Kasetsart J. (Nat. Sci.

) 32 : 333 - 338 (1998)

Bioreactor Performance and Design of Large Scale

Packed-Bed Solid-State Fermentation
Penjit Srinophakun and Thongchai Srinophakhun

ABSTRACT

Bioreactor performance and design was investigated by simulating a transient two-dimensional

heat transfer model. The model was able to describe the microbial growth and death kinetics and the energy
balance in both axial and radial directions at any location in packed-bed solid-state fermentation. The
emphasis of heat conduction, convection and the effect of evaporation made the model particularly useful
and practical for large-scale investigation in which the overheating is critical. The system was simplified
by assuming pseudo-homogeneous and constant values of bed physical and thermal properties. Orthogonal
collocation technique was applied to solve the equations. The characteristic of temporal temperature and
biomass concentration profiles in two dimensions were predicted.
The effects of various design and operating variables on the performance of the 10 m3 were
explored, with the aim of identifying strategies to minimize overheating at the top of the bed. Superficial
velocity and inlet air temperature were found to have significant effects on the bioreactor performance.
Superficial velocities of 0.08 to 0.1 m/s are required for effective convective heat removal. A low inlet air
temperature (308 to 312 K) leads to overheating after 10-20 h which cause all organisms to die. The
geometric ratios of 1.0 and below minimized overheating problems and lead to high growth rates.
Key words : large-scale bioreactor, bioreactor performance and design, solid-state fermentation

INTRODUCTION fermented foods. However, these processes have

been developed and optimized by trial and error
Solid-state fermentation (SSF) involves the over many generations, and therefore the
growth and metabolism of microorganisms on moist fundamental processes occurring are not well
solid materials, in the absence or near absence of understood. Some processes have been
free water (Moo-Young et al., 1983; Mitchell and industrialized, such as the koji stage of soy sauce
Lonsane, 1992). This limited availability of water manufacture, however the technology that has
makes SSF fundamentally different from been developed is largely proprietary.
submerged liquid fermentation (SLF). The performance of SSF processes depends
Solid-state fermentation is relatively rare in on interactions between the 3 main components of
western countries, but has been practiced for the system: the microorganism, the substrate and
thousands of years in Asia, in the production of the bioreactor (Sangsurasak and Mitchell, 1995a;

Department of Chemical Engineering, Kasetsart University, Bangkok 10900, Thailand.

334 Kasetsart J. (Nat. Sci.) 32 (3)
1995b). This study will investigate the process MATHEMATICAL MODEL
from the perspective of bioreactor performance
and design. The mathematical model was developed to
describe the system in Figure 1. The model consists
SYSTEM of two parts; growth and death kinetics and the heat
transfer phenomena. Biomass is divided into two
The bioreactor is a cylindrical bioreactor, types in this investigation, viable biomass or
packed with a bed of substrate particles of uniform reproductive cells and dead biomass or non-
size as shown in Figure 1. These substrate particles reproductive cells. The sum of viable and dead
are assumed to have the properties of starch at 50% biomass, therefore, becomes the total biomass. A
moisture content, such as might be expected if common empirical kinetic model, the logistic
sago-beads are used as a substrate, as was done by equation, was used to describe microbial growth in
Gumbira-Sa’id et al. (1991). The bed is treated as this investigation.
a pseudo-homogeneous matrix. Therefore the The equations are transformed into
thermal and physical properties of the bed are dimensionless form and all the parameters are
calculated as a weighted sum of the properties of grouped and converted into dimensionless terms as
the substrate particles and the air in the bed. The follows.
thermal and physical properties of the pseudo- z
homogeneous bed were assumed to be independent Z = , z = ZH
H
of temperature, which is reasonable given the
r
relative small range of less than 20°C which will be R = , r = RRo
Ro
experienced in the bed.
vz . t z. τ
τ = , t = (1)
z vz

 T − Ti 
θ =   , T = [θ.(Tsurr–Ti)]+Ti
 T surr − T i 

ρ b .C p,b . v z . d p
Pe =
kb
where z is the bed height position
Z is dimensionless bed height
H is the bed height (as indicated in
Figure 1)
r is the radial position
R is dimensionless bed radius
R0 is the overall bed radius (as indicated
in Figure 1)
Figure 1 Packed bed bioreactor. τ is dimensionless time
θ is dimensionless temperature
 Kasetsart J. (Nat. Sci.) 32 (3) 335

T is the temperature The dimensionless heat generation term, Q, is a

Ti is the initial temperature of the bed function of Pe number.
Tsurr is the temperature of the surroundings dp.H
or coolant Q = ( .q (5)
Pe. k.(T b − T i )
Pe is the dimensionless Peclet number
dp is the particle diameter Equation (6), (7) and (8) show the dimensionless
vz is the superficial air velocity total, viable and dead biomass production rate
Cp,b is the bed heat capacity H.µ g
dBt
kb is the bed thermal conductivity = . Bv .(1 − Bt ) (6)
dτ vz
ρb is the bed density

Dimensionless equations
dBv
dτ
=
H
vz
[ ]
(µ g .(1 − Bt ) − µ d ). Bv (7)

By substituting the dimensionless terms

(equation 1) into the equations, a set of dBd H.µ d
= ( ). Bv (8)
dimensionless equations were obtained. The heat dτ vz
transfer equation becomes: where Bt is the dimensionless total biomass
ρma . C p,ma + C p,a . ƒ. λ ∂θ (Xt/Xm)
∂θ dp.H
= −( ) +( ) Bv is the dimensionless viable biomass
∂τ ρ b . C p,b ∂Z Pe, R 20
(Xv/Xm)
Bd is the dimensionless dead biomass
 1 ∂θ ∂θ  d p ∂2θ
 ⋅ +  +( ) +Q (2) (Xd/Xm)
 R ∂R ∂R 2 Pe. H ∂Z 2

NUMERICAL METHOD
initial condition
τ = 0, θ = 0 (3) The equations are solved numerically by a
mixture of orthogonal collocation method
boundary conditions (Finlayson, 1980; Villadsen and Michelsen, 1978;
∂θ Sangsurasak et al., 1993) and the Backward Euler
R = 0, = 0 (4)
∂R method (Hindmarsh, 1974). The model is verified
by comparing with the experimental data of
∂θ
R = 1, = Bi(1–θsurr) Saucedo-Castaneda et al. (1990) Sangsurasak,
∂R
(1996), Sangsurasak et al., (1997).
Z = 0, θ = θa
∂θ RESULTS AND DISCUSSION
Z = 1, = 0
∂Z
where Bi is the dimensionless Biot number The demonstration on how the model can
θsurr is the dimensionless temperature of give an insight into packed bed performance and
the surroundings or coolant the effect of size and operational and design
θa is the dimensionless temperature of parameters on the performance is developed. The
air system of interest is the growth of Rhizopus
oligsporus on the starchy substrates. Simulations
336 Kasetsart J. (Nat. Sci.) 32 (3)
were done for a large bioreactor size of 10 m3 and 298 K leads to the lowest maximum bed temperature
the simulation time is 90 hours. This two- but this bed temperature is close to the optimum for
dimensional geometry enables temperature, and growth (Figure 3a). Therefore, the growth rate is
viable and dead biomass profiles to be predicted for the highest (Figure 3b) and there is negligible death
any location inside the bed. However, only the (Figure 3c). At lower regions of the column the
collocation point in the upper region of the bed temperature will be lower than at the upper region
where the highest temperatures occur will be leading to a slower growth rate. With the inlet air
investigated and discussed in this paper. temperature close to the optimum for growth (308-
Figure 2 demonstrates the effect of air 312 K), rapid growth occurs for a short period of
velocity on temperature together with viable and 10-20 hours (Figure 3b), but the temperature quickly
dead biomass concentration. At an air velocity of increases to values (Figure 3a) which cause all the
0.02 m/s the temperature increases significantly organisms to die (Figure 3c).
during the early stages of fermentation (Figure 2a), Figure 4 indicates that the strategy of
leading to the death of all organisms by 20 hours maintaining the superficial air velocity while
(Figure 2c). Superficial air velocities as high as decreasing the H/D ratio (height per diameter ratio)
0.08 to 0.1 m/s are required to maximize the viable will give significantly better performance for packed
biomass density obtained (Figure 2b). bed bioreactor in SSF. At H/D ratios of 1.0 and
Figure 3 shows the effects of inlet air below, the temperature profiles were less
temperature on system temperature and cell pronounced (Figure 4a) and good growth (Figure
concentrations. The inlet air temperature as low as 4b) and negligible death (Figure 4c) were predicted.

Figure 2 The dynamic effect of air velocity on the system temperature (a), viable biomass (b), and dead
biomass (c).
 Kasetsart J. (Nat. Sci.) 32 (3) 337

Figure 3 The dynamic effect of inlet air temperature on the system temperature (a), viable biomass (b),
and dead biomass (c).

Figure 4 The dynamic effect of bioreactor geometric raio on the system temperature (a), viable biomass
(b), and dead biomass (c).
338 Kasetsart J. (Nat. Sci.) 32 (3)
CONCLUSION Publishers, Essex.
Mitchell, D.A. and B.K. Lonsane. 1992. Definition,
The transient two-dimensional heat transfer characteristics and potential, pp. 1-13. In H.
model developed can be used as a tool aiding the W. Doelle, D. A. Mitchell, and C. E. Rolz
design, operation and scale-up of packed bed which (eds.). Solid Substrate Cultivation. Elsevier
can be used to inform design decisions for larger Science, Publishers, Essex.
scale packed bed bioreactors. Superficial air Sangsurasak, P. 1996. Heat transfer model of packed
velocity, inlet air temperature and geometric ratio bed bioreactors in solid-state fermentation,
are predicted to have significant effects on the Ph.D. thesis, Univ. of Queensland, Queensland,
temperature and growth in the upper regions of the Australia.
packed bed. Sangsurasak, P., D.A. Mitchell, and Md. M.
Hossain. 1993. Transient two dimensional
LITERATURE CITED heat conduction by orthogonal collocation
technique. International Communications in
Finlayson, B.A. 1980. Nonlinear Analysis in Heat and Mass Transfer 20:557-566.
Chemical Engineering. McGraw-Hill, New Sangsurasak, P. and D.A. Mitchell. 1995a. The
York. 366 p. investigation of transient multidimensional
Gumbira-Sa’id, E., H.W. Doelle, P.F. Greenfield, heat transfer in solid-state fermentation. Chem
and D.A. Mitchell. 1991. Protein enrichment Engineering J. 60:199-204.
of sago starch by solid-state fermentation. Sangsurasak, P. and D.A. Mitchell. 1995b.
World J. Microb. and Biotech. 7:419-427. Incorporation of death kinetics into a 2-
Hindmarsh, A.C. 1974. GEAR: Ordinary dimensional dynamic heat transfer model for
Differential Equation System Solver, Lawrence solid-state fermentation. J. Chem Technolo
Livermore Laboratory Report UCID-30001, Biotech. 64: 253-260.
Rev. 3. 167 p. Sangsurasak P., T. Srinophakhun, and D. A.
Moo-Young, M., A.R. Moreira, and R.P. Tengerdy. Mitchell. 1997. Mathematical modelling
1983. Principles of solid-substrate development of packed bed bioreactors in
fermentation, pp. 117-144. In J. E. Smith, D. solid-state fermentation. Kasetsart J. (Nat. Sci.)
R. Berry, and B. Kristiansen (eds.). The 31: 452-458.
filamentous fungi vol IV (Fungal Technology). Villadsen J. and M.L. Michelsen. 1978. Solution of
Edmond Anold. Differential Equation Models by Polynomial
Mitchell, D.A. 1992. Growth patterns, growth Approximation. Prentice-Hall. New Jersey.
kinetics and the modelling of growth in solid- 445 p.
state fermentation, pp. 87-114. In H. W. Doelle,
D. A. Mitchell, and C. E. Rolz (eds.). Solid Received date : 7/08/97
Substrate Cultivation. Elsevier Science, Accepted date : 18/05/98