Validation of a Model for Process

Development and Scale-Up of Packed-Bed

Solid-State Bioreactors
Frans J. Weber,
1,2
Jaap Oostra,
2
Johannes Tramper,
2
Arjen Rinzema
1,2
1
Wageningen Centre for Food Sciences, P.O. Box 557, 6700 AN Wageningen,
The Netherlands
2
Food and Bioprocess Engineering Group, Wageningen University, P.O. Box
8129, 6700 EV Wageningen, The Netherlands; telephone: +31 (0)317 482683;
fax: +31 (0)317 482237; e-mail: Frans.Weber@algemeen.pk.wau.nl
Received 26 December 2000; accepted 13 July 2001
Abstract: We have validated our previously described
model for scale-up of packed-bed solid-state fermenters
(Weber et al., 1999) with experiments in an adiabatic 15-
dm
3
packed-bed reactor, using the fungi Coniothyrium
minitans and Aspergillus oryzae. Effects of temperature
on respiration, growth, and sporulation of the biocontrol
fungus C. minitans on hemp impregnated with a liquid
medium were determined in independent experiments,
and the rst two effects were translated into a kinetic
model, which was incorporated in the material and en-
ergy balances of the packed-bed model. Predicted tem-
peratures corresponded well with experimental results.
As predicted, large amounts of water were lost due to
evaporative cooling. With hemp as support no shrinkage
was observed, and temperatures could be adequately
controlled, both with C. minitans and A. oryzae. In ex-
periments with grains, strong shrinkage of the grains
was expected and observed. Nevertheless, cultivation of
C. minitans on oats succeeded because this fungus did
not form a tight hyphal network between the grains.
However, cultivation of A. oryzae failed because shrink-
age combined with the strong hyphal network formed by
this fungus resulted in channeling, local overheating of
the bed, and very inhomogeneous growth of the fungus.
For cultivation of C. minitans on oats and for cultivation
of A. oryzae on wheat and hemp, no kinetic models were
available. Nevertheless, the enthalpy and water balances
gave accurate temperature predictions when online
measurements of oxygen consumption were used as
input. The current model can be improved by incorpo-
ration of (1) gas-solids water and heat transfer kinetics to
account for deviations from equilibrium observed with
fast-growing fungi such as A. oryzae, and (2) the dy-
namic response of the fungus to changes in tempera-
ture, which were neglected in the isothermal kinetic
experiments. 2002 John Wiley & Sons, Inc. Biotechnol
Bioeng 77: 381393, 2002; DOI 10.1002/bit.10087
Keywords: Aspergillus oryzae; Coniothyrium minitans;
model; packed bed; solid-state fermentation
INTRODUCTION
Many examples of new and promising products from
fungal solid-state fermentation (SSF) are reported in the
literature (Pandey et al., 2000). However, it is unclear
whether these processes can be eectively scaled-up to
industrial scale. One of the major problems to overcome
in large-scale SSF is heat accumulation. Due to the
absence of free-owing water and the low thermal con-
ductivity of solid substrates, removal of the heat pro-
duced by growing microorganisms can be problematic in
SSF. High temperatures must be avoided, as they ad-
versely aect microbial activity. In laboratory-scale
packed-bed reactors, adequate temperature control can
be achieved by wall cooling. However, in larger reactors,
conductive cooling becomes insucient and cooling by
forced aeration has to be used. Cooling by forced ae-
ration is mainly eective due to evaporation (Grajek,
1988; Gutie rrez-Rojas et al., 1996; Oostra et al., 2000;
Sato et al., 1982). This implies that successful tempera-
ture control can only be achieved at the expense of a loss
of moisture from the substrate, which may negatively
aect the cultivation. Furthermore, the shift from con-
ductive cooling to evaporative cooling can result in axial
gradients in packed-bed reactors (Gowthaman et al.,
1993). The axial gradients in industrial-scale packed-
bed reactors can cause dierences in productivity be-
tween lab-scale and industrial packed-bed reactors. A
mathematical model predicting temperature, moisture,
biomass, and/or substrate proles in a large-scale
packed-bed reactor would therefore be a valuable tool
to evaluate scale-up.
Previously, several mathematical models for solid-
state fermenters have been published, so one might
question the need for further modeling studies. We will
briey discuss the shortcomings of these previous
models, to show that there is still an urgent need for
models that correctly describe the predominant physical
phenomena and that have been adequately veried by
experimental work (Mitchell et al., 2000). Only models
that describe spatial and temporal gradients in packed-
bed reactors with forced aeration will be considered. In
most of these models, the eect of evaporation on heat
transfer was not taken into account (Gutie rrez-Rojas Correspondence to: F. J. Weber
2002 John Wiley & Sons, Inc.
et al., 1995; Sangsurasak and Mitchell, 1995; Saucedo-
Castan eda et al., 1990). As discussed above, this is a very
dangerous assumption in a model that will be used to
evaluate scale-up. In addition, these models are ham-
pered by errors in the energy balance (i.e., the formu-
lation of the convection term implies that the solids
move along with the air) (Sangsurasak and Mitchell,
1995; Saucedo-Castan eda et al., 1990) or have been
validated in a bioreactor where conductive radial cool-
ing predominates and axial gradients as well as evapo-
ration losses are much less important then they would be
on an industrial scale (Gutie rrez-Rojas et al., 1995;
Saucedo-Castan eda et al., 1990).
A heat transfer model in which evaporation was in-
cluded was only presented in 1998 (Sangsurasak and
Mitchell, 1998). This model very nicely demonstrated
the importance of evaporation for cooling of unmixed
packed-bed reactors with forced aeration. A desired
improvement of this model was the incorporation of a
water balance, as the authors estimated that about 85%
of the initially available water had evaporated after 40 h
of growth (Sangsurasak and Mitchell, 1998).
Previously, we proposed a strategy to evaluate
whether an SSF-process can be successfully scaled up to
an industrial scale (Weber et al., 1999). This strategy was
based on a model similar to that proposed by Sangsu-
rasak and Mitchell (1998), but extended with a water
balance to predict the water content of the solid sub-
strate in the packed bed. Based on literature and labo-
ratory data in combination with heat and mass balances,
the aeration and evaporation rates required to remove
the metabolic heat produced by microorganisms grow-
ing in an industrial-scale packed-bed reactor were esti-
mated. Whether large-scale production is feasible
depends on (1) the pressure drop resulting from the re-
quired aeration rate, and (2) the eect of evaporation on
the particle volume and water activity. This approach
was used to decide which of the support materi-
alsoats, hemp, bagasse, or perlitecould be used in
large-scale packed-bed reactors for spore production of
Coniothyrium minitans, an eective biological agent for
pest control in many food crops (Whipps and Gerlagh,
1992). It was shown that moisture control is the limiting
factor for cultivation of C. minitans in a packed-bed
reactor. The model predicted that oats could not be used
due to shrinkage and a
w
reduction caused by evapora-
tive cooling. Of the three inert supports tested, hemp
was expected to provide the best spore yield and control
of water activity (Weber et al., 1999).
Theoretical work thus suggests that mathematical
models will be useful tools in the scale-up process.
However, there is an urgent need to test the accuracy
and robustness of the models by applying them within
real process development (Mitchell et al., 2000). In the
current study, our extended model (Weber et al., 1999) is
validated by measuring the temperature and moisture
proles occurring when C. minitans is cultivated in a
properly downscaled packed-bed reactor (i.e., a well-
insulated column with neglegible conductive cooling). In
addition, the applicability of our model to evaluate
other SSF processes employing fungi with a higher
growth rate than C. minitans is reported. Aspergillus
oryzae, an important food fermentation organism, was
used for these studies.
MATERIALS AND METHODS
Inoculum Preparation
C. minitans isolate IVT1 (CBS 148.96) was kindly pro-
vided by M. Gerlagh of IPO-DLO, Wageningen, The
Netherlands. A. oryzae (CBS 570.65) was obtained from
the Centraal Bureau voor Schimmelcultures, The
Netherlands. Both strains were routinely cultured on
potato-dextrose agar (PDA) (Oxoid, Basingstoke, UK)
at 20C and 35C, respectively. A stock solution of
spores was obtained by ooding a PDA agar dish with a
sterile saline solution and gentle scraping of the plate
with a bent glass rod. Glycerol (20% wt/vol, nal con-
centration) was added to the spore suspension, which
was subsequently stored at )80C.
Effect of Temperature
The eect of temperature on growth and sporulation of
C. minitans was determined on impregnated hemp.
Glass culture tubes 1 (diameter 2.5 cm, height 15 cm) with
screw caps and silicon septa were used. The tubes had
three notches at about 1 cm from the bottom, on which
a stainless steel wire-mesh was placed. About 0.2 g of
dry hemp was placed on the wiring in each tube and 3
mL of demineralized water were added. These tubes and
separate solutions of 500-g glucose/L and 100-g yeast
extract/L were sterilized in an autoclave (20 min,
121C). After sterilization, 1-mL aliquots of the glucose
and yeast-extract solutions were added to the tubes with
hemp. The tubes were gently mixed and left at room
temperature for at least 5 h: in this period, the nutrients
could diuse into the hemp. To each tube, 0.4 mL of a
diluted conidial stock solution was added. After 30 min,
the excess of liquid medium was removed by means of
sterile Pasteur pipettes. After 4 h, the remainder of the
nonabsorbed uid that had dripped from the hemp was
removed. Initially, the tubes were incubated at 20C to
allow germination of the conidia. After 2 d, the tubes
were placed in a temperature gradient block (con-
structed by the mechanical workshop of Wageningen
University, The Netherlands). This aluminum block (I
d h: 67 cm 59 cm 18 cm) has temperature condi-
tioning on the left- and right-hand side, by means of
heat exchangers and temperature-controlled water-cir-
culation baths (10 and 32C), and good isolation on the
other sides. A linear temperature gradient exists between
the temperature-controlled sides of the block. The cul-
382 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 4, FEBRUARY 15, 2002
ture tubes were placed in the holes present in the block.
The temperature of the tubes was recorded with a tem-
perature sensor. Growth was assessed by measuring O
2
consumption, using a gas chromatograph as described
previously (Weber et al., 1999).
Packed-Bed Reactor
To mimic an industrial packed-bed reactor, where
temperature losses at the wall are relatively small, we
have constructed a well-isolated pilot-scale reactor. The
reactor is made of polypropylene, and has an internal
diameter of 20 cm and a height of 70 cm (bed height 50
cm). The reactor can be sterilized in an autoclave. At
various heights in the reactor the temperature was
measured with Pt100s. A 2 perforated plate (3-mm-di-
ameter holes) at the bottom of the reactor served to
support the solid substrate. The airow through the
reactor (bottom to top) was controlled with a mass-ow
controller (050 L
N
/min, Brooks, The Netherlands). The
temperature and humidity of the gas entering the reactor
were controlled at 18C and 100% RH (relative hu-
midity). The air entering the reactor was rst humidied
by blowing air through a stainless steel column (diam-
eter 30 cm, height 80 cm) lled with Raschig rings and
water of 23C. Subsequently, the air was cooled down to
18C and the condensed water was collected. A sterile,
0.2-lm PTFE-membrane lter (PolyVent 1000, What-
man, UK) was used to sterilize the air. The humidity of
the o-gas was continuously measured with a cooled-
mirror dewpoint analyzer (Dewmet SD, Michell, UK).
The concentrations of O
2
and CO
2
in the o-gas of the
reactor were measured with in-line gas analyzers (Xentra
4100 paramagnetic oxygen analyzer and Series 1400
infrared CO
2
analyzer, Servomex, The Netherlands). A
small portion of the o-gas (100 mL/min) was passed
through a glass condenser at 5C to remove most of the
water vapor prior to the gas analysis. The reactor was
monitored and controlled with a PC using Fieldpoint
hardware and Labview software (National Instruments,
The Netherlands).
The thick plastic wall of the reactor prevented the
solids inside the reactor from reaching suciently high
temperatures during sterilization in an autoclave. To
secure sterility, the solid substrates were placed in a
thin autoclave bag connected with its opening to the
top of the reactor. After sterilization in the autoclave
(2.5 h, 121C), the contents of the bag were poured in
the reactor, and the bag remained connected to the
reactor. For the cultivations on grain, about 5 kg of
oats or wheat were placed in the bag and 5 kg of
demineralized water was added. A separate ask con-
taining 20 L of demineralized water was also sterilized.
After sterilization, conidia were aseptically added to
the water, which was subsequently pumped in the re-
actor. When the reactor was completely lled, the ex-
cess of water was removed. For the cultivations on
impregnated hemp, about 1.1 kg of hemp (Hemparade,
HempFlax b.v., The Netherlands) were placed in the
autoclave bag and 10 L of demineralized water was
added. To prevent the occurrence of undesired Mail-
lard reactions, the nutrients were sterilized in separate
asks, one containing glucose (Merck, Germany), the
other yeast extract (Technical grade, Difco, USA). For
the standard medium, 2-kg glucose and 400-g yeast
extract in, respectively, 7- and 3-L water were used. In
one experiment, a threefold higher nutrient concentra-
tion was used: 6-kg glucose and 1.2-kg yeast extract.
After heat sterilization, the hemp was poured in the
reactor. The yeast extract and glucose solutions were
mixed and conidia were added. This solution was
pumped in the reactor; when the reactor was com-
pletely lled with medium it was left to stand for at
least 5 h. A period of 5 h is required for the substrates
to diuse into the hemp and reach equilibrium. After
impregnation, the excess medium was removed from
the reactor. Subsequently, the reactor was weighed,
and a sample was aseptically taken to determine the
initial water content. The sample was taken with the
aid of the autoclave bag, functioning as an aseptic
sampler. The sample was collected from the top of the
bed and positioned in a corner of the bag. Two tie-raps
were placed around the bag to separate the sample
from the reactor. The corner of the bag, containing the
sample, was then removed from the reactor by cutting
the bag between the two tie-raps. The packed-bed re-
actor was placed in a cylindrical container (diameter 40
cm, height 100 cm 3 ), which was placed in a tempera-
ture-controlled incubator (18C). The space between
the wall of the cylinder and the reactor was lled with
cork granulate (size: 12 mm) to prevent radial heat
losses. After cultivation, the weight of the reactor was
again determined. Samples were taken from various
heights, to determine water content, water activity, and
spore yield. An additional sample was taken after
thorough mixing of the reactor contents.
ANALYSIS
The moisture content was determined from the weight
loss after drying the sample at 80C for 2 d. The water
activity of the sample was determined in a Novasina
Thermoconstanter (Type TH200, Switzerland). Spores
were liberated from the substrate by blending and were
counted with an electronic particle counter (Casy 1,
Scha rfe-System, Germany) as described previously
(Weber et al., 1999). The water content and spore yield
were expressed per gram of initial dry weight (IDW) of
substrate. For the samples taken at the end of the cul-
tivation, the measured dry weights could be converted
into IDW with the aid of the measured weight of the
total reactor content.
WEBER ET AL.: LARGE-SCALE PACKED-BED SOLID-STATE FERMENTATION 383
MATHEMATICAL MODEL
For large-scale packed-bed reactors, it is valid to assume
that heat losses through the reactor wall are negligible
(Gutie rrez-Rojas et al., 1996; Oostra et al., 2000), and
radial gradients in the bed are therefore unimportant.
When it is furthermore assumed that the air follows ideal
plug-owbehavior, the enthalpy andwater balances read:
o
ot
(1 e)

i
(C
i
h
i
) e

j
(c
j
h
j
)
2 3
= r
///
o
DH
o
F
//
a

o
oz
h
a
(1)
o
ot
(1 e) (x
ws
C
s
x
wx
C
x
) e C
wg

= r
///
w
F
//
a

o
oz
y
w
(2)
C
i
h
i
and C
j
h
j
indicate enthalpies of all components
in the moist solid matrix and the gas phase, respectively.
The term for convective enthalpy transport is based on
the mass ow rate of dry air (F
///
a
kg/m
2
per s), and takes
the enthalpy of dry air 4 and water vapor into account:
h
a
(T) = c
pa
(T T
ref
) y
w
(T) c
pwv
(T T
ref
) DH
w

(3)
The water vapor fraction in saturated air can be calcu-
lated from:
y
w
=
M
w
(water) p
w
M
w
(air) (P
tot
p
w
)
= 0:622
p
w
P
tot
p
w
(4)
and the water vapor pressure can be calculated from the
following empirical expression, which was obtained by
tting Antoine's law to water vapor pressure data at
temperatures between 273 K and 333 K (Kaye and
Laby, 1995):
p
w
= exp 23:59
4045
T 37:70

(5)
The enthalpy and mass balances can be simplied
considerably (Weber et al., 1999). First, in the accu-
mulation term of the balances, the contributions of
gases and all mass accumulation terms are negligible.
Second, a pseudo-steady state with respect to tempera-
ture and oxygen consumption rate can be assumed, be-
cause the characteristic times for changes in these
variables are much larger than those for convective axial
enthalpy transport. Third, we assume that the air is at
equilibrium with the solid matrix at any point in the bed.
These simplications reduce the enthalpy balance to:
0 = r
///
o
DH
o
F
//
a

d
dz
(h
a
) (6)
For the simplication of the water mass balance, it is
also assumed that the water production rate is propor-
tional to the oxygen consumption rate:
r
///
w
= r
///
o
Y
wo
(7)
This simplies the mass balance to:
o
ot
(1 e) (x
ws
C
s
x
wx
C
x
) [ [ = r
///
o
Y
wo
F
//
a

o
oz
y
w
(8)
Note that this model discriminates between water in
biomass and water in the substrate. However, it is hard
to measure these variables independently. The change in
the total amount of water can be calculated from:
o
ot
x
w
= r
///
o
Y
wo
F
//
a

o
oz
y
w
(9)
Table I shows the physical constants used in our
model. An important parameter, the yield coecient
Y
wo
, is unknown and dicult to measure accurately. We
have estimated Y
wo
from the following stoichiometric
equations for, respectively, glucose and starch:
C
6
H
12
O
6
2:59 O
2
2:85 CH
1:8
O
0:5
3:44 H
2
O 3:15 CO
2
C
6
H
10
O
5
2:59 O
2
2:85 CH
1:8
O
0:5
2:44 H
2
O 3:15 CO
2
These equations were set up using an average biomass
composition (CH
1.8
O
0.5
) (Roels, 1983) and the reported
yield of 0.75 kg biomass per kg of oxygen for C. minitans
(Ooijkaas et al., 2000a). The obtained Y
wo
for growth on
glucose (0.75 kg H
2
O per kg O
2
) is higher than for
growth on 5 starch (0.53 kg H
2
O per kg O
2
) as the hy-
drolysis of starch requires water.
Oxygen Consumption
The growth of microorganisms on a solid substrate can
be described by the logistic law (Okazaki et al., 1980):
r
///
X
= l
max
X 1
X
X
max

(10)
The oxygen consumption rate is described with the
linear-growth model (Pirt, 1965):
r
///
o
q
s
=
1
Y
xo
r
///
x
m
o
X

(11)
Table I. Physical properties used in the 9 models.
Parameter Value Unit Reference
c
pa
1,005 J/kg per K Hamblin (1971)
c
pwv
1,857 J/kg per K Hamblin (1971)
DH
o
1.4 10
7
J/kg O
2
Cooney et al. (1968)
DH
w
2.5 10
6
(at T
ref
) J/kg H
2
O Perry et al. (1984)
P
tot
1.01 10
5
Pa
T
ref
273 K
Y
wo
0.75 (glucose),
0.53 (starch)
kg H
2
O/kg O
2
Assumed
q
s
64 kg dry hemp/m
3
reactor
This study
384 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 4, FEBRUARY 15, 2002
It is assumed that Y
xo
and X
max
are temperature inde-
pendent. The parameters l
max
and m
o
are assumed to be
temperature dependent, and the Ratkowsky equation is
used to describe the temperature dependency (Ratkow-
sky et al., 1983):
l
max
(T) = a
1
T T
min
[ [ 1 e
a
2
(TT
max
)
h i
2
& '
(12)
m
o
(T) = b
1
T T
min
[ [ 1 e
b
2
(TT
max
)
h i
2
& '
(13)
The unknown parameters were obtained by tting
Eqs. (10)(13) on O
2
consumption rates measured at
various temperatures using a three-dimensional regres-
sion (SigmaPlot 4.01, SPSS, USA).
RESULTS AND DISCUSSION
To estimate the heat production rate in a solid-state
bioreactor, the growth rate of the microorganisms
should be known. However, reliable methods to quantify
biomass formation on solid substrates are not available
(Ooijkaas et al., 1998). Oxygen consumption, however, is
easily determined and directly correlated with heat pro-
duction (Cooney et al., 1968). We have therefore mea-
sured the O
2
consumption to estimate growth.
Effect of Temperature on Growth
For C. minitans growing on hemp impregnated with a
nutrients solution, the O
2
consumption was determined
at various temperatures between 12 and 30C. In pre-
liminary experiments, it was observed that temperature
had a remarkable eect on the germination rate of the
conidia. In a solid-state reactor, initially no temperature
gradient will exist and all conidia will germinate at the
same temperature. Only when biomass starts to grow
can an axial temperature gradient develop. Therefore,
we used an initial temperature of 20C for the rst 2 d to
allow germination. Subsequently, the cultures were in-
cubated at temperatures between 12 and 30C, and ox-
ygen consumption was measured (Fig. 1).
The logistic and Pirt equations [Eqs. (10) and (11)]
were used to t the O
2
consumption measurements. A
temperature-independent yield coecient of 0.75 kg
biomass per kg O
2
was used (Ooijkaas et al., 2000a). All
other model parameters were obtained by tting; it was
assumed that the maximum amount of biomass (X
max
) is
the same at all temperatures, and that l
max
and m
o
are
temperature dependent [Eqs. (12) and (13)]. A good
correlation (R
2
= 0.977) between measurements and
model description was obtained (Fig. 1B); which was
conrmed by the parity plot of measured values against
predicted values (Fig. 1C). The tted minimum tem-
perature for growth of C. minitans (261 K) is clearly not
a meaningful value (Table II). This is caused by the low
number of measurements at temperatures near T
min
.
Furthermore, the Ratkowsky equation is an arbitrary
function, and the validity of this function to accurately
describe growth and maintenance of C. minitans over
the entire temperature has not been shown yet. Despite
the nonrealistic value for T
min
, a good description
of the temperature-dependent O
2
consumption in the
Figure 1. Measured and tted O
2
consumption of C. minitans growing on hemp (A) at (

) 25.0C, (u) 20.2C, and (e) 15.3C, and (B) at all

temperatures measured. (C) Parity plot of the tted O
2
consumption against the measured consumption.
Table II. Obtained 10 t parameters describing the O
2
consumption of
C. minitans on hemp at temperatures between 12 and 30C [Eqs. (10)
(13)].
Parameter Value ( std. error)
X
0
(7.59 1.65) 10
)4
kg
X
/kg
IDW
X
max
0.194 0.004 kg
X
/kg
IDW
a1 0.0336 0.0022 1/(day
0.5
K)
a2 0.266 0.038 1/K
b1 (7.99 0.41) 10
)3
kg
O2
0:5
/(day
0.5
kg
X
0:5
K)
b2 1.35 3.40 1/K
T
min
260.6 1.7 K
T
max
307.3 0.5 K
WEBER ET AL.: LARGE-SCALE PACKED-BED SOLID-STATE FERMENTATION 385
temperature range of 12 to 30C, which is of interest
for our experiments, is obtained.
It should furthermore be noted that the obtained
descriptions for biomass formation and maintenance
requirements are not validated by independent mea-
surements. However, the obtained values for the main-
tenance coecient (up to 0.12 kg O
2
/kg biomass per day
at 30C) are of the same magnitude as reported for fungi
in submerged cultures (Roels, 1983).
Effect of Temperature on Spore Production
The temperature also aects the formation of C. mini-
tans conidia (Fig. 2). At temperatures above 25C, the
number of conidia is signicantly reduced. For an op-
timal production it is thus required that the temperature
in the bed is adequately controlled.
Predicting PBR Behavior
Due to the dierences in eectiveness of various cooling
mechanisms in small-scale reactors and large-scale re-
actors, scale-up criteria are required. Previously, we
proposed a strategy to evaluate whether (1) the tem-
perature could be adequately controlled, and (2) exces-
sive evaporation of water would not hamper the
cultivation in large-scale packed-bed reactors (Weber
et al., 1999). It was concluded that production of conidia
of the biocontrol fungus C. minitans in a packed-bed
reactor on an industrial scale is feasible. To mimic a
large-scale packed-bed reactor, we have insulated the
reactor to prevent heat losses via the reactor wall. For C.
minitans growing on hemp, we previously predicted that
an air ow rate of 0.025 kg air/m
3
per s would be suf-
cient to prevent axial temperature gradients of more
than 5C. For this prediction we used the measured O
2
consumption rate at 20C, as the highest radial growth
rate was observed at this temperature (McQuilken et al.,
1997). It appears now that the highest O
2
consumption
rate is not observed at 20C but at 25C (Fig. 1). As a
consequence, the activity of C. minitans was underesti-
mated in our previous evaluation. The dierence in the
reported optimal temperatures is probably caused by the
dierent methods used to determine the optimum (radial
growth versus O
2
consumption) and the low number of
temperatures tested by McQuilken. At 25C, the oxygen
consumption rate is maximal 7 10
)5
kg O
2
/m
3
reactor
per s, and an aeration rate of 0.055 kg dry air/m
3
per s
was estimated to result in axial temperature gradients of
5C. Figure 3 shows that a slightly higher maximum
axial temperature dierence (ca. 6C) was measured
when C. minitans was cultivated in the insulated packed-
bed reactor.
When we use the mathematical description of the O
2
consumption as a function of temperature [Eqs. (10)
(13) and Table II] to predict the O
2
consumption and
temperature in the PBR, a reasonable correlation be-
tween the measured and predicted temperatures is ob-
served (results not shown). However, at this aeration
rate, the dierence in oxygen concentration of the inlet
and outlet is very low (<0.12%). To obtain more reli-
able O
2
consumption measurements, we have done
several cultivations at a reduced aeration rate. In Fig. 4,
the predicted temperature and O
2
consumption rate in a
packed-bed reactor with C. minitans aerated at a re-
duced ow rate are shown and compared with the
measured values. A reasonable resemblance between
measured and predicted temperatures is observed when
the kinetic model based on independent measurements is
Figure 2. Number of conidia formed after isothermal cultivation of
C. minitans on hemp for 21 d.
Figure 3. Measured 13 temperatures in an aerated (F
///
a
= 0.055 kg/m
3
per s) packed-bed reactor with C. minitans growing on hemp (experi-
ment 12). Bottom to top: temperature of inlet air, temperatures at 5,
15, 25, 35, and 45 cm bed height, and temperature of outlet air.
386 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 4, FEBRUARY 15, 2002
used. The small dierence is caused by an inaccuracy in
the prediction of the fungal activity. When the measured
O
2
consumption of the reactor is used for the tempera-
ture predictions, a good match between the predicted
and measured temperatures is observed (Fig. 4). The
deviation between predicted and actual O
2
consumption
is not unexpected. The eect of temperature on the O
2
consumption rate has been measured in isothermal ex-
periments. In the reactor, however, the fungus does not
experience isothermal conditions. It is therefore not
surprising that the response to elevated temperatures
during cultivation is dierent from that which occurs
when the microorganism is subjected to the elevated
temperature from the beginning. Studies with Rhizopus
oligosporus showed that the isothermal approach is not
adequate to describe the eects of temperature changes
during SSF (Ikasari et al., 1999).
Control Strategies
Two control strategies can be applied to prevent too-
high temperatures in the PBR (Weber et al., 1999). The
rst option is to choose the air ow rate based on the
highest oxygen consumption rate during the fermenta-
tion, keep it constant, and let the outlet temperature
vary. The second possibility is to let the air ow rate
vary with the oxygen consumption rate and keep the
outlet temperature constant. In the experiment of Fig. 4,
the rst control option is used (constant ow rate).
Figure 5 shows the results of a cultivation where the
Figure 4. Measured and predicted temperature at the outlet of the reactor (left) and oxygen consumption rate (right) in an aerated (F
///
a
= 0.027
kg/m
3
per s; T
air in
= 17.7C) packed-bed reactor (experiment 19). Symbols: () measurements; (- - -) prediction using the kinetic model based on
independent isothermal O
2
consumption measurements; (

) predicted temperature using the measured O

2
consumption in the reactor.
Figure 5. Measured and predicted temperatures (left), air ow rate (middle), and oxygen consumption rate (right) in an aerated (variable airow)
packed-bed reactor (experiment 23). Symbols: () measurements; (- - -) prediction using independent isothermal O
2
-consumption measurements.
WEBER ET AL.: LARGE-SCALE PACKED-BED SOLID-STATE FERMENTATION 387
second control strategy was applied. In this cultivation,
the temperature in the top of the reactor was kept at
27.5C by means of a PID controller that continuously
monitored the temperature at the top and adjusted the
air ow rate. We did not use the measured O
2
con-
sumption rate to control the ow rate, as the noise on
this signal was high due to the small dierence in inlet
and outlet concentrations.
Again, small deviations between predicted and actual
O
2
consumption are observed. As explained before, these
deviations may be caused by the use of independent O
2
consumption measurements at isothermal temperatures.
As a consequence, the predicted and measured ow rates
also dier slightly. Despite this small error, the model
gives a reasonable prediction of the temperatures in the
packed-bed reactor (Fig. 5).
With both control options, the maximum axial tem-
perature gradient was about 10C. The use of control
strategy 2 will economically be the most interesting, as
less energy will be required for the aeration. However, the
axial temperature gradient in this system is constantly
high. With strategy 1, high temperatures are only ob-
served during a small period of time. As the spore yield of
C. minitans on hemp is reduced at temperatures above
25C (Fig. 2), the total number of conidia produced is
lower when control option 2 is used (Table III). When
control option 1 is used, the number of conidia produced
(Table III) is similar to the maximum obtainable yield at
the optimum isothermal temperature (Fig. 2). Appar-
ently, the temporarily high temperatures in the top of the
bed had no negative eect on formation of conidia.
Probably, most conidia are formed after day 6, when bed
temperatures are again optimal for conidiation (Fig. 3).
With control option 2 it is, of course, also possible to
prevent the occurrence of temperatures above 25C.
When these temperatures are prevented, this strategy is
expected to give spore numbers comparable to those
observed in strategy 1.
Effects of Evaporation
Forced aeration is an eective cooling mechanism in
large-scale packed-bed bioreactors. However, forced
aeration results in evaporation of water and desiccation
of the substrate. Desiccation of the substrate can either
lead to an unfavorably low water activity (a
w
), resulting
in poor microbial activity, or to shrinkage of the sub-
strate particles with subsequent channeling in the bed.
Water Activity
Coniothyrium minitans is very sensitive to reduced water
activities: a small reduction in the a
w
has a tremendous
eect on spore formation (Weber et al., 1999). For a
high productivity, it is thus essential that the a
w
remains
close to 1.0 during the cultivation. However, the support T
a
b
l
e
I
I
I
.
O
v
e
r
v
i
e
w
o
f
t
h
e
s
e
t
t
i
n
g
s
a
n
d
o
b
t
a
i
n
e
d
r
e
s
u
l
t
s
i
n
t
h
e
v
a
r
i
o
u
s
p
a
c
k
e
d
-
b
e
d
e
x
p
e
r
i
m
e
n
t
s
.
1
1
E
x
p
e
r
i
-
m
e
n
t
F
u
n
g
u
s
s
u
b
s
t
r
a
t
e
(
t
y
p
e
)
s
u
b
s
t
r
a
t
e
(
k
g
)
A
i
r

o
w
(
k
g
/
m
3
p
e
r
s
)
D
u
r
a
t
i
o
n
(
d
)
O
2
c
o
n
s
u
m
e
d
(
k
g
O
2
/
m
3
)
M
a
x
i
m
u
m
r
O
2
k
g
O
2
/
m
3
p
e
r
s
F
i
n
a
l
a
a w
C
o
n
i
d
i
a
1
/
m
3
D
e
c
r
e
a
s
e
d
w
a
t
e
r
c
o
n
t
e
n
t
(
k
g
)
M
e
a
s
u
r
e
d
P
r
e
d
i
c
t
e
d
b
1
1
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
c
1
.
0
1
0
.
0
2
7
1
4
.
1
2
5
.
2
8
.
2

1
0
)
5
1
.
0
0
6
.
2

1
0
1
4
1
.
7
0
1
.
1
6
1
.
1
2
1
8
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
1
.
0
8
0
.
0
2
7
1
1
.
0
2
2
.
9
8
.
3

1
0
)
5
1
.
0
0
6
.
0

1
0
1
4
1
.
9
7
1
.
2
6
1
.
2
6
1
9
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
1
.
0
1
0
.
0
2
7
1
4
.
0
2
4
.
9
6
.
9

1
0
)
5
1
.
0
0
7
.
9

1
0
1
4
1
.
9
2
1
.
1
6
1
.
1
7
2
5
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
1
.
1
8
0
.
0
2
7
1
4
.
0
2
6
.
4
8
.
3

1
0
)
5
1
.
0
0
5
.
3

1
0
1
4
2
.
2
1
1
.
3
7
1
.
3
6
1
2
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
1
.
0
1
0
.
0
5
5
1
4
.
2
2
4
.
1
7
.
1

1
0
)
5
1
.
0
0
5
.
4

1
0
1
4
1
.
7
3
1
.
1
1
1
.
1
4
1
6
C
.
m
i
n
i
t
a
n
s
H
e
m
p
B
c
1
.
0
1
0
.
0
2
7
1
7
.
0
5
7
.
3
1
4

1
0
)
5
0
.
9
7
1
.
7

1
0
1
4
2
.
9
6

2
.
9
3
2
3
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
1
.
1
8
V
a
r
i
a
b
l
e
1
1
.
6
2
8
.
6
9
.
6

1
0
-
5
1
.
0
0
4
.
7

1
0
1
4
2
.
1
0
1
.
4
1
1
.
1
8
2
4
C
.
m
i
n
i
t
a
n
s
H
e
m
p
A
1
.
1
8
V
a
r
i
a
b
l
e
1
1
.
7
2
8
.
2
9
.
4

1
0
)
5
1
.
0
0
3
.
3

1
0
1
4
2
.
1
5
1
.
4
2
1
.
3
3
1
4
C
.
m
i
n
i
t
a
n
s
O
a
t
s
4
.
3
3
0
.
0
2
7
1
7
.
3
7
1
1
0

1
0
)
5
0
.
9
8
1
1

1
0
1
4
4
.
1
7

4
.
0
5
1
5
C
.
m
i
n
i
t
a
n
s
O
a
t
s
4
.
2
5
0
.
0
2
7
2
4
.
2
8
3
9
.
5

1
0
)
5
0
.
9
8
1
4

1
0
1
4
3
.
8
0

3
.
7
8
2
2
A
.
o
r
y
z
a
e
H
e
m
p
A
1
.
0
1
0
.
0
6
9
4
.
0
1
6
.
7
1
5

1
0
)
5
1
.
0
0

0
.
9
1

0
.
8
8
1
7
A
.
o
r
y
z
a
e
W
h
e
a
t
4
.
5
2
0
.
0
6
9
6
.
0

a
W
a
t
e
r
a
c
t
i
v
i
t
y
o
f
a
t
h
o
r
o
u
g
h
l
y
m
i
x
e
d
s
a
m
p
l
e
o
f
t
h
e
t
o
t
a
l
c
o
n
t
e
n
t
s
o
f
t
h
e
r
e
a
c
t
o
r
.
b
P
r
e
d
i
c
t
e
d
d
e
c
r
e
a
s
e
i
n
w
a
t
e
r
c
o
n
t
e
n
t
c
a
l
c
u
l
a
t
e
d
u
s
i
n
g
t
h
e
p
r
e
d
i
c
t
e
d
O
2
c
o
n
s
u
m
p
t
i
o
n
i
n
t
h
e
r
e
a
c
t
o
r
(

r
s
t
c
o
l
u
m
n
)
o
r
t
h
e
m
e
a
s
u
r
e
d
O
2
c
o
n
s
u
m
p
t
i
o
n
(
s
e
c
o
n
d
c
o
l
u
m
n
)
.
c
H
e
m
p
i
m
p
r
e
g
n
a
t
e
d
w
i
t
h
e
i
t
h
e
r
s
o
l
u
t
i
o
n
A
:
1
0
0
g
/
L
g
l
u
c
o
s
e
a
n
d
2
0
g
/
L
y
e
a
s
t
e
x
t
r
a
c
t
:
o
r
s
o
l
u
t
i
o
n
B
:
3
0
0
g
/
L
g
l
u
c
o
s
e
a
n
d
6
0
g
/
L
y
e
a
s
t
e
x
t
r
a
c
t
.
388 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 4, FEBRUARY 15, 2002
will inevitably lose water during the cultivation due to
evaporation. Ideally, the support should be able to re-
lease this large amount of water without aecting the
water activity. Hemp provided good control of water
activity due to its high water uptake capacity and fa-
vorable sorption isotherm (Weber et al., 1999). A spore
yield of 9 10
14
conidia per m
3
packed-bed was expected
to be feasible (Weber et al., 1999). Our experiments in
the PBR show that the water activity indeed remained
1.0 and that the expected spore numbers were obtained
(Table III). The nal moisture content of impregnated
hemp was suciently high to prevent a drop in the a
w
. It
was anticipated that a higher initial nutrient concen-
tration could be used to impregnate the hemp, which
might lead to an even higher spore yield. However, a
threefold increase in the nutrient concentration resulted
in a drop of the initial water activity. Due to this re-
duction in water activity, the obtained conidia yield was
low (Table III).
The water activity of oats was expected to become
limiting for conidia formation after 24 d of cultivation
(Weber et al., 1999). The water activity at the end of the
cultivation was indeed slightly reduced (a
w
= 0.98), as
was previously predicted (Weber et al., 1999). Despite
the inhibitory a
w
levels at the end of the cultivation, the
conidia production on oats was still about 1.1 10
15
conidia/m
3
(Table III), which is only slightly lower than
that obtained on lab-scale (Weber et al., 1999).
Channeling
Evaporation can negatively aect the cultivation in
various ways. In literature, the attention is primarily
focused on the reduction in water activity. In previous
work, we have emphasized the possibility of shrinkage
and subsequent channeling (Oostra et al., 2000; Weber
et al., 1999). In the current paper, we show that
shrinkage of the solid substrate and channeling can in-
deed have a more signicant eect on the cultivation
than the reduction in water activity.
A disadvantage of grain is that evaporation of water
will cause shrinkage, which we expected to result in
channeling and inhomogeneous aeration (Oostra et al.,
2000; Weber et al., 1999). Indeed, channel formation
was observed when A. oryzae was grown on wheat in the
packed-bed reactor. Evaporation caused shrinkage, and
large air channels appeared between the tightly bound
substrate and the reactor wall (Fig. 6). As a consequence
of the inhomogeneous aeration, the temperature in the
center of the bed could not be controlled and quickly
reached 45C. At the end of the cultivation, almost no
mycelium was visible in the center of the bed. Only near
the channels, where the temperature could be controlled,
was abundant growth observed (Fig. 6). Surprisingly, no
channeling was observed when C. minitans was culti-
vated on oats in the packed-bed reactor. Contrary to
A. oryzae, C. minitans does not form hyphal bridges
between particles. As a consequence, the grains re-
mained free-owing and no channels were observed.
Instead, we observed a reduction of the total bed height.
Our mathematical model cannot be used to predict
temperatures in situations where channeling occurs.
However, it might be extended to predict when shrink-
age will occur. The current prediction of the water
content of the substrate bed can be combined with in-
dependent measurements of substrate volume versus
moisture content (Oostra et al., 2000), to generate a
prediction of bed volume. When combined with obser-
vations from small-scale cultivations on the extent of
hyphal bridge formation, this might be translated into a
rough prediction of channel formation.
Limitations of Evaporative Cooling
Although evaporative cooling is a very ecient way to
remove heat, it has several disadvantages. It may aect
the cultivation by causing a reduction in a
w
or shrinkage
of the substrate, which may cause channeling. Another
disadvantage might be that the evaporation rate of
water cannot be controlled, and might become rate
limiting.
In our model, we assumed that the evaporation rate is
not limiting, and, therefore, the air is expected to be in
equilibrium with the solid matrix at any point in the bed.
The o-gas of the cultivations of C. minitans on hemp
was water-saturated during the whole cultivation, indi-
cating that this assumption is valid. However, when oats
were used as solid substrate, the relative humidity of the
o-gas was only 87% around day 12 (Fig. 7). For this
cultivation our assumption is therefore not valid, and as
a consequence, the measured temperature was higher
than the predicted temperature. Only when the mea-
sured relative humidity of the o-gas was included in the
Figure 6. Channeling in an aerated (F
///
a
= 0.069 kg/m
3
per s) packed-
bed reactor with A. oryzae on wheat (experiment 17). Left: Top view
showing the air channel, (C) between the reactor wall, (W) and the bed
(B) Right: Cross section showing poor mycelium distribution in the
bed resulting from the inhomogeneous aeration.
WEBER ET AL.: LARGE-SCALE PACKED-BED SOLID-STATE FERMENTATION 389
calculations was good agreement between predicted and
measured temperature observed (Fig. 7). The observed
dierence between hemp and oats might be attributed to
the smaller characteristic particle size of the hemp,
which causes (1) a higher specic surface area for mass
transfer, and (2) a smaller mass transfer resistance inside
the particles.
Aspergillus oryzae has a higher growth rate than C.
minitans: consequently, a higher aeration and evapora-
tion rate is required to remove sucient metabolic heat
when A. oryzae is cultivated on hemp. During cultivation
of A. oryzae on hemp in the PBR, it was observed that
evaporation of water also became rate limiting (Fig. 8).
The o-gas was not water-saturated during the whole
cultivation: after about 35 h, the relative humidity was
only 92%. This observation that the air at the outlet is not
always water-saturated implies that the transfer of water
from particle to air is rate limiting. As evaporation of
water is the main mechanism to remove heat, the
eectiveness of evaporative cooling can be seriously
Figure 7. Temperatures (left) and relative humidity of outlet gas (right) in an aerated (F
///
a
= 0.027 kg/m
3
per s) packed-bed reactor with C.
minitans growing on oats (experiment 15). Symbols: () measurements, bottom to top, of temperature of inlet air at 5, 15, 25, and 35 cm bed height
and of outlet air; (

) predicted outlet temperature using measured O

2
consumption and RH of outlet air.
Figure 8. Temperatures (left) and relative humidity (right) in an aerated (F
///
a
= 0.069 kg/m
3
per s) packed-bed reactor with A. oryzae on hemp
(experiment. 22). Symbols: () measurements, bottom to top, of temperature of inlet air at 15 and 35 cm bed height and of outlet air; (

)
predicted outlet temperature using measured O
2
consumption and RH of outlet air.
390 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 4, FEBRUARY 15, 2002
limited by the transfer rate of water from the particle
to air. It is therefore expected that the evaporation rate
required to minimize axial temperature gradients cannot
always be reached. Especially for fast growing fungi,
such as A. oryzae, this means that axial temperature
gradients cannot always be prevented in large-scale
packed-bed reactors. As temperature can have a tre-
mendous eect on the production of, for instance, ex-
tracellular enzymes (Suh et al., 1988) or secondary
metabolites (Cooney et al., 1997), the overall produc-
tivity in a full-scale packed-bed reactor might be lower
than would be expected on the basis of small-scale ex-
periments.
This unexpected limitation demonstrates very clearly
the necessity to experimentally validate a scale-up
strategy. Recently, Mitchell and colleagues (1999) also
proposed a scale-up strategy for packed-bed reactors,
and assumed that air is always water-saturated at su-
percial aeration rates up to 0.10 m/s. We 7 already ob-
served limitations in the evaporation rate at 0.011 m/s
with oats and at 0.021 m/s with hemp. Clearly, the
packed-bed models need to be extended with solids-to-
gas mass transfer kinetics for water.
Predicting Moisture Content
Desiccation of the substrate during cultivation cannot
be prevented and might result in unfavorable conditions.
Our model predicts the moisture content of the substrate
during cultivation. This prediction can be used to eval-
uate whether unfavorably low water activities or sub-
strate shrinkage are expected to occur when a process is
scaled up. The model discriminates between water in
biomass and extracellular water. However, as only the
overall water content could be measured, we have only
compared the predicted and measured overall water
content. Although the model predicts moisture content
during cultivation (Weber et al., 1999), moisture content
measurements can only be done by taking samples from
the reactor bed. As the moisture loss will be highest at
the end of the cultivation, we compared the predicted
and measured decrease in overall water content at the
end of the various cultivations (Table III).
For all cultivations of C. minitans on hemp impreg-
nated with 100 g glucose/L, the measured decrease in
water content was higher than predicted. For all other
cultivations, the measured and predicted values corre-
spond well (Table III). The fact that the deviations be-
tween predicted and measured water content are only
observed for the cultivations of C. minitans on hemp is
remarkable. Both evaporation and metabolic water
production aect the water content of the substrate. As
the model accurately described temperatures in the
PBR, it is expected that the predicted evaporation rates
are correct. The metabolic production of water is small
compared to evaporation. It is therefore not expected
that a fault in this prediction can account for the ob-
served dierence in predicted and measured water con-
tent. The observed dierences could also be caused by
an error in the water-content measurements. Although
the error in the water-content measurements is relatively
large, it is uncertain whether this accounts for the dif-
ferences. The deviation between measurement and pre-
diction is the same for all cultivations of C. minitans
with hemp, and for the other cultivations the predictions
are similar to the measurements. Therefore, it is unlikely
that the error in the moisture analysis causes the ob-
served deviation.
Another striking dierence between the cultivations
on hemp is the higher amount of O
2
consumed by C.
minitans compared to A. oryzae (Table III). The glucose
concentration is expected to be similar for all the ex-
periments, as the same impregnation procedure was
used. C. minitans is known to produce various cell-wall
degrading enzymes (Jones et al., 1974). It is therefore
expected that C. minitans also uses the hemp as sub-
strate, resulting in the higher O
2
consumption. The mi-
crobial degradation of hemp will require water, and this
might cause the observed deviation between measured
and predicted water content. The fact that the nal
water content in the cultivation with C. minitans on
hemp with a higher glucose concentration was correctly
predicted supports this hypothesis. Due to the higher
glucose concentration, it is expected that less hemp is
degraded. However, to explain the large deviation be-
tween measured and predicted water content, the
amount of water used for the degradation of hemp
should have been extremely large. The cause for the
dierence therefore remains uncertain.
CONCLUSIONS
Several models predicting the performance of packed-
bed bioreactors for solid-state fermentation have been
published in recent decades. Most of these models cannot
be used for scale-up studies, as they do not take into
account the most important heat transfer mechanism:
evaporation (Gutie rrez-Rojas et al., 1995; Sangsurasak
and Mitchell, 1995; Saucedo-Castan eda et al., 1990).
Another shortcoming in all previous models is the lack of
a water balance to predict moisture losses (Sangsurasak
and Mitchell, 1998). Also, none of the proposed models
have been properly validated. Several authors tried to
verify their model by tting the model predictions to the
measured temperatures (Sangsurasak and Mitchell,
1998; Saucedo-Castan eda et al., 1990). A t is, of course,
no proof that the models are valid. The results of the
same authors already indicate that their ``validation''
was not properly performed, because when they used
independently determined parameters in their models,
unsatisfactory model predictions were obtained (Sang-
surasak and Mitchell, 1998; Saucedo-Castan eda et al.,
WEBER ET AL.: LARGE-SCALE PACKED-BED SOLID-STATE FERMENTATION 391
1990). The urgent need for proper validation experiments
was also recognized by Mitchell and colleagues (2000) in
their review on design and operation of SSF reactors.
Our model consists of two parts: a biological part and
a physical part. The biological part predicts the activity
of the fungus in response to changing temperatures at
various positions in the bed. The physical part of the
model uses these activities to predict temperatures at
various positions in the bed. Both parts of the model
have been veried.
Reasonable agreement between predicted and mea-
sured O
2
consumption rates was observed. The devia-
tions that were observed may be attributable to the
dynamic response of fungi to changes in temperature,
which was neglected in the model and in the underlying
isothermal kinetic experiments. An eect of the tem-
perature history on fungal growth rates has been re-
ported previously, but remains to be explained (Ikasari
et al., 1999). More research into this area would be re-
quired to further improve the kinetic model.
Very good predictions were obtained when the phys-
ical part of the model was validated using respiration
rates measured in the packed bed. Only at certain con-
ditions was a limitation of the current model identied.
Incorporation of gas-solids water and heat transfer ki-
netics to account for deviations from equilibrium ob-
served with fast-growing fungi would improve the
current model.
Our results clearly show that a proper choice of the
solid medium is essential for a successful process. As-
pergillus oryzae could be successfully cultivated on
hemp, but the cultivation failed when wheat was used as
solid substrate. The high water uptake capacity of hemp
provided good control of water activity, and shrinkage
and channeling were not observed. Also, hemp allowed
higher evaporation rates and thus better control of
temperature. Another important advantage of hemp is
that it can be impregnated with an optimized medium
supporting high product yields (Ooijkaas et al., 2000b).
We believe that our model and the previously pre-
sented design approach using a simplied version of the
current model (Weber et al., 1999) are valuable tools to
evaluate
:
scale-up of SSF processes in packed-bed re-
actors. We hope that continued development of vali-
dated models of this type will contribute to the
development of SSF technology to a feasible alternative
for submerged fermentation.
References
Cooney CL, Wang DI, Mateles RI. 1968. Measurement of heat evo-
lution and correlation with oxygen consumption during microbial
growth. Biotechnol Bioeng 11:269281.
Cooney JM, Lauren DR, Jensen DJ, Perry Meyer LJ. 1997. Eect of
harvest time, temperature, light, and spore inoculum concentra-
tion on 6-n-pentenyl-2H-pyran-2-one production by Trichoderma
spp. J Agric Food Chem 45:28022806.
Gowthaman MK, Raghava Rao KSMS, Ghildyal NP, Karanth NG.
1993. Gas concentration and temperature gradients in a packed
bed solid-state fermentor. Biotechnol Adv 11:611620.
Grajek W. 1988. Cooling aspects of solid-state cultures of mesophilic
and thermophilic fungi. Process Biochem 66:675679.
Gutie rrez-Rojas M, Auria R, Benet CJ, Revah S. 1995. A mathe-
matical model for solid state fermentation of mycelial fungi on
inert support. Chem Eng J 60:189198.
Gutie rrez-Rojas M, Hosn SAA, Auria R, Revah S, Favela-Torres E.
1996. Heat transfer in citric acid production by solid state fer-
mentation. Process Biochem 31:363369.
Hamblin FD. 1971. Abridged thermodynamic and thermochemical
tables: SI units. Oxford: Pergamon press. 79 p.
Ikasari L, Mitchell DA, Stuart DM. 1999. Response of Rhizopus oli-
gosporus to temporal temperature proles in a model solid-state
fermentation system. Biotechnol Bioeng 64:722728.
Jones D, Gordon AH, Bacon JSD. 1974. Co-operative action by endo-
and exo-b-(1-3)-glucanases from parasitic fungi in the degradation
of cell-wall glucans of Sclerotinia scleotiorum (Lib.) de Bary.
Biochem J 140:4755.
NOMENCLATURE
8
a
1
, a
2
b
1
, b
2
Fit parameters see Table II
C
s
concentration of support
in bed
kg dry support/m
3
support
C
wg
concentration of water in
gas phase
kg water/m
3
air
C
x
concentration of biomass in
bed
kg dry biomass/m
3
support
c
pa
specic heat of dry air J/kg dry air per K
c
pwv
specic heat of water vapor J/kg per K
F
a
supercial aeration rate kg dry air/m
2
per s
h
a
enthalpy of (moist) air J/kg dry air
DH
o
reaction enthalpy J/kg O
2
DH
w
evaporation enthalpy water J/kg H
2
O
M
w
molecular weight kg/mole
m
o
maintenance requirements
biomass
kg O
2
/kg dry biomass per s
P
tot
atmospheric pressure Pa
p
w
water vapor pressure Pa
r
///
o
oxygen production biomass kg O
2
/m
3
reactor per s
r
///
w
water production biomass kg H
2
O/m
3
reactor per s
r
///
x
biomass formation rate kg dry biomass/ kg dry
support per s
t time s
T temperature K
T
min
minimum temperature
for growth
K
T
max
maximum temperature
for growth
K
T
ref
reference temperature K
X biomass content kg biomass/kg dry support
X
max
maximum content biomass kg biomass/kg dry support
x
w
water content reactor kg water/m
3
reactor
x
wx
water content biomass kg water/kg dry biomass
x
ws
water content support kg water/kg dry support
Y
xo
yield biomass formation kg biomass/kg O
2
Y
wo
yield water formation kg H
2
O/kg O
2
y
w
water content air kg water/kg dry air
z axial position in bed m
Z
max
height of the reactor bed m
e void fraction m
3
air/m
3
reactor
l
max
maximum specic growth rate 1/s
q
s
density support kg dry support/m
3
reactor
392 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 4, FEBRUARY 15, 2002
Kaye GWC, Laby TH. 1995. Tables of physical and chemical con-
stants. Harlow Essex: Longman. 611 p.
McQuilken MP, Budge SP, Whipps JM. 1997. Eects of culture media
and environmental factors on conidial germination, pycnidial
production and hyphal extension of Coniothyrium minitans. Mycol
Res 101:1117.
Mitchell DA, Krieger N, Stuart DM, Pandey A. 2000. New develop-
ments in solid-state fermentation II. Rational approaches to the
design, operation and scale-up of bioreactors. Process Biochem
35:12111225.
Mitchell DA, Pandey A, Sangsurasak P, Krieger N. 1999. Scale-up
strategies for packed-bed bioreactors for solid-state fermentation.
Process Biochem 35:167178.
Okazaki N, Sugama S, Tanaka T. 1980. Mathematical model for
surface culture of Koji mold. J Ferment Technol 58:471476.
Ooijkaas LP, Buitelaar RM, Tramper J, Rinzema A. 2000a. Growth
and sporulation stoichiometry and kinetics of Coniothyrium mini-
tans on agar media. Biotechnol Bioeng 69:292300.
Ooijkaas LP, Tramper J, Buitelaar RM. 1998. Biomass estimation of
Coniothyrium minitans in solid-state fermentation. Enzyme Mi-
crobiol Technol 22:480486.
Ooijkaas LP, Weber FJ, Buitelaar RM, Tramper J, Rinzema A. 2000b.
Dened media and inert supports: their potential as solid-state
fermentation production systems. Trends Biotechnol 18:356
360.
Oostra J, Tramper J, Rinzema A. 2000. Model-based bioreactor se-
lection for large-scale solid-state cultivation of Coniothyrium
minitans spores on oats. Enzyme Microbiol Technol 27:652663.
Pandey A, Soccol CR, Mitchell D. 2000. New developments in solid
state fermentation: lbioprocesses and products. Process Biochem
35:11531169.
Perry RH, Green DW, Malony JO. 1984. Perry's chemical engineers'
handbook. New York: McGraw-Hill. 2301 p.
Pirt SJ. 1965. The maintenance energy of bacteria in growing cultures.
Proc Soc London 163:224231.
Ratkowsky DA, Lowry RK, McMeekin TA, Stokes AN, Chandler
RE. 1983. Model for bacterial culture growth rate throughout the
entire biokinetic temperature range. J Bacteriol 154:12221226.
Roels JA. 1983. Energetics and kinetics in biotechnology. Amsterdam:
Elsevier. 330 p.
Sangsurasak P, Mitchell DA. 1995. Incorporation of death kinetics
into a 2-dimensional dynamic heat transfer model for solid state
fermentation. J Chem Technol Biotechnol 64:253260.
Sangsurasak P, Mitchell DA. 1998. Validation of a model describing
two-dimensional heat transfer during solid-state fermentation in
packed bed bioreactors. Biotechnol Bioeng 60:739749.
Sato K, Nagatani M, Sato S. 1982. A method of supplying moisture to
the medium in a solid-state culture with forced aeration. J Ferment
Technol 60:607610.
Saucedo-Castan eda G, Gutie rrez-Rojas M, Bacquet G, Raimbault M,
Viniegra-Gonza lez G. 1990. Heat transfer simulation in solid
substrate fermentation. Biotechnol Bioeng 35:802808.
Suh DH, Becker TC, Sands JA, Montenecourt BS. 1988. Eects of
temperature of xylanase secretion by Trichoderma reesei. Bio-
technol Bioeng 32:821825.
Weber FJ, Tramper J, Rinzema A. 1999. A simplied material and
energy balance approach for process development and scale-up of
Coniothyrium minitans conidia production by solid-state cultiva-
tion in a packed-bed reactor. Biotechnol Bioeng 65:447458.
Whipps JM, Gerlagh M. 1992. Biology of Coniothyrium minitans and
its potential for use in disease biocontrol. Mycol Res 96:897
907.
WEBER ET AL.: LARGE-SCALE PACKED-BED SOLID-STATE FERMENTATION 393