BIOTECHNOLOGY AND BIOENGINEERING, VOL.

XIX, PAGES 425433 (1977)

An Analysis of Extended and Exponentially-Fed-Batch Cultures

HENRY C. LIM, BILL J. CHEN, and CHRISTOPHER C. CREAGAN, School of Chemical Engineering, Purdue University, West Lafayette, Indiana 47907

Summary
Under certain conditions it is shown that an extended culture is equivalent to an exponentially-fed-batch culture, that an exponentially-fed-batch culture (and an extended culture) can be maintained a t a steady state and that an exponentially-fed-batch culture may be mimicked by a continuous-flow culture with a constant dilution rate. Operational conditions required to maintain steady states are specified.

INTRODUCTION
Semicontinuous operation of batch cultures in which a medium containing the growth-limiting substrate is fed continuously has been developed empirically for such industrial applications as penicillin-3 and bakers yeast productions3 and waste disposals. This type of culture operation was termed by Yoshida et al.4 as Fed-Batch Culture, who tried fed-batch fermentation of hydrocarbon and reported that the cell/substrate yield during the linear growth phase was about twice as high as that during the exponential growth phase. Edwards e t aL5 gave a brief treatment of what they termed extended culture, a fed-batch culture in which the limiting substrate concentration is kept constant by manipulating the feed rate of the substrate. They also gave various examples of industrial practices of extended culture. Apparently, the fedbatch culture may be operated at a constant feed rate, a programmed feed rate, or a n exponentially increasing feed rate.4-6 I n fact, a device had been proposed which delivers exponentially increasing amounts of nutrients and allows precise control of bacterial growth rate.6 Pirt~8 presented arguments to show that when the specific growth rate of organisms follows that of a Monod type, a fed-batch culture may reach a quasi-steady state, which is defined as dx/dt

425

01977 by John Wiley & Sons, Inc.

426

LIM, CHEN, AND CREAGAN

0 and ds/dt G 0 and called this a Quasi-Steady-State Fed-Batch Culture. Ryu and Humphreyg used substrate feeding to control the oxygen demand of a culture while Calam and Russell10 discussed a computer model of a penicillin fermentation. A means of overcoming catabolite repression of product formation by substrate feed rate has been considered by Demain. I n spite of the importance of fed-batch cultures and their unique features, theoretical analysis of fed-batch cultures has not been considered fully, with the exception of the work of Dunn and Mor,I2 who discussed and gave analyses of constantly-fed-batch cultures including the analogy with a continuous-flow culture of decreasing feed flow rate. Apparently exponentially-fed-batch culture, in spite of its ability to precisely control the bacterial growth rate, has not been analyzed fully. Edwards et al.5 gave a brief treatment of extended culture. It is the purpose here to consider theoretically extended cultures and exponentially-fed-batch cultures and to show that: 1) under certain conditions a n extended culture is a n exponentially-fed-batch culture, 2) operational conditions under which a n exponentially-fed-batch culture and a n extended culture may be reduced to ((steady-state fed-batch cultures ( d z l d t = 0 = d s / d t ) , and 3) the performance of a n exponentially-fed-batch culture may be mimicked by a continuousflow culture operating at a constant dilution rate.

ANALYSIS
We begin by writing pertinent material balance equations for fed-batch cultures. Henceforth we shall assume that the cultures are run under constant environmental conditions including pH, temperature, and other nutrients except the limiting substrate. We shall also assume that the culture is pure and there is only one limiting substrate. For convenience we will not consider any product formation and consider the biomass (cell) as the product as in the case of single-cell protein production.

Fed-Batch Culture Consider a batch culture into which a nutrient medium containing a fixed concentration of a limiting substrate SF is fed continuously at a flow rate of Ffb. Initially the culture volume is Vo and the culture contains fixed concentrations of cells and substrate, zo and so. The maximum culture volume is VmaX. The appropriate balance equations during the filling time period, 0 5 t _< t,
(/fa=

~Ib(r)

dT

_<

vmaX v,),are-

EXTENDED AND EXPONENTIALLY-FED-BATCH CULTURES 427

and

where u is the specific substrate consumption rate, p is the specific growth rate, s is the limiting substrate concentration a t any time t, x is the cell concentration, and V f b is the fed-batch culture volume a t any time t. Any change in density is assumed negligible. Substitution of eq. (3) into eqs. (1) and (2) yields

(-+)
and

(sp

- s)

ux =

-dt

ds

s(0) = so

(4)

Provided that u and p depend on the limiting substrate concentration under constant environment but not on the cell age distributions, eqs. (3)-(5) describe the dynamics of fed-batch culture. Otherwise, an additional cell age distribution equation as well as the precise dependency of u and p on the age distribution must be known. We assume that under constantly kept environmental conditions u and p are dependent only on the limiting substrate concentration and ignore the age distribution.

Exponentially-Fed-Batch Cultures Apparently the fed-batch cultures are operated either at a constant feed rate or a t an exponentially increasing feed rate.4-6 Constantlyfed cultures have been thoroughly analyzed by Dunn and Morl* and no further analysis is made here. Instead, exponentially-fedbatch cultures are treated. For a fed-batch culture which is fed at an exponential feed rate, F f b e = aebt, where a and b are arbitrary positive numbers, eq. (3) reduces to

428

LIM, CHEN, AND CREAGAN

and the dynamics are given by eqs. (7) and (8),

b(SF - 8 ) ( ( V o b - a)e-bf/a 1

ux =

- dt

ds

s(0) = so

(7)
(8)

bx ( ( V o b - a)e-bt/a 1

- px

=-

dx dt

x(0)

= 50

Thus, the dynamics of exponentially-fed-batch cultures is described by eqs. (6)-(8). These equations are utilized below to analyze limiting behaviors of exponentially-fed-batch cultures.

Extended Culture and Analogy to Exponentially-Fed-Batch Culture An extended culture is a fed-batch culture in which the feed flow rate is continuously manipulated to keep the limiting substrate concentration in the culture constant at all times. Therefore, the analysis of extended cultures can be made using the equations for fed-batch cultures, eqs. (1)-(5) , with appropriate subscripts. Let us denote by sex the desired limiting substrate concentration, F,, is the flow rate required to keep the substrate concentration constant a t sex,and V e xis the volume of extended culture. Then, the required flow rate is obtained by setting ds/dt = 0 in eq. (4) : Fex = Q X V ~ X / - sFx )(~ e (9) Even with this flow rate there would be a transient period in the substrate concentration unless the initial concentration is also a t the desired value, i.e., s(0) = sex. Therefore, we need to start initially with the desired substrate concentration. If u depends on only the limiting substrate concentration which is kept at sex, u evaluated a t s = sex, u(sex)] is also constant. When the concentration of substrate in the feed is constant, eq. (9) suggests that the flow rate F,, should vary in proportion to the total biomass in the culture xVex,i.e., Fex
=

[u(sex)/(sF- sex)IzVex

S(OX . Sex=

SO

(10)

Hence, if the total biomass increases exponentially, the flow rate must also be increased exponentially to keep the substrate concentration constant. Under the assumption that the specific growth and substrate consumption rates depend on the limiting substrate concentration only, keeping the limiting substrate concentration constant implies that a(seX) and p(s,,) are also constant. Therefore, eq. ( 2 ) can be integrated to obtain
ZV,, =
= so zoVoep(*e~)f sex

(11)

EXTENDED AND EXPONENTIALLY-FED-BATCH CULTURES 429

which states that the total biomass increases exponentially. Substitution of eq. (11) into eq. (10) yields

Fex = [(T(s,,)zOVO/(S~ sex)]er(sex)t= FexOer(sex)f-

(12)

Therefore, when (T and p depend only on the limiting substrate concentration, the extended culture, in which the substrate concentration is kept constant at its initial value, so = sex, is equivalent to an exponentially-fed-batch culture with the flow rate given by eq. (12). Note that Edwards et al.5 in their interpretation of the observation made by Martin and Felsenfeld6 assumed that there was a n arbitrary exponential function of the form Fex = F o p t , while the above analysis shows definite restrictions on the parameters Fo and a to keep the limiting substrate concentration constant at sex for all time. Indeed, the initial flow rate FexOdepends on the initial amount of inoculum zOVO, the initial substrate concentration sex = so, the initial specific substrate consumption rate c(seX),and the feed substrate concentration S F . The cell concentration history is obtained by solving eq. (11) for z,
z = zo(Vo/V)ea~t
=

zo/[(l -

c)e-(sex)t

+ c]
- (FexolVo)

(13)

where
c=
Q(S e x ) zo p(Sex)(Sp

- sex)

Sex

SO

p(sex)

where Y ( s e x ) is the biomass/substrate yield coefficient evaluated a t s = sex. Equation (13) suggests that the cell concentration can increase, decrease, or remain constant a t ZO depending upon the numerical value of c. When c is negligibly small, the cell concentration can increase exponentially since the increase in the culture volume over the initial volume is negligible over a short time interval. Physically, this can happen when the initial dilution rate required t o keep the substrate concentration a t sex as predicted by eq. (12) is negligibly small as compared to the initial specific growth rate (FeXo/Vo << p ( s e x ) ) . Alternatively c is negligibly small when the potential cell production, Y(s,,)(sF - sex) /: Fex d7, is much greater than that which is needed t o keep the cell concentration at ZO, zo Fex dr. When the initial dilution rate is greater than the initial specific growth rate of cells, i.e., c > 1, the cell concentration

430

LIM, CHEN, AND CREAGAN

decreases as it should and reaches asymptotically xo/c. A very interesting situation can arise if c can be chosen to be unity. This case is unique and will be considered in the next section.

Extended Culture and Analogy to Steady-State Fed-Batch Culture According to eqs. (13) and (14), if one can match the initial dilution rate of the exponential feed Fexo/Vowith the initial specific growth rate p(so), which can happen when Y ( s e x ) ( s ~ sex) = ZO,then c = 1 and not only the substrate concentration but also the cell concentration may be kept constant a t the initial value x0. I n other words, one can feed the limiting substrate at such a rate that the cell generated and substrate consumed will keep up with the increasing culture volume in such a way as to keep the cell and substrate concentrations constant. Under this condition the extended culture becomes a steady-state fed-batch culture. Since x = xo and s = so = sex,the flow rate is Fex(t)= p(sex)Voe'(de')t p ( ~ o ) V ~ e " ( ~ " )(15) ~=
and the cell and substrate balance equations (eqs. (4) and (5)) require that
20

c((sex)(sF- sex)/a(sex) = y(sex)(sF -

Sex)

= Y(SO)(SF

- so)

(16)

which is a constraint among xo, sF, and so. Not all arbitrary initial conditions are allowed. With the above feed rate and starting initial conditions the culture volume increases exponentially,

Vo

the cell and substrate concentration remain invariant, x = zo and s = so, and the dilution rate is constant,

1:

= Voec(80)1 F e x ( r )dr

(17)

FexIVex

= ~(80)

(18)

Thus, the extended culture of initial volume V o and the initial cell and substrate concentrations x0 and so, respectively, which meet the constraint of eq. (16), when fed with the exponential feed rate of eq. (15), reduce to a steady-state fed-batch culture in which the substrate and cell concentrations remain constant a t the initial values. This type of operation not only keeps the environmental factors constant but also keeps the cell concentration constant, eliminating any potential interaction factor of changing cell densities.

EXTENDED AND EXPONENTIALLY-FED-BATCH CULTURES 431

Indeed, this situation is analogous t o a steady-state chemostat with the exception that there is a potential difference in age distribution as will be discussed later. It should be possible to utilize this situation which is unique to exponentially-fed-batch culture as one would in taking advantage of the quasi-steady state for constantly-fed-batch cultures. However, it should be noted that with constantly-fed-batch cultures the steady state may be reached only after a suitably long time, while with exponentially-fed-batch cultures the steady state can be established for all times.

Analogy Between Exponentially-Fed-Batch Culture and Continuous-Flow Culture The analogy between constantly-fed-batch cultures and dynamic continuous flow cultures was discussed by Dunn and Mor.12 Interesting observations unique to exponentially-fed-batch cultures that have not been pointed out previously will be considered here. As pointed out by Dunn and Mor,12eqs. (4) and (5) also represent unsteady-state substrate and biomass equations for a continuousflow culture provided that the dilution rates are equal, i.e., Fc/Vc = Ffb/Vfb (19)

For a constantly-fed-batch culture, Ffb/Vfb decreases continuously with time so that in order for the analogy t o hold the flow rate to the continuous-flow culture must also be made t o decrease with time. Therefore, it can be stated that the transients are caused by the initial concentrations xo and so and also b y the decreasing flow rate F,. For the exponentially-fed-batch culture with the flow rate given by Fibs = aebt, the corresponding flow rate for the continuous culture is given by

F, =

bVc (Vob - a)edbt/a

+1

(vmaX-

+ 1) (20)

The flow rate starts a t a V , / V o and approaches bV, .according t o eq. (20). Unlike constantly-fed-batch cultures which can be mimicked by continuous-flow cultures with a decreasing flow rate, the performance of exponentially-fed-batch cultures can be mimicked by continuous-flow cultures with decreasing as well as increasing or constant flow rates, depending upon the quantity Vob - a. When Vob - a > 0, the flow rate must increase with time, while the flow rate must decrease when Vob - a

< 0. When Vob = a, the

432

LIM, CHEN, AND CREAGAN

corresponding flow rate t o the continuous-flow culture must be constant. Thus the performance of the exponentially-fed-batch culture with the feed rate given by

can be mimicked by the use of a continuous-flow culture of constant volume V c and a constant flow rate

Unlike the mimicking of constantly-fed-batch cultures by continuousflow cultures, in this situation the transients in the continuous-flow culture are caused strictly by the initial substrate and biomass concentrations and not by the flow rate (therefore, the dilution rate) which must be kept constant. It is apparent that at least in theory the dynamics of any fed-batch culture, a constant-feed fed-batch culture, a n extended culture, a n exponential-feed fed-batch culture, or a steady-state fed-batch culture may be mimicked by the performance of equivalent continuous-flow culture as long as the feed rate t o the fed-batch culture as well as the initial conditions are known. Of course, the earlier assumptions that u and p are functions of the limiting substrate must hold. Otherwise the analogy breaks down.

DISCUSSION
Various types of fed-batch cultures were analyzed and certain equivalences were established under the assumption that u and p are functions of the limiting substrate concentration only. Under specified conditions it was shown that an extended culture may be equivalent to an exponentially-fed-batch culture. Operational conditions were also specified which should force a n exponentiallyfed-batch culture (and therefore also an extended culture) to behave as a steady-state fed-batch culture in which the biomass and substrate concentrations remain constant for all times. I n practice, however, establishment of these conditions requires exact knowledge of u and p a priori. Under the same assumption on u and p the substrate and cell concentration dynamics of fed-batch cultures may be mimicked by constant-volume continuous-flow cultures in which the transients are caused by the initial subst,rate and cell concentrations and/or the feed flow rate which may decrease, increase, or remain constant.

EXTENDED AND EXPONENTIALLY-FED-BATCH CULTURES 433

It is particularly interesting to observe that a continuous-flow

culture with a constant dilution rate can mimic the performance of an exponentially-fed-batch culture. However, if the mixing in the continuous-flow culture is not perfect, i.e., instantaneous and homogeneous, then the age distribution in the continuous-flow culture from which the cells are continually removed may be different from that in the fed batch from which no cells escape. Under this situation the difference in cell age distributions could alter the specific substrate consumption and specific growth rates and make each culture uniquely different. Conversely, the differences, if any, between the fed-batch culture and the corresponding continuous culture could be attributed to the differences in the properties of cell populations with different cell age distributions.
This work waa supported in part by a grant from the National Science Foundation, Grant No. ENG 75-17796.

References
1 A. J. Moyer, U.S. Patent 2,442,141 (May 25, 1948).. 2. P. Hosler and M. J. Johnson, Ind. Eng. Chem., 45, 871 (1953). 3. S. C. Prescott and C. D. Dunn, Industrial Microbiology, 3rd Ed., McGrawHill, Inc., New York, 1959. 4. Y . Yoshida, T. Yamane, and K. Nakamoto, Biotechnol. Bioeng., 15, 257 (1973). 5. V. H. Edwards, M. J. Gottachalk, A. Y . Noojin, 1 1 L. B. Tuthidl, and1, A. L.Tannahill, Biotechnol. Bioeng., 12,975 (1970). 6. R. G. Martin and G. Felsenfeld, Anal. Biochem., 8, 43 (1964). 7 S. John Pirt, J. Appl. Chem. Biotechnol., 24. 415 (1974).. 8. S.John Pirt, Principles of Microbe and Cell Cultivation, John Wiley & Sons, New York, 1975. 9. D. Y . Ryu and A. E. Humphrey, J . Femnent. Technol. Osaka, 50,424 (1972). 10. C. T. Calam and D. W. Russell, J . Appl. Chem. Biotechnol., 23,225 (1973). 11. A. L. Demain, J . Appl. Chem. Biotechnol., 22, 345 (1972). 12. I. J. Dunn and J. R. Mor, Biotechnol. Bioeng., 17, 1805 (1975).

Accepted for Publication October 25, 1976