Original Research
Modeling and Simulation of Substrate and Product Inhibitions
of Fermentation of Cassava (Manihot Esculenta) Extract
Nana Y. Asiedu,1 Neba F. Abunde,2 Ahmad Addo2
1
Department of Chemical Engineering and 2Department
of Agricultural Engineering, Kwame Nkrumah University
of Science and Technology, Kumasi, Ghana
Abstract
The fermentation of cassava extracts in brewing is often
hampered by microbial inhibitors and oscillations from prod-
uct and substrate inhibitions, which often decrease ethanol
yields. Previous evaluation of the modeling of ethanol inhibi-
tion during fermentation of cassava extracts showed low model
accuracy. This low model accuracy is due to substrate inhi-
bition, which increases process control challenges. The inhi-
bition patterns considered in this work were linear, sudden
growth stop, and exponential. Mathematical models were de-
veloped to simulate substrate and product inhibitions during
the fermentation of cassava extracts for alcohol production.
The results obtained showed substrate and product inhibitions
exist during fermentation of cassava extracts, with inhibition
patterns being linear substrate-exponential product, exponen-
tial substrate-exponential product, and linear substrate-linear
product inhibitions. The obtained models are described with a
high degree of accuracy (99% confidence interval). The de-
veloped models thus agree well with experimental data.
Keywords: fermentation, mathematical modeling, inhibition
patterns, simulation
Introduction
F
ermentation principles consist of exploiting the meta-
bolic reactions that take place in the cell of a micro-
organism for the production of valuable products.1 For
ethanol fermentation, the three most common yeast
strains usually used include: ale yeasts, which are top-fermenting,
can ferment at higher temperatures, and tend to produce more
esters; lager yeasts, which are bottom fermenting, withstand lower
temperatures, and produce a crisper taste; and wild yeasts, which
produce a lot of unusual compounds and contribute to a horse
sweat flavor that is more acidic. In order to activate the
metabolic pathways of interest within the cell, specific envi-
ronmental conditions, namely temperature, pH, and nutrient
concentration, are applied to enable the yeast cell to grow and
produce ethanol. However, due to the dynamic nature of the
DOI: 10.1089/ind.2016.0001 M A R Y A NN L I E B E R T , I N C .  VOL. 12
culture medium, yeast cells often suffer from various stresses
from both the environmental conditions and from product and/
or substrate imbibition.
Cassava (Manihot esculenta) is one of the most important
food crops in the humid tropics, and its high starch content
makes it extremely suitable for alcohol industries seeking a
local substitute for imported barley.2 Though alcohol pro-
duction using cassava extracts involves several unit opera-
tions, the fermentation step requires near-optimal process
conditions for microorganisms to grow, multiply, and produce
the desired ethanol quality.3 However, studies regarding bio-
reactors and cassava extracts have shown that suboptimal
conditions resulting from intrinsic microbial inhibitors and
oscillations from product and substrate inhibition often affect
fermentation performance.49 These inhibitions increase the
residual sugar at the end of the fermentation, decreasing raw
material consumption, and correspondingly, the ethanol yield
if no economically acceptable attenuation strategies are de-
veloped.9 In a typical procedure for modeling ethanol fer-
mentation, if inhibition is considered, it is often conventional
to predefine the inhibition pattern. However, this practice
increases uncertainties in the model since inhibition pattern
varies depending on the type of microorganism and on the type
and strength of fermentation wort.10 It also increases the un-
reliability of process controllers and simulators, since these
automatic tools are usually based on a mathematical repre-
sentation of the considered system.3
In recent years, there has been extensive research to ex-
plore locally sourced raw materials as alternatives to barley in
brewing to create more affordable beer and more opportuni-
ties for local stake holders. Cassava has been among the can-
didate raw materials, and many breweries are already using it
for beer production. However, several studies examining the
dynamics of bioreactors have reported oscillations in both aer-
obic and anaerobic cultures of Saccharomyces cerevisiae; such
instability is more complicated in ethanol fermentation sys-
tems due to ethanol inhibition and the lag response of yeast
cells to this inhibition.59 Most research geared toward explor-
ing these problems has been qualitative in naturefocusing
on the intracellular mechanisms and how they affect cellular
growthand often predefine the mathematical pattern of the
inhibition.
This paper therefore describes dynamic models, incorporat-
ing the various effects of substrate- and product-inhibition pat-
terns. These models could be helpful in determining optimal
control systems that can minimize the effect of such inhibitions
during fermentation processes.
NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 303
ASIEDU, ABUNDE, AND ADDO

Materials and Methods Table 2. Mathematical Expressions for the Substrate

KINETICS OF GROWTH AND PRODUCT FORMATION and Product Inhibition Factors
Considering the Monod Equation for cell growth and prod-
uct formationEquations 12three inhibition patterns were INHIBITION FACTOR MATHEMATICAL EXPRESSION
considered in modeling product inhibition. These include linear, Linear product (LP) (1 - K2 P)
sudden growth stop, and exponential as shown in Table 1. Four
Exponential product (EP) (e - K1 P)
inhibition patterns were considered in modeling the kinetics of
growth and product formation: Linear Substrate and Linear Linear substrate (LS) (1 - K2 S)
Product Inhibition Model (LS-LP), Linear Substrate and Ex-
Exponential substrate (ES) (e - K1 S)
ponential Product Inhibition Model (LS-EP), Exponential
Substrate and Linear Product Inhibition Model (ES-LP), and
Exponential Substrate and Exponential Product Inhibition
Model (ES-EP) inhibitions (Table 2).
lmax S
l S lS; P = 1 - Kisx Sexp - Kix P (Equation 5)
lS = max (Equation 1) Ksx + S
Ksx + S
where lmax is the maximum specific growth rate (h-1) and Ksx is

qmax S
the substrate saturation (Monod) constant for cell growth (g/ qS; P = 1 - Kisp S exp - Kip P (Equation 6)
Ksx + S
100 g).
qpmax S
qp S = (Equation 2) where Kisp is the substrate inhibition coefficient on product
Ksp + S formation
-
Kinetics with exponential substrate and linear product
where qpmax is the maximum rate of product formation (h-1) and inhibitions
Ksp is the substrate saturation (Monod) constant for product
formation (g/100 g). lmax S
lS; P = 1 - Kix Pexp - Kisx S (Equation 7)
Km + S
SUBSTRATE AND PRODUCT INHIBITION MODELS
By incorporating the effect of substrate and product inhibi-
qmax S
tions into the Monod equation and using the inhibition factors in qS; P = 1 - pPexp - Kisp S (Equation 8)
Km + S
Table 1, the following kinetic models were obtained:
-
Kinetics with linear substrate and linear product inhibitions -
Kinetics with exponential substrate and exponential product
inhibitions
lmax S
lS; P = 1 - Kix P1 - Kisx S (Equation 3)
Ksx + S lmax S
lS; P = exp - Kix P exp - Kisx S (Equation 9)
Ksx + S
where Kix is product inhibition coefficient on cell growth
and Kisx is substrate inhibition coefficient on cell growth.

qmax S


qmax S qS; P = exp - Kip P exp - Kisp S (Equation 10)
qS; P = 1 - Kip P 1 - Kisx S (Equation 4) Ksp + S
Ksx + S
Introducing the effect of product inhibition into the Monod
where Kip is product inhibition coefficient on product for- equation, and using the respective inhibition factors, yielded the
mation. following kinetic models:
-
Kinetics with linear substrate and exponential product
inhibitions
-
Kinetics with linear product inhibition (also known as the
Hinshelwood-Dagley model)11
Table 1. Mathematical Expressions for the Product lmax S
Inhibition Factors lS; P = 1 - Kix P (Equation 11)
Ksx + S
PRODUCT INHIBITION FACTOR MATHEMATICAL EXPRESSION
Linear (1 - K2 P)
qmax S
  qS; P = 1 - Kip P (Equation 12)
Sudden growth stop P
1 - Pmax Ksx + S
Exponential e - K1 P
-
Kinetics sudden growth stop product inhibition (also known
as Ghose-Tyagi model)12

304 INDUSTRIAL BIOTECHNOLOGY O C T O B E R 2 0 1 6

 MODELING AND SIMULATIONS OF INHIBITION PATTERNS

 
P lmax S dP
lS; P = 1 - (Equation 13) = qX (Equation 17)
Pmax Ksx + S dt

  dS 1 dX 1 dP
P qmax S =- - - Gs X - Ms X (Equation 18)
qS; P = 1 - (Equation 14) dt Yx dt Yp dt
Ppmax Ksx + S
where Gs is the yield coefficient of the cell based on sub-
where Ppmax is product concentration when cell growth strate utilization (g/g$h); Ms is the cell growth coefficient on
ceases (g/100 g) substrate (g/g$h); Yx is the yield coefficient of the cell based
-
Kinetics with exponential product inhibition (also known as on substrate utilization (g/g); and Yp is the yield coefficient
the Aiba and Shoda model)13 of the cell based on substrate utilization (g/g).
lmax S Using the batch kinetic models Equations 15 and 16, and
lS; P = exp - Kix P (Equation 15)
Ksx + S substituting l and q in Equations 1618 with each of their
product-inhibition expressions, the approximate representation

qmax S of the fermentation process in each inhibition scenario could be
qS; P = exp - Kip P (Equation 16) described by the following Equations 1921:
Ksx + S
-
Dynamics with linear substrate and linear product inhibi-
MATERIAL BALANCE AND DEVELOPMENT tions
OF DIFFERENTIAL EQUATIONS
The ordinary differential equations that describe cell growth, dX l S
product formation, and substrate utilization are presented in = 1 - Kix P1 - Kisx S max X (Equation 19)
dt Ksx + S
Equations 1618:
dX dP

qmax S
= lX (Equation 16A) = 1 - Kip P 1 - Kisp S X (Equation 20)
dt dt Ksp + S

Table 3. Model Parameters for Alcoholic Fermentation of Cassava Extract

INHIBITION
MODEL MONOD LINEAR S LINEAR S EXPONENTIAL S EXPONENTIAL S
PARAMETERS (NO INHIBITION) LINEAR P EXPONENTIAL P LINEAR P EXPONENTIAL P
lmax 0.0100 0.0100 3.0000 2.4342 2.9996

qpmax 8.4851 14.5352 7.1381 7.1395 183.4561

Kisx 0.3131 0.0001 0.0001 0.0001

Kisp 0.0010 0.0021 0.0010 0.2002

Kix 0.9539 0.5411 0.1264 0.5453

Kip 0.1475 0.8968 0.1450 1.2437

Ksx 249.9991 1.0090 67.607 249.9969 66.2282

Ksp 199.9997 198.8989 9.5495 199.9999 11.9572

Yx 0.1002 2.9782 3.0000 3.0000 2.9999

Yp 0.5464 0.6497 0.6131 0.6389 0.6131

Gs 0.0010 0.0017 0.0001 0.0011 0.0001

Ms 0.0100 0.0017 0.0001 0.0011 0.0001

Model error 94.0344 7.3934 2.8012 7.5115 2.7975

M A R Y A N N L I E B E R T , I N C .  VOL. 12 NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 305

Fig. 1. Experimental results and model fitting, LS-EP.

Fig. 2. Experimental results and model fitting, LS-LP.

306 INDUSTRIAL BIOTECHNOLOGY O C T O B E R 2 0 1 6

Fig. 3. Experimental results and model fitting, ES-EP.

Fig. 4. Experimental results and model fitting, ES-LP.

M A R Y A N N L I E B E R T , I N C .  VOL. 12 NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 307

ASIEDU, ABUNDE, AND ADDO

Table 4. Model Statistical Validity with Kinetics Exponential Substrate and Exponential Product Inhibition,
with Two Sample F-Tests for Variance
BIOMASS PRODUCT SUBSTRATE
EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL
(XOBS) MODEL (XPRED) (POBS) MODEL (PPRED) (SOBS) MODEL (SPRED)
Mean 2.093 2.113 5.881 5.886 5.275 5.280

Standard Error 0.245 0.228 0.403 0.397 0.722 0.727

Standard Deviation 0.979 0.913 1.613 1.588 2.889 2.909

Observations 16 16 16 16 16 16

Confidence Interval 0.990 0.990 0.990

f 0.8695 0.9682 1.0143

Pr(F < f) Two-tailed 0.7900 0.9509 0.9785

dS 1 dX 1 dP -
Dynamics with exponential substrate and linear product
=- - - G s X - Ms X (Equation 21) inhibition
dt Yx dt Yp dt

Dynamics with linear substrate and exponential product dX l S

-
= 1 - Kix Pexp - Kisx S max X (Equation 25)
inhibition dt Ksx + S

dX l S dP l S
= 1 - Kisx Sexp - Kix P max X (Equation 22) = 1 - Kix Pexp - Kisx S max X (Equation 26)
dt Ksx + S dt Ksp + S

dP

qmax S
= 1 - Kisp S exp - Kip P X (Equation 23) dS 1 dX 1 dP
dt Ksp + S =- - - G s X - Ms X (Equation 27)
dt Yx dt Yp dt

dS 1 dX 1 dP Dynamics with exponential substrate and exponential

=- - - G s X - Ms X (Equation 24) -

dt Yx dt Yp dt product inhibitions

Table 5. Model Statistical Validity with Kinetics Linear Substrate and Exponential Product Inhibition
with Two Sample F-Tests for Variance
BIOMASS PRODUCT SUBSTRATE
EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL
(XOBS) MODEL (XPRED) (POBS) MODEL (PPRED) (SOBS) MODEL (SPRED)
Mean 2.093 2.112 5.881 5.886 5.275 5.280

Standard Error 0.245 0.228 0.403 0.397 0.722 0.727

Standard Deviation 0.979 0.914 1.613 1.588 2.889 2.910

Observations 16 16 16 16 16 16

Confidence Interval 0.990 0.990 0.990

f 0.8719 0.9682 1.0144

Pr(F < f) Two-tailed 0.7941 0.9508 0.9783

308 INDUSTRIAL BIOTECHNOLOGY O C T O B E R 2 0 1 6

 MODELING AND SIMULATIONS OF INHIBITION PATTERNS

Fig. 5. Product dynamics versus substrate utilization and time, ES-EP inhibition model.

dX l S -
Dynamics with linear product inhibitions
= exp - Kix P max exp - Kisx SX (Equation 28)
dt Ksx + S
dX l S
= 1 - Kix P max X (Equation 31)
dt Ksx + S
dP
qmax S

= exp - Kip P exp - Kisp S X (Equation 29)
dt Ksp + S dP
qmax S
= 1 - Kip P X (Equation 32)
dt Ksp + S
dS 1 dX 1 dP
=- - - G s X - Ms X (Equation 30)
dt Yx dt Yp dt dS 1 dX 1 dP
=- - - Gs X - Ms X (Equation 33)
dt Yx dt Yp dt
PRODUCT INHIBITION MODELS
With the help of batch kinetic models stated above, and -
Dynamics with sudden growth stop product inhibitions
substituting l and q with each of their product-inhibition ex-
pressions, the approximate representation of the fermentation  
process in each inhibition model could be described by the dX P lmax S
= 1- X (Equation 34)
following differential equations: dt Pmax Ksx + S

Fig. 6. Dynamics of cell growth versus substrate utilization and time, ES-EP inhibition model.

M A R Y A N N L I E B E R T , I N C .  VOL. 12 NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 309

ASIEDU, ABUNDE, AND ADDO

 
dP P qmax S STATA (StataCorp, College Station, TX) at a confidence in-
= 1- X (Equation 35) terval of 99%, to find out the confidence level for the developed
dt Ppmax Ksp + S
mathematical model.
The three-dimensional profiles using the proximal interpolant
dS 1 dX 1 dP method, implemented using the Matlab curve fitting tool, are
=- - - G s X - Ms X (Equation 36)
dt Yx dt Yp dt shown in Figs. 56). They present the dynamics of product in-
terpolated with substrate and time.
-
Dynamics with exponential product inhibitions
RESULTS OF MODEL PARAMETERS AND SIMULATIONS
dX l S
= exp - Kix P max X (Equation 37) Table 3 presents the parameters for the four different models
dt Ksx + S used to describe the dynamics of fermentation, and Figs. 14
show the fitting of the models with respect to the experimental
dP
qmax S data. The results were unexpected. First both substrate and
= exp - Kip P X (Equation 38) product inhibition exists in the alcoholic fermentation of cassava
dt Ksp + S
extracts, with the LS-EP and ES-EP models describing the in-
hibition dynamics with relatively high accuracy (having lower
dS 1 dX 1 dP model errors). Second, the results confirmed our previous ob-
=- - - G s X - Ms X (Equation 39) servations that modeling only product inhibition but did not
dt Yx dt Yp dt
describe with high accuracy the alcoholic fermentation of cas-
sava extracts, suggesting the presence of substrate inhibition.
The dynamics described by conventional Monod equationin
Results and Discussion the case of no inhibitionshowed a very high error, confirming
PARAMETER ESTIMATION AND MODEL VALIDATION the presence of inhibitions.
The identification of model parameters for the different sys- Another important parameter in fermentation kinetics is the
tems of equations was made with Matlab (Mathworks, Natick, specific rate of ethanol accumulation in the medium. The pa-
MA), and the ode45 solver was used to simulate the differential rameter estimation results indicated high maximal rate of
equations. This was done by minimizing the overall sum of product formation183.4561 with the ES-EP model com-
squared error using Equation 40 between the model simulation pared to 7.1381 for LS-EP model. This suggests that even
and experimental data points of the process variables (Biomass though both models showed similar accuracy in describing
(X), Substrate (S) and Product (P)). fermentation dynamics, designing a control policy with the ES-
EP model will result in a high process productivity in terms of
e = min +[X(k1 ; k2 ; . . . , kn - X e )2 ethanol accumulation. It should be noted that the product in-
hibition had a higher effect on the fermentation process than
+ S(k1 ; k2 ; . . . , kn - Se )2 + P(k1 ; k2 ; . . . , kn - Pe )2 ] substrate inhibition; this is apparent due to the product inhi-
bition coefficients being very high compared to those for
The model parameters were estimated, and the capability of substrate inhibition.
the mathematical model to describe the fermentation process This is also explained by the fact that most of the cyanide in
was tested statistically using the Ftests. This was done using the cassava, which could result in substrate toxicity, is removed

Fig. 7. Dynamics of cell growth versus substrate utilization and cell growth, ES-EP inhibition model.

310 INDUSTRIAL BIOTECHNOLOGY O C T O B E R 2 0 1 6

 MODELING AND SIMULATIONS OF INHIBITION PATTERNS

Table 6. Model Parameters for Fermentation Using Cassava Extracts

INHIBITION
MODEL PARAMETERS NO INHIBITION LINEAR SGS EXPONENTIAL
lmax 0.0100 0.3326 2.9996 4.9991

qpmax 8.4851 51.3560 5.8755 2.6067

Pxmax 6.8884

Ppmax 6.7637

Kix 0.1466 0.8290

Kip 0.1013 0.6347

Ksx 249.9991 13.5297 231.0673 52.4603

Ksp 199.9997 249.9655 199.9957 25.2593

Yx 0.1002 19.9983 1.0000 0.4297

Yp 0.5464 0.7589 1.1878 6.9998

Gs 0.0010 0.0010 0.0010 0.0010

Ms 0.0100 0.0100 0.0100 0.0100

Model error 94.0344 4.8107 7.8655 5.5755

during upstream processing. The high product inhibition is in
accordance with the work by Ingledew, which showed that
substrate and product were among the environmental stresses
to S. cerevisiae during alcoholic fermentation.10 Table 4 and
Table 5 present the model statistical validity using two sample
F-tests for variance. The results show that at a 99% confidence
interval, LS-EP and the ES-EP model predictions showed no
significant difference with the experimental data. In the early
stages of the fermentation process, product formation occurs
quickly. This decreases in intensity as substrate concentration
decreases in the bioreactor. This can be attributed to the high
sugar concentrations the cells encounter after the hydrolysis
process. This exerts osmotic stress on the cells, resulting in
transient instabilities in the reactor.9,10,14 This stress decreases
along with the sugar concentration, up to a point where the
systems achieves a stable steady state and the rate of product
formation becomes constant. The interpolation results for cell
growth were similar (Figs. 57).
PRODUCT INHIBITION RESULTS
The product inhibition and model statistical validation results
are shown in Tables 68 and Figures 811.
Understanding how the substrate and product vary in the
fermenter as the cells grow is also important in a fermentation
process. The linear and exponential inhibition models were

Table 7. Model Statistical Validity with Kinetics Linear Substrate and Linear Product Inhibitions, Two Sample F-test for Variance
BIOMASS PRODUCT SUBSTRATE
EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL
(XOBS) MODEL (XPRED) (POBS) MODEL (PPRED) (SOBS) MODEL (SPRED)
Mean 2.296 2.255 5.881 5.855 5.291 5.299

Standard error 0.261 0.183 0.403 0.416 0.788 0.816

Standard deviation 1.044 0.733 1.613 1.666 3.151 3.266

Observations 16 16 16 16 16 16

Confidence interval 0.990 0.990 0.990

F 0.4934 1.0665 1.0742

Pr(F < f) two-tailed 0.1832 0.9024 0.8915

M A R Y A N N L I E B E R T , I N C .  VOL. 12 NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 311

ASIEDU, ABUNDE, AND ADDO

Table 8. Model Statistical Validity with Kinetics Exponential Substrate and Linear Product Inhibitions,
Two Sample F-Test for Variance
BIOMASS PRODUCT SUBSTRATE
EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL
(XOBS) MODEL (XPRED) (POBS) MODEL (PPRED) (SOBS) MODEL (SPRED)
Mean 2.296 2.180 5.880 5.910 5.291 5.308

Standard error 0.261 0.227 0.403 0.375 0.788 0.826

Standard deviation 1.044 0.909 1.613 1.501 3.151 3.306

Observations 16 16 16 16 16 16

Confidence interval 0.990 0.990 0.990

F 0.7572 0.8654 1.1005

Pr(F < f) two-tailed 0.5969 0.7831 0.853

both used to simulate these variations. Figures 12 and 13
showed a decrease in substrate concentration and an increase
in product concentration with cell growth. However, with the
linear model, a curve-like behavior was observed at the edges
of the plot, being more intense with the product profile. There
are two theories explain this behavior: ethanol inhibition or
stationary/decline phases in batch growth kinetics. In the
case of ethanol inhibition, the linear model has a relatively
high maximal rate of product formation, and ethanol rapidly
accumulates to inhibitory levels, leading to disruption of
the cell membranes and thus nonlinearity in the substrate-
consumption and product-formation profiles.59,14 Regard-
ing the stationary or decline theory in batch growth kinetics,
the curve-like behavior observed in the linear model can be

Fig. 8. Experimental results and model fitting, no inhibition (Monod-Model).

312 INDUSTRIAL BIOTECHNOLOGY O C T O B E R 2 0 1 6

Fig. 9. Experimental results and model fitting, linear inhibition model.

Fig. 10. Experimental results and model fitting, Sudden growth stop inhibition model.

M A R Y A N N L I E B E R T , I N C .  VOL. 12 NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 313

Fig. 11. Experimental results and model fitting, exponential inhibition model.

Fig. 12. Simulation of substrate and product as a function of biomass during fermentation, linear inhibition kinetics.

314 INDUSTRIAL BIOTECHNOLOGY O C T O B E R 2 0 1 6

 MODELING AND SIMULATIONS OF INHIBITION PATTERNS

Fig. 13. Simulation of substrate and product as a function of biomass during fermentation using exponential inhibition kinetics.

attributed to the fact that cells get to the stationary and de-
cline phases. At these stages the cells no longer consume
substrate to produce ethanol as in the exponential phase,
hence the observed nonlinearity.
The smallest discrepancy between the model and experi-
mental data was for the linear model, which was somewhat
unsatisfactory compared to previous studies modeling ethanol
fermentation. This high model error could be attributed to the
existence of substrate inhibition that was not considered in the
modeling. Cassava contains intrinsic microbial inhibitors such
as linamarin, which is converted to hydrocyanic acid during
the upstream processes.4 Figure 14 presents product formation
with cell growth and substrate consumption and further confirms
this assertion. Nonlinearities are observed at high substrate

Fig. 14. Dynamics of product formation with cell growth and substrate consumption.

M A R Y A N N L I E B E R T , I N C .  VOL. 12 NO. 5  OCTOBER 2016 INDUSTRIAL BIOTECHNOLOGY 315

ASIEDU, ABUNDE, AND ADDO
concentration, which gradually decrease in intensity as substrate
concentration decreases.
Conclusions
A mathematical approach to study the substrate product in-
hibitor on the kinetics of alcoholic fermentation has been pre-
sented. The results show that there exist both product and
substrate inhibition in the alcoholic fermentation of cassava
extracts, with product inhibition having a higher effect. The
kinetics of the fermentation process was determined using
at a 99% confidence interval using two mathematical mod-
els. There are differences with respect to process productiv-
ity with the ES-EP model showing to result in a higher ethanol
yield if used to design a control policy. 3D simulations showed
that there exist transient instabilities at the start of fermentation
process attributed to high sugar concentrations encountered
immediately after hydrolysis which exert osmotic stress on yeast
cells resulting in transient instabilities in the bioreactor.
The modeling of product inhibition in the alcoholic fermen-
tation of cassava extracts has been presented. The results show
that ethanol inhibition can be described with a 99% confidence
interval as being either a linear or exponential inhibitory impact
on ethanol concentration. The linear model resulted in a very
high maximal rate of ethanol accumulation, hence rapidly ap-
proaching inhibitory levels. We recommend developed models
should be used to determined optimal controls that minimize the
effects of such inhibitions. We also recommend substrate inhi-
bition to be modeled on the kinetics of cell growth and product
formation in an attempt to further decrease the discrepancy
between model and experimental data.
The obtained models described with high accuracy with 99%
confidence interval. It is also recommended that an optimal
control problem using the developed models should be set up
and solved to determine the fermentation conditions that mini-
mize the effect of such inhibitions during the fermentation of
cassava extracts.
Acknowledgments
Our team expresses sincere thanks to the Intra-ACP under the
project Strengthening African Higher Education through Aca-
demic Mobility (STREAM). We are also thankful to Breweries
Ghana for supplying us with industrial data to validate our model.
Author Disclosure Statement
No competing financial interests exist.
316 INDUSTRIAL BIOTECHNOLOGY OCTOBER 2016
REFERENCES
1. Olivier B. Mass balance modelling of bioprocesses. (2001) Available at: www
.sop.inria.fr/comore/ICTP_Modelling_Bernard.pdf (Last accessed August 2016).
2. Tonukari NJ. Cassava and the future of starch. African J Biotechnol 2004;7(1).
3. Alford JS. Bioprocess control: Advances and challenges. Computers Chem Eng
2006;30:14641475.
4. Philbrick DJ, Hill DC, Alexander JC. Physiological and biochemical changes
associated with linamarin and administration to rats. Toxicol Asql Pharmacol
1977;42:539.
5. Chen CI, McDonald KA. Oscillatory behavior of Saccharomyces cerevisiae in
continuous culture: I. Effects of pH and nitrogen levels. Biotechnol Bioeng
1990;36:1927.
6. Chen CI, McDonald KA. Oscillatory behavior of Saccharomyces cerevisiae in
continuous culture: II. Analysis of cell synchronization and metabolism.
Biotechnol Bioeng 1990;36:2838.
7. Beuse MRB, Kopmann AHD, Thoma M. Effect of dilution rate on the mode of
oscillation in continuous culture of Saccharomyces cerevisiae. J Biotechnol
1998;61:1531.
8. Beuse MRB, Komann AHD, Thoma M. Oxygen, pH value, and carbon source
induced changes of the mode of oscillations in Saccharomyces cerevisiae.
Biotechnol Bioeng 1999;63:410417.
9. Fengwu B. 2007. Process oscillations in continuous ethanol fermentation with
Saccharomyces cerevisiae. (Doctoral thesis). Ontario, Canada: University of Waterloo.
10. Russell AD. Similarities and differences in the responses of microorganisms to
biocides. J Antimicrob Chemother 2003;52:750763.
11. Hinshelwood CN. The chemical kinetics of the bacterial cell. London, UK:
Clarendon Press Oxford, 1946.
12. Ghose TK, Tyagi RD. Rapid ethanol fermentation of cellulose hydrolysate. II.
Product and substrate inhibition and optimization of fermentor design.
Biotechnol Bioeng 1979;21:14011420.
13. Aiba S, Shoda M, Nagatani M. Kinetics of product inhibition in alcohol
fermentation. Biotechnol Bioeng 1968;10:845864.
14. Sutton KB. (2011) A Novozymes short report: Fermentation inhibitors. Available
at: http://bioenergy.novozymes.com/Documents/Ferm_SR_Inhibitors.pdf (Last
accessed August 2016).
Address correspondence to:
Nana Y. Asiedu, PhD
Department of Chemical Engineering
College of Engineering
Kwame Nkrumah University of Science
and Technology (KNUST)
Kumasi
Ghana
Phone: +23 3 54236 4021
E-mail: nasiedusoe@yahoo.co.uk