rate). Experimental 4.1.2. Biofilm detachment


evidence for this hypothesis was produced by Biofilm detachment, the interphase transport of
Tijhuis et al. (1994) for the formation of hetero- biomass from an attached microbial film to the
trophic aerobic biofilms on small suspended bulk liquid phase, has generally been attributed to
basalt particles in airlift reactors. These authors four different processes (Characklis, 1990), includ-
also showed that, during biofilm development in ing grazing (the consuming of bacteria from the
the start-up of BAS reactors, more than 95% of outer surface of the biofilm by protozoa), slough-
the attached biomass production is detached to ing (the periodic loss of large patches of biofilm),
the liquid phase and washed out of the reactor. erosion (the continuous removal of small particles
The amount of detached biomass, and therefore, from the surface of the biofilm, primarily caused
biofilm accumulation, is strongly influenced by by liquid shear stress), and abrasion (analogous to
abrasion processes in the reactor, which are erosion, but caused by collisions of particles).
mainly due to particle interactions. It was found Among the mechanisms controlling biofilm reac-
for the BAS reactor that the rate of biofilm spe- tor performance, biofilm detachment is one of the
cific surface detachment rate is a function of the least studied and understood. The biofilm detach-
bare carrier concentration, independent of the ment rate is a complicated function of many
surface substrate loading rate and biofilm thick- variables, including hydrodynamics of the liquid
ness. Because the bare carrier concentration de- flow, biofilm morphology, and support
creases during the start-up, abrasion and characteristics.
The mechanisms controlling biomass loss have
therewith detachment rate decreases. This results
been investigated for different biofilm reactors,
in a strong non-linear increase of the biofilm
including solid/liquid BFB reactors (Chang et al.,
thickness with time.
1991; Nicolella et al., 1996), three-phase BFB
The influence of carrier type on adhesion and
reactors (Rittmann et al., 1992), and BAS reactors
biofilm formation of pure and mixed cultures was
(Tijhuis et al., 1995b; Gjaltema et al., 1995,
studied by Gjaltema et al. (1997a) using sus-
1997b,c).
pended carriers (standard, roughened, hydropho-
Chang et al. (1991) made direct measurements
bic and positively charged glass beads, sand and
for the specific biofilm loss rate coefficient and the
basalt grains) in laboratory airlift reactors. The total biofilm accumulation in a laboratory BFB
results clearly show that in airlift reactors hydro- reactor with varying liquid velocities and support
dynamic conditions and particle collisions control particle concentrations. Multiple regression analy-
biofilm formation. Increased surface roughness of sis of the results showed that increased particle-to-
the carriers promoted biofilm accumulation on particle attrition and increased turbulence caused
suspended carriers, whilst the physico-chemical the biofilm to be denser and thinner. The specific
characteristics of the carrier surface proved to be detachment rate coefficient increased as inert par-
less important. ticle concentration and particle Reynolds number
It was shown that the biofilm morphology is (i.e. turbulence) increased, and the turbulence and
influenced to a large extent by the surface sub- attrition of bed fluidization appeared to be domi-
strate loading and applied detachment forces (Ti- nant detachment mechanisms. Similar results are
jhuis et al., 1995a; Kwok et al., 1998). A moderate reported by Gjaltema et al. (1995) for the detach-
surface substrate loading and a high detachment ment of biomass from suspended biofilm pellets in
force yielded smooth and strong biofilms, whilst the presence of bare carrier particles in BAS
the combination of a high surface substrate load- reactors under non-growth conditions. The de-
ing and low detachment forces led to rough bio- tachment rate was dominated by collisions be-
films. The strength of biofilms also appeared to be tween bare carrier particles and biofilm pellets,
related to the detachment forces applied during whilst hydrodynamic conditions and substrate
biofilm formation, in combination with the sub- loading rates have a minor effect (Tijhuis et al.,
strate loading. 1995b). The collisions between biofilm pellets and
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 15

bare carrier particles cause an on-going abrasion Successful design and operation of particle type
of the biofilm pellets, leading to a reduction in biofilm reactors rely on information on settling
pellet volume, whilst the breakage of the biofilm and fluidization characteristics, such as fluidized
pellets is negligible (Gjaltema et al., 1997b). These bed-height as a function of liquid velocity. Infor-
findings were confirmed by Kwok et al. (1998) for mation on fluidized-bed height is important be-
growing biofilm suspensions. cause it establishes the solid’s residence time and
The experimental data on biofilm detachment the specific biofilm surface area in the biologically
rate in BFB reactors reported by Nicolella et al. active zone. The relationships valid for fluidiza-
(1996) showed that the specific detachment rate tion of rigid particles, readily available in the
coefficient strongly increases with increasing liq- chemical engineering literature (e.g. Di Felice,
uid velocity. Other parameters, such as particle 1995), have to be modified to take into account
concentration and liquid shear stress, were found the effect of the biofilm layer.
to be less significant. Nicolella et al. (1996) found
that the biofilm detachment rate is very low under 4.2.1. Terminal settling 6elocity
the flow conditions used in their experiments, The terminal settling velocity of a single spheri-


which are comparable to those of many industrial cal particle in an infinite expanse of fluid is:
applications of fluidized bed technology in
wastewater treatment processes (e.g. liquid veloc- ut =
4g(rS − rL) n
0.5
(1)
3CDrL
ities in the range 1– 10 mm s − 1).
The high detachment rate caused by bare carri- Depending on the biofilm thickness and on the
ers has some consequences for BAS reactor opera- carrier type, values for the equivalent density of
tion: while bare carrier are necessary for biofilm biofilm particles (rS) range typically from 1100 to
formation, they can at the same time severely 1500 kg m − 3. The drag coefficient CD is a func-

 
hinder or delay biofilm formation, especially dur- tion of the particle Reynolds number
ing the start-up phase. On the other hand, bare
rLdSut
carrier particles can be used to prevent excessive Ret =
mL
biofilm formation, which can lead to reduced
reactor performance. Contrary to airlift reactors Biofilm particles are in the intermediate flow
where solids are completely mixed in the reactor, regime (1B Ret B 100) for the vast majority of
in BFB, bare particles will accumulate at the cases, invariably obtained when sand (0.5–1 mm)
bottom of the bed, whilst thick fluffy biofilms will or similar material is used as inert support. In the
accumulate at the top of the bed (Di Felice et al., intermediate flow regime, the drag coefficient for
1997). In BFB, control of biofilm development by a smooth, rigid sphere is (Perry and Green, 1997):
particle shear is therefore not as easy as in an
CD = 18.5Ret− 0.6 (2)
airlift reactor.
Biofilm particles are of nearly spherical shape,
but they are not smooth or rigid, and therefore
4.2. Hydrodynamic characteristics of particulate Eq. (2) cannot be used to calculate their settling
biofilms velocity. For this reason, other empirical correla-
tions have been suggested for the estimation of
A sound knowledge of the hydrodynamic be- CD for biofilm particles. A selection of these
haviour of biofilm-coated particles is an essential correlations is presented in Table 4 and plotted in
step for the correct design of particle-supported Fig. 5. The drag coefficient of the biofilm particles
biofilm reactors. In these systems, biofilm growth is clearly higher than that of smooth, rigid
alters particle size, apparent density, shape and spheres, also reported in Fig. 5, over the whole
roughness, and all these factors have a strong range of Ret investigated.
influence on the hydrodynamic behaviour of the Nicolella et al. (1999b) could estimate the ter-
reactors. minal settling velocity of particles used in their
16 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

Table 4
Selected correlations for CD, n and ui for biofilm particles

Reference Ret CD n ui

Ngian and Martin (1980) 4.4Re−0.1


t ut
Hermanovicz and Ganczarczyk (1983) 50–100 17.1Re−0.47
t
Thomas and Yates (1985) 20–100 30Re−0.505
t ut
Mulcahy and Shieh (1987) 40–90 36.66Re−0.67
t 10.35Re−0.18
t ut
Harada et al. (1987) 10–50 8.733Re−0.341
t ut
−0.21
Hermanovicz and Cheng (1990) 50–100 9.11Ret ut
24
Ro and Neethling (1990) 15–87 +21.55Re−0.518
t
Ret
24
Yu and Rittmann (1997) 40–90 +14.55Re−0.48
t ut
Ret 4.526Re−0.0126
t
Nicolella et al. (1999b) 7–90 29.6Re−0.6
t 4.4Re−0.1
t 0.8ut

work and those reported by other authors with an Richardson and Zaki (1954) showed that the
average error of 10% by: extrapolation of the velocity to o= 0, ui, and the
single particle settling velocity, ut, are virtually
(CD)biocovered coincident for systems where the particle diameter
=1.6 (3)
(CD)clean is considerably smaller than the column diameter.
The expansion index n was found to be a function
Surface roughness has generally been invoked
of the flow regime only, and, for the range of
as the reason for the increase in the drag coeffi-
Reynolds number of interest in the case of biofilm
cient of biofilm particles (Hermanovicz and
particles (1BReB 500), Richardson and Zaki
Ganczarczyk, 1983; Mulcahy and Shieh, 1987).
(1954) proposed:
Ro and Neethling (1990) introduced a form factor
and a shape factor. The experimental value of
both parameters, however, was very close to one, n= 4.4Ret− 0.1 (5)
so that their effective contribution was not tested.
Nicolella et al. (1999b) found that the ratio of
drag coefficient for biofilm particles to drag coeffi-
cient for smooth rigid solid is independent of
biofilm thickness, and concluded that particle de-
formability has a negligible effect on CD. They
also showed that as the Reynolds number de-
creases (up to 0.001), the experimental measure-
ments for CD become closer to the correlation for
rigid smooth particles, thereby indicating that the
surface roughness indeed plays an dominant role
in the determination of CD.

4.2.2. Expansion characteristics


The empirical equation of Richardson and Zaki
(1954) is widely used to describe the bed expan-
sion characteristics of concentrated suspensions of
rigid spherical particles:
uL
o nL = (4) Fig. 5. Selected correlations for drag coefficient of particulate
ui biofilms.
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 17

their findings for the two numerical parameters


characterising the expansion law (Eq. (4)),
Nicolella et al. (1999b) were able to predict the
overall bed height of the biological fluidised bed
as a function of the fluidising velocity for systems
whose physical parameters are reported in litera-
ture (Ngian and Martin, 1980; Mulcahy and
Shieh, 1987).
Experimental data on bed expansion character-
istics can be used in combination with the empiri-
cal equation of Richardson and Zaki (1954) (Eq.
(4)) to give an estimate of biofilm thickness
(Nicolella et al., 1995) and reactor biomass con-
centration (Nicolella et al., 1997) in BFB reactors
through a non-linear parameter estimation rou-
tine. This unique estimation does not require any
measurement other than bed height versus liquid
velocity, and constitutes a feature of BFB sys-
Fig. 6. Selected correlations for the expansion index of partic- tems, as in other biofilm systems it is impossible
ulate biofilms.
to give an estimation of biofilm thickness and
reactor biomass concentration without measuring
For BFB, the experimental evidence published these variables experimentally.
so far (Andrews and Tien, 1979; Ngian and Mar-
tin, 1980; Mulcahy and Shieh, 1987; Hermanovicz 4.2.3. Effects of biofilm o6ergrowth
and Cheng, 1990; Csizkor et al., 1994; Nicolella et A problem with particulate biofilm reactors is
al., 1999b) suggests that bed expansion character- that the biofilm has a tendency to change in
istics can still be represented by an equation like density, thickness, and shape over the cycle of
that of Richardson and Zaki (i.e. a linear relation- operating life. The major result of this change is
ship in logarithmic co-ordinate between voidage that the mass transfer, reaction and hydrody-
and liquid velocity). However, the dependence of namic characteristics of the bed alter over time.
the parameters n and ui on the system’s physical All these effects can dramatically influence the
characteristics has yet to be fully established. Fig. operation of the unit. In steady state operation,
6 reports the estimates of the expansion index n as biomass would tend to grow continuously around
a function of the terminal Reynolds number as the inert support, and this growth would be only
suggested by different authors for biofilm parti- partially balanced by loss due to liquid shear and
cles. The values obtained by Nicolella et al. particle attrition. Since the biomass has a density
(1999b) are well in line with those found by usually lower than the support (typical value for
Richardson and Zaki (1954) for liquid beds of the wet density range around 1100 kg m − 3), the
smooth rigid particles but much smaller than the increase in biofilm thickness due to biological
estimates reported for biological beds (e.g. growth modifies the mixing and fluidization char-
Thomas and Yates, 1985; Mulcahy and Shieh, acteristics, resulting in an expansion of the bed.
1987; Hermanovicz and Cheng, 1990; Yu and The overall particle density reduction can cause
Rittmann, 1997). Nicolella et al. (1999b) found carry over of the particles, leading to problems
that the ratio between ui and ut for the particle such as loss of particles and active biomass into
considered in their experimental work, is around pumps (Wright and Raper, 1996). Hence particles
0.8, in contrast with the work of Richardson and need to be filtered out, the biofilm removed and
Zaki (1954), but in agreement with a more re- the clean particles returned to the reactor. This
cently published study (Di Felice, 1995). Using practice requires additional capital expenditure on
18 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

particle separation equipment such as vibrating portance. Reactor geometry (height, diameter,
screens (Shieh et al., 1981; Fan, 1989). In response draught tube dimensions), gas supply rate and
to biofilm overgrowth, a number of different reac- solids loading and properties will all determine
tor configurations have been proposed to enable the liquid circulation velocity and therewith mix-
the removal of the excess biofilm within the ing, gas hold-up and solid suspension.
fluidized bed (Wright and Raper, 1996). This is
done by creating areas of high turbulence on the 4.3.1. Liquid circulation
top of the reactor to shear the biofilm off the Several hydrodynamic models have been devel-
support particles. As a result, two solid phases are oped to derive theoretical relationships between
present at the same time in the reactor: one fully the gas hold-up, circulation pressure drop, and
covered with the biofilm and the other where the liquid velocity in two-phase airlift reactors (Bello
biofilm is thin or even absent. Due to different et al., 1985; Chisti et al., 1988; Siegel and
hydrodynamic characteristics, these two solid Robinson, 1992; Ayazi Shamlou et al., 1994).
phases tend to segregate within the bed. These models were derived using either energy
balances on the gas input or macroscopic momen-
4.3. Flow regimes in airlift suspension reactors tum balances. A key parameter is the total fric-
tional resistance to fluid flow in the circulating
In BAS reactors, gas sparging is used to keep liquid loop, including the flow reversal at the top
the biofilm particles in suspension. In most full- and bottoms of the columns. Friction coefficients
scale applications, this is achieved by using an have been estimated by applying standard rela-
internal draught tube (riser). The riser is sparged tionships for single-phase flow and two-phase
with gas, and the resulting gas hold-up difference flow in pipes and bends, or by experimental fitting
between the gas-sparged riser and the unsparged (Siegel and Robinson, 1992).
downcomer leads to a difference in the bulk densi- Most of the hydrodynamic models of three-
ties of the fluids in the two zones and hence an phase airlift reactors concern external loop sys-
induced fluid circulation (upflow in the riser, tems (Douek et al., 1994), systems with small solid
downflow in the downcomer) is realised. The particles of density close to that of water (whose
magnitude of liquid circulation is one of the most flow behaviour is similar to that of two-phase
important design and scale-up parameters for air- systems, Lu et al., 1995; Hwang and Cheng,
lift reactors: liquid circulation influences the gas 1997), or lab-scale systems (where the hydrody-
hold-up in the vessel, mass transfer coefficients, namic regime 1 and 2 predominate, Livingston
the extent of mixing in the reactor and the pre- and Zhang, 1993). Heijnen et al. (1997) have
vailing flow regimes for gas, liquid and solid developed a model based on a momentum balance
phases (Chisti et al., 1988; Heijnen et al., 1997). to predict data measured in large scale internal-
Depending on liquid circulation velocity, different loop airlift reactors (pilot-scale, 0.4 m3, and full-
flow regimes exist with respect to gas bubble scale, 300 m3) with high gas recirculation through
entrainment into the downcomer. For aerobic sys- the downcomer (regime 3). Garcia-Calvo et al.
tems, the liquid velocity should be sufficient to (1999) presented a similar model for the descrip-
ensure entrainment of bubbles into the down- tion of flow behaviour in three-phase airlift reac-
comer and from the downcomer into the riser tors, which is based on an energy balance. This
again. This gas recirculation is desired to achieve model predicts liquid and solid circulation veloc-
maximal aeration (gas hold-up). In addition, there ity, gas hold-up and solid distribution within the
is a minimal gas supply rate below which the reactor, and describes the three solid flow regimes
liquid circulation velocity is insufficient to main- (packed bed, fluidized bed, and complete recircu-
tain a stable particle suspension. The liquid circu- lation). Experimental data obtained by various
lation rate should at least be well above the authors for reactors of different shape (internal
settling velocity of the particles. Knowledge of the and external loops) and size (0.02–284 m3), differ-
liquid circulation rate is therefore of obvious im- ent solid particles (sizes from 0.1 to 3 mm and
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 19

densities from 2550 to 3700 kg m − 3) and different The behaviour of BAS reactors in the fluidized
experimental conditions are simulated regime is similar to that of three-phase BFB reac-
satisfactorily. tors. The boundary between the fluidized bed and
van Benthum (1998) has presented a new reac- the circulating bed regimes is established by the
tor concept, the biofilm airlift suspension exten- condition:
sion, for simultaneous nitrification and
uLr ] ut (7)
denitrification of wastewater. The reactor consists
of a three-phase (biofilm particles, water and air) Riser liquid velocities higher than the particle
internal-loop airlift reactor for nitrification, which settling velocity result in solid recirculation
is extended with a two-phase (biofilm particles through the downcomer. As the circulating bed
and water) external concentric downflowing bed, regime is the normal operation mode, the opera-
the extension for denitrification. Due to the spe- tional conditions to reach this regime are among
cial design of the reactor, the liquid velocity in the the main design criteria for BAS reactors.
extension can be manipulated by the overpressure
in the headspace of the aerobic compartment, 4.3.3. Gas flow regimes
therewith controlling the liquid recirculation ratio Based on observations at different scales
between aerobic compartment and extension (van (10 − 3 –102 m3), three flow regimes have been
Benthum et al., 1999a, 2000). noted for three-phase airlift reactors when gas
velocity increases (Heijnen et al., 1997):
4.3.2. Solid flow regimes 1. No gas entrainment in the downcomer. This
Three solid regimes are reported in the litera- occurs at low gas velocities when the liquid
ture (Douek et al., 1994) for three phase airlift circulation velocity is insufficient to entrain
systems. These are, with increasing gas flow rate, bubbles in the downcomer. The downcomer
packed bed (settled solid), fluidized bed (no solids liquid velocity is lower than the bubble swarm
recirculation) and circulated bed (complete sus- velocity.
pension regime). In the packed bed regime, the 2. Gas entrainment in the downcomer, but no
riser may be assumed as a gas – liquid bubble gas circulation. When the liquid velocity in the
column, while the downcomer contains static liq- downcomer becomes about equal to the bub-
uid. This condition is undesirable for operation of ble swarm velocity, gas is entrained into the
biofilm reactors. The condition to suspend the downcomer. Often an axial bubble distribution
solid settled in the bottom may be written as is present in the downcomer. Complete solids
(Garcia-Calvo et al., 1999): suspension is commonly but not always

 
present.
rS 3. Complete gas circulation. All the gas bubbles
oG − − 1 o 0S ]0 (6) that are entrained into the downcomer are
rL
carried all the way down the downcomer and
The initial solid distribution is decisive in the into the riser again. There is no axial bubble
change from packed to fluidized or circulating bed distribution in the downcomer. The liquid ve-
regimes. When the gas velocity, uG, is increased locity in the downcomer is significantly higher
from zero to the minimum value needed for than the bubble swarm velocity, which leads to
fluidization, uGmf, o 0S corresponds to the solid ini- a large downward gas flow in the downcomer.
tially settled in the bottom of the riser. If uG The riser gas flow rate, combining the sparged
decreases from high values to reach uGmf, the gas and the recirculated gas, is significantly
condition of Eq. (6) is fulfilled when the apparent higher than the injection flow rate. Complete
densities in the riser and downcomer are equal, solids suspension is common, but incomplete
and o 0S = oSr −oSd. The different o 0S values corre- suspension cannot be excluded.
sponding to these two alternatives are related to Fig. 7 compares the gas hold-up measured for a
the hysteresis effect (Heck and Onken, 1988). two-phase lab-scale airlift reactor by Nicolella et
20 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

al. (1998a) and Bakker et al. (1993). Both sets of 4.4. Mass transfer
data show the same trend. The slope of the graph
of oG versus uG changes drastically on two occa- Particle-supported biofilm reactors can be
sions, and the curves present a platform, corre- classified as three-phase gas–liquid–solid reac-
sponding to the transition between the tors. The solid phase is constituted by biofilm
hydrodynamic regime of no gas entrainment in particles suspended in the bulk liquid. The gas
the downcomer (regime 1) and complete gas recir- phase is constituted by air (or oxygen) in aerobic
culation (regime 3). Such transitions are indicated systems or biochemical reaction products (CH4,
by vertical lines in Fig. 7. Data on liquid circula- CO2, H2S, and N2) in anaerobic systems, and
tion velocity reported by Bakker et al. (1993) are moves in the form of dispersed bubble through
also plotted in Fig. 7, where it can be observed the bulk liquid. In aerobic systems, oxygen is
that the liquid circulation velocity changes drasti- transferred from the air bubbles through the bulk
cally when the flow regimes changes from 2 to 3, liquid to the biofilm particles, where it is con-
remaining almost constant in regime 2 and rapidly sumed by biochemical reactions (Fig. 2). Gas–liq-
increasing with increasing gas velocity in regime 3. uid (G–L) and liquid solid (L–S) mass transfer
Similar results are reported by van Benthum et al. coefficients (MTCs) are, together with reaction
(1999b) for a pilot-scale (1 m3) internal-loop airlift kinetic parameters, important design parameters
reactor with two phases (water and air) and three of gas–liquid–solid bioreactors for biochemical
phases (water, air and polystyrene particles). applications.
In general, the importance of regimes 1 and 2 Reliable techniques exist for the estimation of
decreases with increasing scale. In large-scale re- the volumetric G–L MTC (kLa) and kinetic
actors regime 1 is hardly ever realised, and the parameters in biofilm reactors. The well known
transition from regime 2 to regime 3 depends transient gassing technique (Fuchs et al., 1971) is
strongly on sparger design and reactor geometry often used to measure kLa; this technique was
(Heijnen et al., 1997). recently applied specifically to BAS reactors
(Nicolella et al., 1998b). The biofilm kinetics can
be measured by means of various techniques such
as the measurement of the oxygen uptake rate
(Tijhuis, 1994). A new method has been recently
presented for the measurement of L–S MTC in
biofilm systems (Nicolella et al., 1998b, 1999a).
This method is based on the analysis of oxygen
consumption rate data in biofilm reactors and
biological oxygen monitoring systems (Tijhuis,
1994), and allows the independent estimation of
G–L MTC, L–S MTC and biofilm reaction rate
parameters.

4.4.1. Gas hold-up and gas–liquid mass transfer


The determination of the gas hold-up and volu-
metric G–L MTC is an important factor in the
design and operation of three-phase biofilm reac-
tors and has been the subject of much research
interest. Extensive data are available for the esti-
mation of these design parameters in two-phase
G–L systems (see for example Chisti, 1989 and
Fig. 7. Gas hold-up and circulation velocities for different flow Bello et al., 1985 for airlift reactors; Linek et al.,
regimes in airlift reactors. 1987 for aerated agitated vessels; Shah et al.,
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 21

1982, Heijnen and van’t Riet, 1984, and Deckwer Ryhner et al. (1988) reported that the gas–liquid
and Schumpe, 1993 for bubble columns) and in volumetric mass transfer coefficient in a three-
G – L slurry reactors (Chaudari and Ramachan- phase biofilm fluidized sand bed reactor decreased
dran, 1980; Pandit and Joshi, 1986; Siegel and (in the range 0.02–0.04 s − 1) with increasing
Robinson, 1992; Beenackers and van Swaaij, amounts of clean sand and was almost indepen-
1993), where the liquid and solid phases are con- dent of the sand fraction with biofilm-covered
sidered as one pseudo-homogeneous phase. For sand. Direct information on biofilm airlift suspen-
the specific case of three-phase gas – liquid – solid sion reactors is provided by Nicolella et al.
reactors, there is still no universally applicable (1998a). The gas hold-up in aerated suspensions
relation describing the influence of all types of of solid particles is generally reported to decrease
particles in any weight fraction in any liquid and significantly with increasing solids loading. Miya-
for any reactor scale (Beenackers and van Swaaij, hara and Kawate (1993) showed that for low-den-
1993). sity particles the gas hold-up decreased
The volume fraction of gas in gas – liquid – solid significantly when the solid hold-up was larger
dispersions (gas hold-up, oG) has a strong influ- than about 0.2. Some of the available information
ence on the performance of pneumatic biofilm on gas hold-up and G–L mass transfer in three-
reactors, such as the BAS and the IC reactors. phase sparged reactors is reported in Table 5.
The residence time of the gas in the liquid, the Empirical (Koide et al., 1985) and theoretical
gas – liquid contact area for mass transfer and the (Livingston and Zhang, 1993; Lu et al., 1995)
design volume of the reactor depend on the gas models have also been proposed to describe the
hold-up which occurs under given operating con- hydrodynamics of small-scale systems (0.001–0.01
ditions. In addition, the gas hold-up in conjunc- m3). Only recently, Heijnen et al. (1997) proposed
tion with the knowledge of mean bubble diameter a model for the hydrodynamic behaviour of large
allows the determination of the interfacial area scale three-phase airlift reactors (0.1–500 m3).
and thus leads to the mass transfer rate between The effect of solid size should be mainly related
the gas and the liquid phase. In pneumatic reac- to the effect of the particle terminal settling veloc-
tors, the principal operating influence on gas ity (ut). The particle terminal settling velocity has
hold-up is the gas velocity in the vessel. Conse- a direct influence on the difference in solid hold-
quently the effect of gas velocity on gas hold-up up between the riser and the downcomer (DoS). In
and G–L MTC has been much investigated (Shah turn, the difference DoS strongly influences the
et al., 1982; Heijnen and van’t Riet, 1984; Chisti hydrodynamics of the system: if the solid hold-up
and Moo-Young, 1987, 1988; Fan et al., 1987). in the riser is larger than in the downcomer, the
Hydrodynamics and mass transfer have been presence of solids lowers the driving head of the
studied extensively for gas – liquid – solid fluidiza- system, and thereby the liquid circulation rate and
tion (Muroyama and Fan, 1985), and to a less the gas recirculation, i.e. the gas residence time
extent for three-phase pneumatic reactors. Most (Heijnen et al., 1997). Nicolella et al. (1998a)
of the papers on the hydrodynamics and mass measured the gas hold-up in a concentric tube
transfer of three-phase gas – liquid – solid reactors airlift reactors in the presence of basalt and bio-
have dealt with systems of relatively small scale. film particles of different sizes. The terminal set-
The influence of the presence of solids has been tling velocities of basalt particles varied in a wider
reported for various types of particles including range (19–129 mm s − 1) than those of biofilm-
glass beads (Koide et al., 1985), plastic beads coated particles (30–49 mm s − 1), and the influ-
(Miyahara and Kawate, 1993), polystyrene cylin- ence of particle size was more noticeable in the
ders (Hwang and Lu, 1997), activated carbon first case than in the latter. In both cases, the gas
particles (Muroyama et al., 1985), Raney nickel hold-up decreases with increasing particle settling
particles (Gavroy et al., 1995), calcium alginate velocity. The observed influence of solid size on
beads (Lu et al., 1995), and basalt (Nicolella et al., the hydrodynamics is consistent with that found
1998a). For the case of biofilm coated particles, by other authors for similar systems (Koide
22
Table 5
Selected gas hold-up and G–L mass transfer correlations for three-phase (gas–liquid–solid) sparged reactors

System Range of variables Gas hold-up correlationa G–L mass transfer correlation(*) Reference

AL. L, water, electrolyte, glycerol dS =0.08–0.2 mm;  n, n


oG oG
  
CS 0.069
Koide et al. (1985)

C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33


rS =4680–8770 kg m−3;
       n
and ethylene glycol aqueous (1−oG)4 (1−oG)4 [kLa]/[kLa]0 = 1+0.099
0
rS
solutions. G, air; S, glass and ut =0.165–2.44 m s−1; CS 0.091 rLs 3L 0.023
ut 0.046 −1
bronze spheres uG =0.03–0.15 m s−1; hc = = 1+0.17

    n
rS gm 4L uG
1.5 m; dc =100–300 mm;
dr =60–190 mm rLs 3L 0.043
ut 0.067 −1

gm 4L uG

  n
DTFB. L, water; G, air; S, glass dS =0.3–0.7 mm; [kLa]/[kLa]0
Fan et al. (1987)
beads, activated carbon rS =1500–2500 kg m−3; oSRet
particles ut =0.023–0.0935 m s−1; = 0.143 1−31.5

 
ReS
uG =0–17.069 m s−1; hc =1.22
oSRet 0.002

 
m; dc =760 mm; dr =500 mm × (1−oS)10.31
ReS
FB. L, water; G, air; S, glass dS =0.05–8 mm; rS =2510–2950 oS
kLa= 0.39 1− u 0.67 Nguyen-Tien et al.
spheres kg m−3; oS =0.005–0.53; 0.58
g
(1985)
uL =0.046–0.116 m s−1; hc =m;
dc =140 mm

'  
AL. L, water; G, air; S, plastic dS =2.6–12.7 mm; 0.4
Fr% Miyahara and
oG =
beads rS =1046–1135 kg m−3; uL Kawate (1993)
uG =0.03–0.15 m s−1; uL =2–9 1+0.4 Fr% 1+
uG
m s−1; hc =1 m; dc =148 mm;
dr =60–100 mm
FB. L, water; G, air; S, glass dS =0.53–0.755 mm; rS =2338 kLa= 0.99u 0.5 0.97 −0.27
G uL dS (1−oS)0.53 Zheng et al. (1995)
beads, activated carbon kg m−3; uG =0.01–0.1 m s−1; (1−0.89
h+0.2h)
particles uL =0.05–0.5 m s−1; hc =4.1 m;
dc =285 mm; oS =0.02–0.165

a
Subscript 0 refers to the two-phase gas–liquid system.
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 23

where o is the energy dissipation rate.


One approach to analysing the mass transfer
data is to assume that the biofilm reactor may be
treated as having essentially uniform energy dissi-
pation rates throughout the entire volume. The
energy dissipation rate so calculated is:
o= uGg (10)
Kolmorgoff’s theory has been used extensively
in correlating mass transfer data for bubble
columns (Sano et al., 1974; Sanger and Deckwer,
1981), fluidized beds (Arters and Fan, 1986; Liv-
ingston and Chase, 1990) and airlift reactors
(Nicolella et al., 1998b). An exponent of 1/3 in
Eq. (8) is often chosen for the Schmidt number.
Published correlations valid for different types of
three-phase reactors are compared in Fig. 8. The
characteristics of the systems used by different
Fig. 8. Selected correlations for liquid–solid mass transfer
coefficient in three-phase sparged reactors. authors are reported in Table 6. For the case of
mass transfer to biofilm particles in a BAS reac-
et al., 1985) and that predicted by the model of tor, Nicolella et al. (1998b) correlated their exper-

 
Livingston and Zhang (1993). imental data with the following relationship:
The effect of the ratio of downcomer to riser od 4S 0.241

cross-section area (Ad/Ar) on hydrodynamic and Sh= 2.0+ 0.265 Sc1/3 (11)
n3
mass transfer in airlift reactors was studied by
Bello et al. (1985). They measured gas hold-up The L–S MTC measured for particle-supported
and G–L MTC in two-phase internal and exter- biofilms was found to be smaller (by a factor of
nal loop reactors with Ad/Ar varying in the range approximately 15%) than the values reported for
0.11 –0.69, and proposed unifying generalised cor- rigid particles, and it was therefore concluded that
relations for both parameters. L–S mass transfer should be regarded as critical
process in biofilm systems.
4.4.2. Liquid solid mass transfer 4.4.3. Intraparticle mass transfer and reaction
Mass transfer to particles submerged in a fluid Particle-supported biofilm reactors are charac-
is generally described by equations based on the terised by microorganisms being attached to a
boundary layer theory, which leads to a correlat- solid surface in the form of a biofilm. Biofilms are
ing equation of the form: dense layers through which substrates have to be
Sh = 2.0+ CRenScm (8) transported for biochemical reaction to occur.
This transport takes place by molecular diffusion,
Correlations of this form involve calculating the which is a slow process. In practice it proves that
particle Reynolds number. In problems involving the general rule is that the removal is limited by
liquid–particle mass transfer in three-phase sus- diffusion, and one of the major disadvantages of
pension the Reynolds number is frequently biofilm reactors is the low efficiency of thick
defined according to Kolmorgoff’s theory of biofilms (Harremoes and Henze, 1995).
turbulence: The substrate concentration in a biofilm is de-
termined by the substrate conversion process and
od 4S diffusion. The latter can be described by Fick’s
Re = (9)
n3 law with an effective diffusion coefficient. In gen-
24
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33
Table 6
Selected liquid–solid mass transfer correlations in aerated suspensions

Reference System Particles uG (mm s−1) Sc Re= od 4/n 3 Sh

Sano et al. BC Ion-exchange resin, benzoic acid, KMnO4, b-naphthole; 0–170 217–1410 1–107 2+0.4Re1/4Sc1/3
(1974) oS =0.03–0.07; dS =0.06–0.833 mm
Arters and Fan FB Benzoic acid; rS =1298 kg m−3; dS =1.8–4.4 mm 0–260 1960–3500 2.5×105 2+0.695Re0.2Sc1/3
(1986) –4.8×109
Goto et al. BCDT Ion-exchange resin; oS =0–0.015; dS =0.55–0.92 mm 1–10 143 103–106 2+1.01Re0.173Sc1/3
(1989)
Livingston and BCDT Ion-exchange resin; oS =0.04–0.1; dS =0.655–1.119 mm 0–60 400 1.5×104 2+0.275Re0.274Sc1/3
Chase (1990) –6.77×105
Kikuchi et al. BC, FB Ion-exchange resin; oS =0–0.4; dS =0.32–2.59 mm 0–100 300–600 10–104 2+0.47Re0.21Sc1/3
(1995)
Nicolella et al. BAS Biofilms; oS =0.05–0.15; dS =0.47–1.95 mm 2–27 400 103–2×106 2+0.275Re0.274Sc1/3
(1998a)
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 25

eral the effective diffusion coefficient is 80 – 90% sorting effect where particles with higher settling
of the diffusion coefficient in water. The substrate velocity are found at the bottom of the bed (Di
flux at the biofilm surface (if the concentration is Felice, 1995).
well above the substrate affinity coefficient, which The most common technique to study the mix-
is often the case for biofilm processes) can be ing characteristics of fluidized bed and airlift sus-
calculated by considering two distinct reaction pension systems is the injection and response
regions with abrupt transition in the order of the method (Levenspiel, 1972).
reaction (Harremoes, 1978). This transition is
characterised by: 4.5.1. Fluidized bed systems

b=
' 2DeC i
f
(12)
In BFB reactors, solid stratification occurs ow-
ing to the difference in size and density among the
k0d 2 biofilm particles. Thicker biofilms are usually
For b \ 1, the biofilm is fully penetrated by the found at the top of the bed, whilst bare carriers
substrate and the substrate flux is zero order with and thin biofilms remain in the bottom region of
respect to the substrate concentration at the bio- the column (Shieh et al., 1981; Hermanovicz and
film surface: Ganczarczyk, 1983; Boaventura and Rodrigues,
1988). Particle stratification has been attributed to
b \ 1 [ NS =k0d (13) differences in drag and buoyancy that affect parti-
For b B1, the biofilm is partially penetrated by cle terminal settling velocity (Di Felice, 1995).
the substrate and the substrate flux is half order Simple models have been developed to simulate
with respect to the substrate concentration at the solid stratification in BFBs (Di Felice et al., 1997).
biofilm surface: As a consequence of stratification and size distri-
bution, biodegradation rate, biofilm composition
b B1 [ NS =
2Dek0C if (14) and biofilm specific activity changes along the
The zero-order reaction rate constant is defined height of the bed. For example, in anaerobic
as (Harremoes, 1978) BFBs, the glucotrophic activity decreases along
the bed from the bottom to the top, whilst the
k0 = mmaxXf/Y (15) methanogenic activity increases and is maximum
at the top of the bed (thickest biofilms) (Buffiere
4.5. Mixing and scale-up effects et al., 1998a). In the upper portion of the bed,
where the thickest biofilms are present, diffusional
Liquid mixing in particulate biofilm reactors mass transfer limitations are particularly impor-
affects the interphase mass transfer, the reactant tant and reduce the conversion capacity of the
concentration distribution, and ultimately reac- reactor (Schreyer and Coughlin, 1999).
tant conversions. Information on axial liquid mix- Axial liquid mixing in liquid–solid fluidized
ing is therefore crucial to reactor design and beds and three-phase fluidized beds have been
optimisation. Fluidized bed systems and airlift extensively studied for systems with particles hav-
suspension systems present different mixing char- ing densities significantly greater than that of the
acteristics. Airlift suspension reactors are usually liquid medium (e.g. Davidson et al., 1985;
considered ideally mixed for liquid and solids Nguyen-Tien et al., 1984). Only a few studies
(Heijnen et al., 1993). Axial liquid mixing in liq- (Kikuchi et al., 1984; Tang and Fan, 1990; Zanin
uid – solid and gas– liquid – solid fluidized beds is et al., 1993) focused on axial liquid mixing in
often analysed using the axial dispersion model fluidized beds containing particles akin to particu-
(Levenspiel, 1972; Davidson et al., 1985; Fan, late biofilms. In particular, Tang and Fan (1990)
1989). In these systems, a distribution of solid size measured axial dispersion coefficients in liquid
and/or density results in a distribution of terminal fluidized beds of low density particles (1.05–1.3
settling velocities leading to a classification within kg m − 3). The axial dispersion coefficient was
the fluidized bed. This classification is caused by a found to be independent of bed height and to
26 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

increase monotonically with increasing liquid su- 1994). The degree of mixing in the riser is re-
perficial velocity. The extent of axial liquid mixing ported to be higher than that in the downcomer,
in beds of light particles is significantly less than and mixing in the two-phase systems is higher
that in beds of heavy particles, and therefore the than that in three-phase systems (Lu et al.,
particle density effect on axial liquid mixing is 1994a).
important. Gas production affects the hydrody- The axial dispersion model is generally used to
namics and mixing characteristics of BFBs describe the overall mixing phenomena in airlift
(Buffiere et al., 1998b), resulting in a bed contrac- reactors. The overall axial dispersion coefficient
tion and in an increase of dispersion. These effects increases as aeration rate increase (Lu et al.,
have in turn an influence on reactor performance, 1994b). This result is expected since a higher
which, at high reactor loading, can be reduced by aeration rate corresponds to a higher circulation
10 – 15% (Buffiere et al., 1998c). velocity, and, consequently, to a higher degree of
The axial dispersion coefficient in liquid turbulence.
fluidized bed reactors decreases with increasing Axial dispersion in airlift reactors is higher than
ratio of height to diameter ratio (Thommes et al., in fluidized beds. For lab-scale reactors of com-
1995). This results in different mixing characteris- parable size, Schoutens et al. (1986) reports axial
tics in lab-scale reactors and full-scale applica- dispersion coefficients of 4–16× 10 − 4 and 3–
tions. As the H/D ratio is larger in lab-scale 10× 10 − 3 m2 s − 1 for fluidized beds and airlift
reactors (H/D \10) than in pilot-scale and full- columns, respectively. The degree of mixing is
scale applications (H/D B5, see Table 2), the therefore higher in airlift reactors than in fluidized
liquid axial mixing is normally greater in indus- beds.
trial columns than it is in small bench reactors. Heijnen et al. (1993) report that a pilot-scale
This effect considerably affects the scale-up of (0.25 m3) and a full-scale (280 m3) BAS reactor
BFBs. have nearly the same gas hold-up. This implies
Though USB and EGSB present some hydro- that in airlift reactors there is a negligible effect of
dynamic similarities with fluidized systems, the scale on mass transfer and mixing characteristics.
simple axial dispersion model is insufficient to The effect of reactor scale on biomass structure,
describe their mixing characteristics. Based on biofilm detachment, liquid mixing, gas–liquid
experience with a model with all dimensions 1/20 mass transfer and liquid–solid mass transfer in
of a full-scale USB plant, Bolle et al. (1986) two- and three-phase particulate biofilm reactors
observed that in USB reactors the sludge bed and has been analysed extensively by Heijnen (1996).
the sludge blanket behave like mixed tank reac- In particular:
tors with short-circuiting flows, whilst the settling 1. The importance of undesired biomass struc-
volume can be described as a plug flow reactor. tures (e.g. growth in suspension, growth on
The short-circuiting flows over the sludge bed and static reactor surface growth) is clearly scale-
the sludge blanket (i.e. the mixing characteristics dependent. In particular the concentration of
of these compartments) were affected by the su- suspended organisms and debris in the reactor
perficial gas velocity. An optimal bed height of depends on their settling velocity relative to
3.5 – 4 m was found corresponding to minimal the wash-out liquid velocity in the three-phase
short-circuiting flows. separator. It is important to maintain identical
settler liquid velocities at different reactor
4.5.2. Airlift suspension systems scales to eliminate undesired biomass
The airlift reactor can be considered as a system structures.
consisting of interrelated zones: riser, downcomer, 2. As discussed above, most biofilm detachment
bottom section, top section and settler. Mixing in particulate biofilm reactors is due to parti-
efficiency is due to the circulation of liquid and cle–particle interactions in collisions. In two-
solid particles around the airlift column, and to phase systems, abrasion only occurs between
the backmixing in the settler (Obradovic et al., biofilm particles, because bare carriers are con-
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 27

centrated at the bottom of the reactor due to Appendix A. Design criteria for particle-biofilm
stratification. The rate of abrasion in two- reactors
phase systems increases at increasing particle
concentrations, i.e. at decreasing liquid veloc- A.1. Condition for biofilm de6elopment
ity. Liquid velocity should therefore be main-
tained constant in scaling-up a two-phase Biofilms will form in a bioreactor when the
reactor. For three-phase systems it is noted dilution rate (DH = 1/tH = Q/V) is larger than the
that, contrary to two-phase systems, biofilm maximum growth rate (mmax) (Heijnen, 1984; Ti-
particles are mixed up with bare carriers. If a jhuis, 1994):
heterogeneous flow regime is maintained,
scale-up effects are not significant. 1 Q
= \mmax (A1)
3. In three-phase BAS reactors, there is an im- tH V
portant effect of scale on the oxygen transport,
In some cases biofilm formation is already ob-
which is related to the circulation of gas bub-
served at a dilution rate below the maximum
bles in the downcomer. As discussed above,
growth rate. Upon analysis of the system it will be
gas regime 1 (no bubble circulation) and 2
generally observed that despite the biofilm accu-
(presence of bubbles in the downcomer but no
mulation the majority of the conversion still takes
bubble circulation) only occur in small reac-
place in suspension.
tors. Full-scale reactors always operate with
For an aerobic system where the rate limiting
full bubble recirculation (gas regime 3), which
factor is the gas–liquid oxygen mass transfer, the
is obviously the most favourable for oxygen
reactor size can be designed as a function of
transfer due to the presence of gas flow
organic loading rate FC = Q/C, oxygen yield (YO)
everywhere.
and oxygen transfer rate (OTR):
FC Y QC
V= = O (A2)
OTR OTR
5. Conclusions
By substituting Eq. (A2) and rearranging terms
A number of particulate-biofilm reactor tech- in Eq. (A1), the condition for biofilm develop-
niques (USB, BFB, EGSB, IC, BAS) have been ment (Eq. (A1)) can be rewritten as:
developed in the last two decades for applications OTR
to industrial and municipal wastewater treatment. CB (A3)
YOmmax
Several full-scale applications have been presented
and illustrated in this review, and the engineering The line in Fig. 1 delimiting the region of
aspects of their design and operation have been applications of particle-biofilms and flocs (line 1)
analysed and discussed. Most fundamentals of is defined assuming a maximum oxygen transfer
biofilm formation, hydrodynamics, mixing, mass rate of 10 kg m − 3 day − 1, an oxygen yield coeffi-
transfer and chemical kinetics of these systems are cient of 1 and a maximum specific growth rate of
well understood, and empirical correlations and 1 day − 1.
mathematical models are available for the predic-
tions of relevant parameters and variables needed A.2. Condition for particle retention
in the design and operation of full-scale reactors.
One of the least understood aspects in particu- For particle retention in a bioreactor, the su-
late-biofilm reactors is detachment of biomass. perficial liquid velocity u=Q/A must be lower
Although the mechanism seems to be particle – than the particle settling velocity (ut):
particle collisions, there is no design rule for rates
of detachment. This is an important lack of Q
u= B ut (A4)
knowledge, which requires further research. A
28 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

Using Eq. (A4), and assuming a cylindrical is obtained which limits the region of applicability
geometry for the bioreactor, the reactor volume of static biofilm reactors.
can be expressed as a function of the liquid veloc-
ity and the height to diameter ratio a = H/D:
'
A.4. Limitations on biomass separation
Q 4Q
V=a (A5) The separation step (usually by sedimentation)
u pu
is the main limitation in processes based on the
By combining Eq. (A5) with Eq. (A2), the use of flocs (e.g. activated sludge processes). Floc
condition for particle retention in the biofilm re- settling characteristics are rather poor (ut B 5 m
actor (Eq. (A4)) can be rewritten as: h − 1) and limits the maximum volatile suspended
pu 3t C 2Y 2O solid concentration in the liquors to about 3 kg
QB (A6) m − 3 (Metcalf and Eddy, 1991). This in turn
4a 2OTR2
results in a limitation of the reactor capacity,
The lines in the diagram C – Q of Fig. 1 delimit- which can be expressed as a function of the
ing the regions of applications of particle-biofilm reactor biomass concentration by considering the
suspensions and floc suspension (line 2 and 3, food to microorganism ratio (defined as the ratio
respectively) are obtained by assuming settling between the organic load to the reactor and the
velocities of 30 and 5 m h − 1 for particle biofilms reactor biomass concentration: F/M=QC/VX):
and flocs, respectively, and an aspect ratio a =5
(the maximum oxygen transfer rate is fixed at 10 OTR= (F/M)XYO (A9)
kg m − 3 day − 1 as in the previous section).
The food to microorganism ratio is an empiri-
A.3. Liquid– solid mass transfer limitations cal parameter frequently used for the design of
activated sludge processes. Typical values for con-
In biofilm reactors, the process rate may be ventional systems range from 0.3 to 1.5 kgCOD
−1
limited by liquid–solid mass transfer. For aerobic kgSS day − 1 (Metcalf and Eddy, 1991). Assuming
−1
processes, assuming for the sake of simplicity that a value of 1 kgCOD kgSS day − 1, and taking YO as
oxygen concentration at the biofilm interface is 1, the maximum oxygen transfer rate in an acti-
negligible, the oxygen flux (N) from the bulk vated sludge process can be estimated as 3 kg
liquid to the biofilm surface is given by: m − 3 day − 1, which, using Eq. (A3) results in a
vertical line in the diagram C–Q (line 5 in Fig. 1).
N =kLSCL (A7)
When the solid loading to the settler increases,
The volumetric oxygen transfer rate (OTR) is the settling area required to achieve a desirable
the product of flux (N) and biofilm specific sur- effluent quality in a reasonable time may become
face area (a). unrealistic. There is no general rule, since in prin-
ciple the settling volume could be chosen as large
OTR = Na (A8)
as required, but a sensible design would probably
For systems with low values of biofilm specific not consider settling volumes much larger than
surface area (e.g. static biofilm reactors), the liq- the oxidation volumes. If, for example, it is as-
uid – solid mass transfer may become the rate lim- sumed that the settler to reactor volume ratio is
iting process. Assuming values of 1 m day − 1 for limited to a given value (let g be this ratio), then
kLS and 8 g m − 3 for CL (dissolved oxygen concen- it can be shown (with calculations similar to those
tration in water at equilibrium using air at ambi- performed in the previous section) that the region
ent temperature and pressure), the maximum of applicability of activated sludge processes is
oxygen transfer rate in a static biofilm reactor delimited in the plane C–Q by the line:
with a specific surface area of 200 m2 m − 3 is 2 kg
m − 3 day − 1. Introducing this value in Eq. (3), a pq 3C 2Y 2O
Q= (A10)
vertical line in the diagram C – Q (line 4 in Fig. 1) 4(a/g)2OTR2
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 29

where q is the overflow rate to the settler. Values Buffiere, P., Fonade, C., Moletta, R., 1998b. Liquid mixing
for q and a in Eq. (A10) can be obtained from the and phase hold-ups in gas producing fluidized bed bioreac-
tors. Chem. Eng. Sci. 53, 617 – 627.
practice of wastewater engineering (Metcalf and Buffiere, P., Fonade, C., Moletta, R., 1998c. Mixing and phase
Eddy, 1991). The line 6 of Fig. 1 is plotted hold-ups variations due to gas production in anaerobic
assuming q = 2 m h − 1, a =0.5, g =5. fluidized bed digesters: influence on reactor performance.
Biotechnol. Bioeng. 60, 36 – 43.
Chang, H.-T., Rittmann, B.E., Amar, D., Heim, R., Ehlinger,
O., Lesty, Y., 1991. Biofilm detachment mechanisms in a
References liquid fluidized bed. Biotechnol. Bioeng. 38, 499 – 506.
Characklis, W.G., 1990. Biofilm processes. In: Characklis,
W.G., Marshall, K.C. (Eds.), Biofilms. Wiley, New York.
Andrews, G.F., Tien, C., 1979. The expansion of a fluidized
Chaudari, R.V., Ramachandran, P.A., 1980. Three phase
bed containing biomass. AIChE J. 25, 720–723.
slurry reactors. AIChE J. 26, 177 – 201.
Arters, D.C., Fan, L.-S., 1986. Solid–liquid mass transfer in a
Chisti, M.Y., 1989. Airlift Reactors. Elsevier, Amsterdam.
gas – liquid – solid fluidized bed. Chem. Eng. Sci. 41, 107–
Chisti, M.Y., Moo-Young, M., 1987. Hydrodynamics and
115.
oxygen transfer in pneumatic bioreactor devices. Biotech-
Austermann, U., Meyer, H., Seyfried, C.F., Rosenwinkel,
nol. Bioeng. 31, 487 – 494.
K.H., 1999. Full scale experiences with anaerobic/aerobic
Chisti, M.Y., Moo-Young, M., 1988. Gas hold-up in pneu-
treatment plants in the food and beverage industry. Water
Sci. Technol. 40, 305 –312. matic reactors. Chem. Eng. J. 38, 149 – 152.
Ayazi Shamlou, P., Pollard, D.J., Ison, A.P., Lilly, M.D., Chisti, M.Y., Halard, B., Moo-Young, M., 1988. Liquid circu-
1994. Gas hold-up and liquid circulation rate in concentric lation in airlift reactors. Chem. Eng. Sci. 43, 451 – 457.
tube airlift bioreactors. Chem. Eng. Sci. 49, 303–312. Cooper, P.F., 1981. The use of biological fluidised beds for the
Bakker, W.A.M., van Can, H.J.L., Tramper, J., de Gooijer, treatment of domestic and industrial wastewaters. Chem.
C.D., 1993. Hydrodynamics and mixing in a multiple Eng. 371, 373 – 376.
air-lift loop reactor. Biotechnol. Bioeng. 42, 994–1001. Cooper, P.F., Sutton, P.M., 1983. Treatment of wastewaters
Badot, R., Coulom, T., de Longeaux, N., Badard, M., Sibony, using biological fluidized beds. Chem. Eng. 392, 392.
J., 1994. A fluidized bed reactor: the biolift process. Water Cooper, P.F., Wheeldon, D.H.V., 1982. Complete treatment
Sci. Technol. 29, 329 –338. of sewage in a two-stage fluidized bed system. Part 1.
Beenackers, A.A.C.M., van Swaaij, W.P.M., 1993. Mass trans- Water Pollut. Cont. 81, 447 – 464.
fer in gas – liquid slurry reactors. Chem. Eng. Sci. 48, Csizkor, Z., Mihaltz, P., Czako, L., Hollo, J., 1994. New
3109 – 3139. interpretation of expansion in biofilm-coated particle
Bello, R.A., Robinson, C.W., Moo-Young, M., 1985. Gas fluidization. Appl. Microbiol. Biotechnol. 41, 608 – 614.
holdup and overall volumetric oxygen transfer coefficient Davidson, J.F., Clift, R., Harrison, D., 1985. Fluidization.
in airlift contactors. Biotechnol. Bioeng. 27, 369–381. Pergamon, Oxford.
Bernard, J., Rovel, J.-M., 1989. Fluidized bed reactor with Deckwer, W.-D., Schumpe, A., 1993. Improved tools for
means for ensuring homogeneous distribution of the fluid bubble column reactor design and scale-up. Chem. Eng.
to be treated. U.S. Patent No. 4,869,815. Sci. 48, 889 – 911.
Boaventura, R.A., Rodrigues, A.E., 1988. Consecutive reac- Denac, M., Uzman, S., Tanaka, H., Dunn, I.J., 1983. Mod-
tions in fluidized bed biological reactors: modelling and elling and experiments on biofilm penetration effects in a
experimental study of wastewater denitrification. Chem. fluidized bed nitrification reactor. Biotechnol. Bioeng. 25,
Eng. Sci. 43, 2715 – 2728. 1841 – 1861.
Bolle, W.L., van Breugel, J., van Eybergen, G.C., Kossen, Di Felice, R., 1995. Hydrodynamics of liquid fluidisation.
N.W.F., Zoetemeyer, R.J., 1986. Modeling the liquid flow Chem. Eng. Sci. 50, 1213 – 1245.
in up-flow anaerobic sludge blanket reactors. Biotechnol. Di Felice, R., Nicolella, C., Rovatti, M., 1997. Mixing and
Bioeng. 28, 1615 – 1620. segregation in water fluidized bed bioreactors. Water Res.
Boonaert, C.J.-P., Dupont-Gillain, C.C., Dengis, P.B., 31, 2392 – 2396.
Dufrene, Y.F., Rouxhet, P.G., 1999. Cell separation, floc- Douek, R.S., Livingston, A.G., Johansson, A.C., Hewitt,
culation. In: Flickinger, M.C., Drew, J.W. (Eds.), Encyclo- G.F., 1994. Hydrodynamic of an external-loop three-phase
pedia of Bioprocess Technology. Wiley, New York. airlift reactor. Chem. Eng. Sci. 49, 3719 – 3737.
Bryers, J.D., Characklis, W.G., 1990. Biofilms in water and Dunn, I.J., Tanaka, H., Uzman, S., Denac, M., 1983. Biofilm
wastewater treatment. In: Characklis, W.G., Marshall, fluidized bed reactors and their application to wastewater
K.C. (Eds.), Biofilms. Wiley, New York. nitrification. Ann. N. Y. Acad. Sci. 413, 168 – 183.
Buffiere, P., Steyer, J.P., Fonade, C., Moletta, R., 1998a. Durot, J., Prevot, C., 1987. Fluid bed reactor for the biological
Modeling and experiments on the influence of biofilm size treatment of water. U.S. Patent No. 4,707,252.
and mass transfer in a fluidized bed reactor for anaerobic Fan, L.-S., 1989. Gas – Liquid – Solid Fluidization. Butter-
digestion. Water Res. 32, 657–668. worth, Guildford, UK.
30 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

Fan, L.-S., Fujie, K., Long, T.R., Tang, W.-T., 1986. Charac- Harada, H., Ando, H., Momonoi, K., 1987. Process analysis
teristics of draft tube gas–liquid–solid fluidized bed biore- of fluidized bed biofilm reactor for denitrification. Water
actor with immobilized living cells for phenol degradation. Sci. Technol. 19, 151 – 162.
Biotechnol. Bioeng. 30, 498–504. Harremoes, P., 1978. Biofilm kinetics. In: Mitchell, R. (Ed.),
Fan, L.-S., Ramesh, T.S., Tang, W.-T., Long, T.R., 1987. Water Pollution Microbiology, vol. 2. Wiley, New York,
Gas – liquid mass transfer in a two-stage draft tube gas–liq- pp. 71 – 109.
uid – solid fluidized bed. Chem. Eng. Sci. 42, 543–553. Harremoes, P., Henze, M., 1995. Biofilters. In: Henze, M.,
Frankin, R., Koevoets, W.A.A., van Gils, W.M.A., van der Harremoes, P., Jansen, J.C., Arvin, E. (Eds.), Wastewater
Pas, A., 1992. Application of the Biobed upflow fluidized- Treatment. Springer, Berlin, pp. 143 – 192.
bed process for anaerobic waste water treatment. Water Heck, J., Onken, U., 1988. Characteristics of solid suspension
Sci. Technol. 25, 373–382. in a bubble column with and without draft tube. Chem.
Frijters, C.T.M.J., Eikelboom, D.H., Mulder, A., Mulder, R., Eng. Sci. 44, 1743 – 1745.
1997. Treatment of municipal wastewater in a Circox airlift Heijnen, J.J., 1984. Biological industrial wastewater treatment
reactor with integrated denitrification. Water Sci. Technol. minimizing biomass production and maximizing biomass
36, 173 – 181. concentration. Ph.D. Thesis, Delft University of Technol-
Fuchs, R., Ryu, D.D.Y., Umphrey, A.E., 1971. Effect of ogy, Delft.
surface aeration on scale-up procedures for fermentation Heijnen, J.J., 1985. Process for preparing biomass attached to
processes. Ind. Eng. Chem. Proc. Des. Dev. 10, 190–196. a carrier. US Patent No. 4,560,479.
Furusaki, S., 1988. Engineering aspects of immobilized biocat- Heijnen, J.J., 1996. Scale up aspects of immobilized cell reac-
alysts. J. Chem. Eng. Jpn. 21, 219–230. tors. Prog. Biotechnol. 11, 497 – 510.
Garcia-Calvo, E., Rodriguez, A., Prados, A., Klein, J., 1999. Heijnen, J.J., van’t Riet, K., 1984. Mass transfer and heat
A fluid dynamic model for three-phase airlift reactors. transfer phenomena in low viscosity bubble column reac-
Chem. Eng. Sci. 54, 2359–2370. tors. Chem. Eng. J. 28, B21 – B42.
Gavroy, D., Joly-Vuillemin, C., Cordier, G., Delmas, H.,
Heijnen, J.J., Mulder, A., Enger, W., Hoeks, F., 1989. Review
1995. Gas hold-up, liquid circulation and gas–liquid mass
on the application of anaerobic fluidized bed reactors in
transfer in slurry bubble-columns. Chem. Eng. Res. Des.
wastewater treatment. Chem. Eng. J. 41, B37 – B50.
Part A 73, 637 – 642.
Heijnen, J.J., Mulder, A., Weltevrede, R., Hols, P.H., van
Gjaltema, A., Tijhuis, L., van Loosdrecht, M.C.M., Heijnen,
Leeuwen, H.L.J.M., 1990. Large-scale anaerobic/aerobic
J.J., 1995. Detachment of biomass from suspended non-
treatment of complex industrial wastewater using immobi-
growing spherical biofilms in airlift reactors. Biotechnol.
lized biomass in fluidized bed and airlift suspension reac-
Bioeng. 46, 258 – 269.
tors. Chem. Eng. Technol. 13, 202 – 208.
Gjaltema, A., van der Marel, N., van Loosdrecht, M.C.M.,
Heijnen, J.J., Mulder, A., Weltevrede, R., Hols, J., van
Heijnen, J.J., 1997a. Adhesion and biofilm development on
leeuwen, H.L.J.M., 1991. Large scale anaerobic – aerobic
suspended carriers in airlift reactors: Hydrodynamic condi-
treatment of complex industrial waste water using biofilm
tions versus surface characteristics. Biotechnol. Bioeng. 55,
880 – 889. reactors. Water Sci. Technol. 23, 1427 – 1436.
Gjaltema, A., Vinke, J.L., van Loosdrecht, M.C.M., Heijnen, Heijnen, J.J., van Loosdrecht, M.C.M., Mulder, R.,
J.J., 1997b. Abrasion of suspended biofilm pellets in airlift Weltevrede, R., Mulder, A., 1993. Development and scale
reactors: Importance of shape, structure, and particle con- up of an aerobic biofilm airlift suspension reactor. Water
centrations. Biotechnol. Bioeng. 53, 88–99. Sci. Technol. 27, 253 – 261.
Gjaltema, A., van Loosdrecht, M.C.M., Heijnen, J.J., 1997c. Heijnen, J.J., Hols, J., van der Lans, R.G.J.M., van Leeuwen,
Abrasion of suspended biofilm pellets in airlift reactors: H.L.J.M., Mulder, A., Weltevrede, R., 1997. A simple
Effect of particle size. Biotechnol. Bioeng. 55, 206–215. hydrodynamic model for the liquid circulation velocity in a
Godia, F., Sola’, C., 1995. Fluidized bed bioreactors. Biotech- full scale two and three phase internal airlift reactor oper-
nol. Prog. 11, 479 – 497. ating in the gas recirculation regime. Chem. Eng. Sci. 52,
Gorur, S., Vereijken, T., Tielbaard, M., 1995. Strict effluent 2527 – 2540.
treatment design criteria at brewery with novel anaerobic- Hermanovicz, S.W., Cheng, Y.W., 1990. Biological fluidized
aerobic process. WEFTEC Exhibition and Conference, bed reactor: hydrodynamics, biomass distribution and per-
Miami Beach, FL. formance. Water Sci. Technol. 22, 193 – 202.
Goto, S., Matsumoto, Y., Gaspillo, P., 1989. Mass transfer Hermanovicz, S.W., Ganczarczyk, J.J., 1983. Some fluidiza-
and reaction in bubble column slurry reactor with draft tion characteristics of biological beds. Biotechnol. Bioeng.
tube. Chem. Eng. Commun. 85, 181–191. 25, 1321 – 1330.
Habets, L.H.A., Engelaar, A.J.H.H., Groenveld, N., 1997a. Hickey, R.F., Owens, R.W., 1981. Methane generation from
Anaerobic treatment of inuline effluent in an internal high-strength industrial wastes with the anaerobic biologi-
circulation reactor. Water Sci. Technol. 35, 189–197. cal fluidized bed. Biotechnol. Bioeng. Symp. 11, 399 – 413.
Habets, L.H.A., Engelaar, A.J.H.H., Groenveld, N., 1997b. Holst, T.C., Truc, A., Pujol, R., 1997. Anaerobic fluidized
Anaerobic treatment of sugarbeet and inuline effluent in an beds: ten years of industrial experience. Water Sci. Tech-
internal circulation reactor. EuroTechLink 97, UK. nol. 36, 415 – 422.
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 31

Hoyland, G., Robinson, P.J., 1983. Aerobic treatment in Livingston, A.G., Chase, S.F., 1990. Liquid – solid mass trans-
Oxitron biological fluidized-bed plant at Coleshill. Water fer in a three-phase draft tube fluidized bed reactor. Chem.
Pollut. Cont. 82, 479–493. Eng. Commun. 92, 225 – 244.
Hwang, S.J., Cheng, Y.-L., 1997. Gas holdup and liquid Livingston, A.G., Zhang, S.F., 1993. Hydrodynamic be-
velocity in three-phase internal loop airlift reactors. Chem. haviour of three-phase (gas – liquid – solid) airlift reactors.
Eng. Sci. 52, 3943 – 3960. Chem. Eng. Sci. 48, 1641 – 1654.
Hwang, S.J., Lu, W.-J., 1997. Gas–liquid mass transfer in an Lourens, P., Zoetemeyer, R.J., 1992. Fluidized bed process.
internal loop airlift reactor with low density particles. U.S. Patent No. 5,116,505.
Chem. Eng. Sci. 52, 853–857. Lu, W.-J., Hwang, S.-J., Chang, C.-M., 1994a. Liquid mixing
Jeris, J.S., 1983. Industrial wastewater treatment using anaero- in internal loop airlift reactors. Ind. Eng. Chem. Res. 33,
bic fluidized bed reactors. Water Sci. Technol. 15, 169– 2180 – 2186.
176. Lu, W.-J., Hwang, S.-J., Chang, C.-M., 1994b. Liquid mixing
Jeris, J.S., Beer, C., Mueller, J.A., 1976. Waste treatment in two and three-phase airlift reactors. Chem. Eng. Sci. 49,
apparatus. U.S. Patent No. 3,956,129. 1465 – 1468.
Jeris, J.S., Owens, R.W., Hickey, R., Flood, F., 1977. Biologi- Lu, W.-J., Hwang, S.-J., Chang, C.-M., 1995. Liquid velocity
cal fluidized bed treatment for BOD and nitrogen removal. and gas holdup in three-phase internal loop airlift reactors
J. Water Pollut. Cont. Fed. 49, 816–831. with low-density particles. Chem. Eng. Sci. 50, 1301 – 1310.
Jewell, W.J., Switzenbaum, M.S., Morris, J.W., 1981. Munici- Melcer, H., Nutt, S., Marvan, I., Sutton, P., 1984. Combined
pal wastewater treatment with the anaerobic attached mi- treatment of coke plant wastewater and blast furnace
crobial film expanded bed process. J. Water Pollut. Cont. blowdown water in a coupled biological fluidized bed
Fed. 53, 482 – 491. system. J. Water Pollut. Cont. Fed. 56, 192 – 198.
Kikuchi, K.-I., Konno, H., Kakutani, S., Sugawara, T., Metcalf and Eddy, 1991. Wastewater Engineering. Treatment,
Ohashi, H., 1984. Axial dispersion of liquid in liquid Disposal, Reuse. McGraw-Hill, New York.
Mishra, P.N., Sutton, P.M., 1991. Biological fluidized beds for
fluidized beds in the low Reynolds number region. J.
water and wastewater treatment: a state of the art review.
Chem. Eng. Jpn. 17, 362–367.
In: Rossmoore, H.W. (Ed.), Biodeterioration and
Kikuchi, K.-I., Takahashi, H., Suggawara, T., 1995. Liquid–
Biodegradation. Elsevier, New York.
solid mass transfer in a slurry bubble column and a
Miyahara, T., Kawate, O., 1993. Hydrodynamics of a solid-
gas – liquid – solid three-phase fluidized bed reactor. Chem.
suspended bubble column with a draught tube containing
Eng. Commun. 73, 313–321.
low-density particles. Chem. Eng. Sci. 48, 127 – 133.
Koide, K., Horibe, K., Kawabata, H., Ito, S., 1985. Gas
Mousseau, F., Liu, S.X., Hermanovicz, S.W., Lazarova, V.,
holdup and volumetric liquid-phase mass transfer coeffi-
Manem, J., 1998. Modeling of Turboflo — A novel bio-
cient in solid-suspended bubble column with draught tube.
film reactor for wastewater treatment. Water Sci. Technol.
J. Chem. Eng. Jpn. 18, 248–254.
37, 177 – 181.
Kwok, W.K., Picioreanu, C., Ong, S.L., van Loosdrecht,
Mulcahy, L.T., Shieh, W.K., 1987. Fluidization and reactor
M.C.M., Ng, W.J., Heijnen, J.J., 1998. Influence of biomass characteristics of the denitrification fluidized bed
biomass production and detachment forces on biofilm biofilm reactor. Water Res. 21, 451 – 458.
structures in a biofilm airlift suspension reactor. Biotech- Muroyama, K., Fan, L.-S., 1985. Fundamentals of gas – liq-
nol. Bioeng. 58, 400 –407. uid – solid fluidization. AIChE J. 31, 1 – 34.
Lazarova, V., Manem, J., 1996. An innovative process for Muroyama, K., Mitani, Y., Yasunishi, A., 1985. Hydrody-
wastewater treatment: the circulating bed floating reactor. namics characteristics and gas liquid mass transfer in a
Water Sci. Technol. 34, 89–99. draft tube slurry reactor. Chem. Eng. Commun. 34, 87 – 98.
Lazarova, V., Meyniel, J., Duval, L., Manem, J., 1997. A Ngian, K.-F., Martin, W.R.B., 1980. Bed expansion character-
novel circulating bed reactor: hydrodynamics, mass trans- istics of liquid fluidized particles with attached microbial
fer and nitrification capacity. Chem. Eng. Sci. 52, 3919– growth. Biotechnol. Bioeng. 22, 1843 – 1856.
3927. Nguyen-Tien, K., Patwari, A.N., Schumpe, A., Deckwer,
Lettinga, G., van Velsen, A.F.M., Homba, S.W., de Zeeuw, W.D., 1984. Liquid dispersion in 3 phase fluidized beds. J.
W., Klapwijk, A., 1980. Use of the upflow sludge blanket Chem. Eng. Jpn. 17, 652 – 653.
reactor concept for biological wastewater treatment espe- Nguyen-Tien, K., Patwari, A.N., Schumpe, A., Deckwer,
cially for anaerobic treatment. Biotechnol. Bioeng. 22, W.D., 1985. Gas-liquid mass transfer in fluidized bed
699 – 734. particle beds. AIChE J. 31, 194 – 201.
Levenspiel, O., 1972. Chemical Reaction Engineering. Wiley, Nicolella, C., Converti, A., Di Felice, R., Rovatti, M., 1995.
New York. The estimation of the solid size and density in liquid
Linek, V., Vacek, V., Benes, P., 1987. A critical review and fluidised-bed biological reactors. Chem. Eng. Sci. 50,
experimental verification of the correct use of the dynamic 1059 – 1062.
method for the determination of oxygen transfer in aerated Nicolella, C., Di Felice, R., Rovatti, M., 1996. An experimen-
agitated vessels to water, electrolyte solutions and viscous tal model of biofilm detachment in liquid fluidized bed
liquids. Chem. Eng. J. 34, 11–34. biological reactors. Biotechnol. Bioeng. 51, 713 – 719.
32 C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33

Nicolella, C., Di Felice, R., Rovatti, M., 1997. Biomass con- Schugerl, K., 1989. Biofluidization: Application of the fluidiza-
centration in fluidised bed biological reactors. Water Res. tion technique in biotechnology. Can. J. Chem. Eng. 67,
31, 936 – 940. 178 – 184.
Nicolella, C., van Loosdrecht, M.C.M., van der Lans, Schugerl, K., 1997. Three-phase-biofluidization: Application
R.G.J.M., Heijnen, J.J., 1998a. Hydrodynamic characteris- of the fluidization technique in the biotechnology. A re-
tics and gas – liquid mass transfer in a biofilm airlift suspen- view. Chem. Eng. Sci. 52, 3661 – 3668.
sion reactor. Biotechnol. Bioeng. 60, 627–635. Shah, Y.T., Kelkar, B.G., Godbole, S.P., Deckwer, W.D.,
Nicolella, C., van Loosdrecht, M.C.M., Heijnen, J.J., 1998b. 1982. Design parameters estimations for bubble column
Mass transfer and reaction in a biofilm airlift suspension reactors. AIChE J. 28, 353 – 379.
reactor. Chem. Eng. Sci. 53, 2743–2753. Shieh, W.K., Sutton, P.M., Kos, P., 1981. Predicting reactor
Nicolella, C., van Loosdrecht, M.C.M., Heijnen, J.J., 1999a. biomass concentration in a fluidized bed system. J. Water
Identification mass transfer parameters in three-phase bio- Pollut. Cont. Fed. 53, 1574 – 1584.
film reactors. Chem. Eng. Sci. 54, 3143–3152. Siegel, M.H., Robinson, C.W., 1992. Applications of airlift
Nicolella, C., van Loosdrecht, M.C.M., Di Felice, R., Rovatti, gas – liquid – solid reactors in biotechnology. Chem. Eng.
M., 1999b. Terminal settling velocity and bed expansion Sci. 47, 3215 – 3229.
characteristics of biofilm-coated particles. Biotechnol. Bio- Sutton, P.M., Mishra, P.N., 1990. Fluidized bed biological
eng. 62, 63 – 70. wastewater treatment: effects of scale-up on system perfor-
Nutt, S.G., Melcer, H., Pries, J.H., 1984. Two-stage biological mance. In: P.M. Sutton and Associates, Biological Flu-
fluidized bed treatment of coke plant wastewater for nitro- idized Beds for Water and Wastewater Treatment: a User’s
gen control. J. Water Pollut. Cont. Fed. 56, 851–857. Forum. Conference Proceedings, Ann Arbor, pp. 1 – 12.
Obradovic, B., Dudukovic, A., Vunjaknovakovic, G., 1994. Sutton, P.M., Mishra, P.N., 1994. Activated carbon based
Local and overall mixing characteristics of the gas–liquid– biological fluidized beds for contaminated water and
solid airlift reactor. Ind. Eng. Chem. Res. 33, 698–702. wastewater treatment: a state of the art review. Water Sci.
Oliva, E., Jacquart, J.C., Prevot, C., 1990. Treatment of
Technol. 29, 309 – 317.
wastewater at the El Aguila brewery. Methanization in
Sutton, P.M., Shieh, W.K., Kos, P., Dunning, P.R., 1980.
fluidized bed reactors. Water Sci. Technol. 22, 483–490.
Dorr-Oliver’s Oxitron system fluidized bed water and
Pandit, A.B., Joshi, J.B., 1986. Mass and heat transfer charac-
wastewater treatment process. In: Cooper, P.F., Atkinson,
teristics of three-phase sparged reactors. Chem. Eng. Res.
F. (Eds.), Biological Fluidized Bed Treatment of Water
Des. 64, 125 – 157.
and Wastewater. Ellis Horwood, Chichester.
Pereboom, J.H.F., Vereijken, T.L.F.M., 1994. Methanogenic
Sutton, P.M., Hurvid, J., Hoeksema, M., 1999. Biological
granule development in full-scale internal circulation reac-
fluidized bed treatment of wastewater from byproduct
tors. Water Sci. Technol. 30, 9–21.
coking operations: full-scale history. Water Environ. Res.
Perry, R.H., Green, D.W., 1997. Chemical Engineers’ Hand-
71, 5 – 9.
book. McGraw-Hill, New York.
Tang, W.-T., Fan, L.-S., 1987. Steady state phenol degrada-
Richardson, J.F., Zaki, W.N., 1954. Sedimentation and fluidi-
sation. Part I. Trans. Inst. Chem. Eng. 32, 35–53. tion in a draft tube gas – liquid – solid fluidized bed bioreac-
Rittmann, B.E., Trinet, F., Amar, D., Chang, H.T., 1992. tor. AIChE J. 33, 239 – 249.
Measurement of the activity of a biofilm: the effect of Tang, W.-T., Fan, L.-S., 1990. Axial liquid mixing in liquid –
surface loading and detachment on a three-phase, liquid- solid and gas – liquid – solid fluidized beds containing low
fluidized-bed reactor. Water Sci. Technol. 26, 585–594. density particles. Chem. Eng. Sci. 45, 543 – 551.
Ro, K.S., Neethling, J.B., 1990. Terminal settling velocity of Thomas, C.R., Yates, J.G., 1985. Expansion index for biolog-
bioparticles. Res. J. Water Pollut. Cont. Fed. 62, 901–906. ical fluidised beds. Chem. Eng. Res. 63, 67 – 70.
Ryhner, G., Petrozzi, S., Dunn, I.J., 1988. Operation of a Thommes, J., Weiher, M., Karau, A., Kula, M.R., 1995.
three-phase biofilm fluidized sand bed reactor for aerobic Hydrodynamics and performance in fluidized bed adsorp-
wastewater treatment. Biotechnol. Bioeng. 32, 677–688. tion. Biotechnol. Bioeng. 48, 367 – 374.
Sanger, P., Deckwer, W.D., 1981. Liquid–solid mass transfer Tijhuis, L., 1994. The biofilm airlift suspension reactor. Ph.D.
in aerated suspensions. Chem. Eng. J. 22, 179–186. Thesis, Delft University of Technology, Delft.
Sano, Y., Yamaguchi, N., Adachi, T., 1974. Mass transfer Tijhuis, L., van Loosdrecht, M.C.M., Heijnen, J.J., 1994.
coefficients for suspended particles in agitated vessels and Formation and growth of heterotrophic aerobic biofilms
bubble columns. J. Chem. Eng. Jpn. 7, 155–161. on small suspended particles in airlift reactors. Biotechnol.
Schoutens, G.H., Guit, R.P., Zieleman, G.J., Luyben, Bioeng. 44, 595 – 608.
K.C.A.M., Kossen, N.W.F., 1986. A comparative study of Tijhuis, L., Hijman, B., van Loosdrecht, M.C.M., Heijnen,
a fluidized bed reactor and a gas lift loop reactor for the J.J., 1995a. Influence of detachment, substrate loading and
IBE process. Part II: hydrodynamics and reactor mod- reactor scale on formation of biofilms in airlift reactors.
elling. J. Chem. Technol. Biotechnol. 36, 415–426. Appl. Microbiol. Biotechnol. 45, 7 – 17.
Schreyer, H.B., Coughlin, R.W., 1999. Effects of stratification Tijhuis, L., van Loosdrecht, M.C.M., Heijnen, J.J., 1995b.
in a fluidized bed bioreactor during treatment of metal Dynamics of biofilm detachment in biofilm airlift suspen-
working wastewater. Biotechnol. Bioeng. 63, 129–140. sion reactors. Biotechnol. Bioeng. 45, 481 – 487.
C. Nicolella et al. / Journal of Biotechnology 80 (2000) 1–33 33

Trinet, F., Heim, R., Amar, D., Chang, H.T., Rittmann, B.E., Vellinga, S.H.J., 1986. Anaerobic purification equipment for
1991. Study of biofilm and fluidization of bioparticles in a waste water. US Patent No. 4,609,460.
three phase liquid fluidized bed reactor. Water Sci. Tech- Wolmarans, B., Driessen, W., 1996. Full scale BIOPAQ
nol. 23, 1347 – 1354. UASB installation treating distillery effluent. Proceeding of
van Benthum, W.A.J., 1998. Integrated nitrification and deni- WISA Conference, Port Elisabeth, South Africa.
trification in biofilm airlift reactors: biofilm development, Wolmarans, B., Nell, B., 1996. Biopaq UASB treatment of
process design and hydrodynamics. Ph.D. Thesis, Delft distillery effluent — From pilot to full scale at SFW
University of Technology, Delft. Wellington, RSA. AFRIWATER 96, Johannesburg, South
van Benthum, W.A.J., van den Hoogen, J.H.A., van der Lans, Africa.
R.G.J.M., van Loosdrecht, M.C.M., Heijnen, J.J., 1999a. Wright, P.C., Raper, J.A., 1996. A review of some parameters
The biofilm airlift suspension reactor. Part I: design and involved in fluidized bed bioreactors. Chem. Eng. Technol.
two-phase hydrodynamics. Chem. Eng. Sci. 54, 1909–1924. 19, 50 – 64.
van Benthum, W.A.J., van der Lans, R.G.J.M., van Loos-
Yu, H., Rittmann, B.E., 1997. Predicting bed expansion and
drecht, M.C.M., Heijnen, J.J., 1999b. Bubble recirculation
phase holdups for three-phase fluidized bed reactors with
regimes in an internal loop airlift reactor. Chem. Eng. Sci.
and without biofilm. Water Res. 31, 2604 – 2616.
54, 3995 – 4006.
Zanin, G.M., Neitzel, I., de Morales, F.F., 1993. Axial
van Benthum, W.A.J., van der Lans, R.G.J.M., van Loos-
dispersion in a liquid fluidized bed of particles akin to
drecht, M.C.M., Heijnen, J.J., 2000. The biofilm airlift
immobilized enzymes. Appl. Biochem. Biotechnol 39/40,
suspension reactor. Part II: three-phase hydrodynamics.
Chem. Eng. Sci. 55, 699–711. 477 – 489.
van Houten, R.T., Pol, L.W.H., Lettinga, G., 1994. Biological Zheng, C., Chen, Z., Feng, Y., Hofmann, H., 1995. Mass
sulphate reduction using gas-lift reactors fed with hydrogen transfer in different flow regimes of three-phase fluidized
and carbon dioxide as energy and carbon source. Biotech- beds. Chem. Eng. Sci. 50, 1571 – 1578.
nol. Bioeng. 44, 586 –594. Zoutberg, G.R., de Been, P., 1997. The Biobed EGSB (ex-
van Houten, R.T., Yun, S.Y., Lettinga, G., 1997. Ther- panded granular sludge bed) system covers shortcomings
mophilic sulphate and sulphite reduction in lab-scale gas- of the upflow anaerobic sludge blanket reactor in the
lift reactors using H2 and CO2 as energy and carbon chemical industry. Water Sci. Technol. 35, 183 – 188.
source. Biotechnol. Bioeng. 55, 807–814. Zoutberg, G.R., Frankin, R., 1996. Anaerobic treatment of
van Loosdrecht, M.C.M., Eikelboom, D., Gjaltema, A., Mul- chemical and brewery waste water with a new type of
der, A., Tijhuis, L., Heijnen, J.J., 1995. Biofilm structures. anaerobic reactor: the Biobed EGSB reactor. Water Sci.
Water Sci. Technol. 31, 163–171. Technol. 34, 375 – 381.

You might also like