Jour of Adv Research in Dynamical & Control Systems, Vol.

11, Special Issue-12, 2019

Modified Kinetic Model For Ethanol

Fermentation From Oil Palm Trunk Sap
Sharmin Sultana1, Norazaliza Mohd Jamil1*, Abu Yousuf2, Norhazimah Abdul Halim3
1Faculty of Industrial Sciences & Technology, Universiti Malaysia Pahang, Malaysia
2Faculty of Engineering Technology, Universiti Malaysia Pahang, Malaysia
3Department of Science and Mathematics, Universiti Tun Hussein Onn, Malaysia

Abstract. This paper extended and improved the current mathematical model for the batch fermentation
process. By varying the initial cell concentration, the model predicted the profile of cell growth and ethanol
production throughout the fermentation process. The kinetic models take into account the following factors
which are substrate limitation, substrate inhibition, product inhibition, and cell death simultaneously for the
production of ethanol from the OPT sap. The mathematical model was formulated using a set of ordinary
differential equations to describe the profiles of sugar, cell and ethanol during the fermentation process. The set
of equations were solved numerically by using the 4th order Runge-Kutta method. The result showed that the
rate of sugar utilization and ethanol production are depended on the initial cell concentration. For low initial cell
concentration, the conversion rate was increased gradually. On the other hand, for high initial cell concentration,
sugar conversion to ethanol was augmented sharply and depleted after a short duration due to the access of the
ethanol, which might inhibit the cell growth. The combined consideration of the substrate limitation and
inhibition, growth and non-growth associated product formation, product inhibition and cell death rate increased
the accuracy of the model by means of rRMSE. The proposed model has better predictive capabilities. This
approach has increased our understanding of the theory behind the OPT sap fermentation.
Keywords: Kinetic model, initial cell concentration, fermentation, bio-ethanol, oil palm trunk sap

INTRODUCTION
Ethanol is an alternative fuel which supports a sustainable economy by reducing the use of petroleum, carbon
dioxide (CO2) accumulation particulate matter and nitrous/nitric oxide (NOx) emission from combustion
(Srimachai et al., 2015). However, the chemical properties of biomass material and the reaction kinetics for the
degradation of biomass is not well understood (Jamil & Wang, 2016). For this reason, to understand, to oper ate,
to optimize and to control the ethanol fermentation process, complete knowledge of dynamic behaviour is
required (Oliveira et al., 2016).
The fermentation step is an essential part of the biomass to the bio-ethanol conversion process. An appropriate
kinetic model of ethanol fermentation would be a powerful instrument for increasing fermentation efficiency
and process optimization (Liu & Li, 2014). The amount of the initial cell concentration is a crucial well-known
process parameter in microbial fermentation (Papagianni & MooYoung, 2002; Rao et al., 2004). The impact of
the initial cell concentration in ethanol formation has been very little studied and not in an easily predictable
way. (Eker & Sarp, 2017) studied the analytical effects of initial sugar and biomass concentrations of H2 gas
production. Some of the researchers showed that a higher level of inoculum resulted in a higher yield of ethanol
(Carrau et al., 2010; Mateo et al., 2001). However, different overall behaviour was acknowledged to cell strain
(Carrau et al., 2008) but the kinetic behaviour of different ratios of initial cell concentrations was not taken into
account. Mathematical modelling is a cognitive tool used to describe the cellular response of microbial cells to
changes in nutrient inputs and other environmental factors (Tijani et al., 2018). Currently, there exists no
accurate model that simultaneously incorporates the essential factors such as substrate limitation and inhibition,
growth and non-growth associated product formation, product inhibition and cell death with the combined effect
of culture parameter temperature and initial cell concentration.
The fundamental model for microbial growth activities was proposed by Monod (Monod, 1949) and the kinetic
model accounts only the factor of substrate limitation through an equation called as Monods equation. On the
other hand, the models of Hinshelwood (1946), Hoppe and Hansford (1982) account only for ethanol inhibition.
Other than those factors, Aiba and Shoda (1969) includes also product inhibition in their model. Ghaly and
ElTaweel (1994) were concerned about substrate limitation, substrate inhibition, product inhibition from cheese
whey by the yeast Candida pseudo-tropicalis. However, all of these models were not concerned about the effect
of some culture parameters such as temperature, pH and inoculum size.
Some of the recent kinetic models in oenology (the study of wine) have been proposed by (Jin et al., 2012);
(Kelkar & Dolan, 2012); (Mohamad et al., 2016) and (Oliveira et al., 2016) which are unstructured kinetic
models. Specifically, Kelkar and Dolan (2012) studied the mutual effect of primary nitrogen concentration and
temperature on fermentation and concluded that yeast cell growth is controlled by nitrogen and sugar
concentration. Jin et al. (2012) applied the Hinshelwood model to explained the roles of preliminary reducing
*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]
Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
280
 Jour of Adv Research in Dynamical & Control Systems, Vol. 11, Special Issue-12, 2019

sugar content on the kinetic behaviour of immobilized Saccharomyces cerevisiae, where the medium was sweet
sorghum stalk juice. They observed that there was a major inhibition on the maximum specific growth rate.
However, no significant effect was found on the maximum specific ethanol production rate with the increase of
initial reducing sugar. Mohamad et al. (2016) include limiting effect of xylose (substrate) and oxygen on cell
growth and inhibition effect of oxygen on the product. Oliveira et al. (2016) accounted for the inhibition effect
of substrate and product on cell growth but did not consider cell death in their proposed model. The
Phisalaphong model (2006) was modified from the Monod model considering the essential factors of the cane
molasses fermentation. By this model, they observed the inhibition effect of substrate and product, including
cell death rate. However, they did not show any influential impact of cell death rate and initial cell
concentration.
With the proposed model, a significant effect was found for the different ratios of initial cell concentrations of
higher ethanol production. In this study, OPT sap was used as the raw material for ethanol production by batch
fermentation at different temperature and initial cell concentrations. The initial cell concentration was varied
between 1 and 3g/L-1 as the low cell concentration and between 10 and 20 g/L-1 as the high cell concentration,
while sugar concentration was kept constant at 69 g/L-1 to determine the effects of initial cell concentration on
ethanol production at the 25◦C, 30◦C, and 35◦C temperature.
This study aims to propose a modified kinetic model for ethanol fermentation from oil palm trunk sap based on
the existing model from Phisalaphong and Luedeking-Piret (LP) dynamic model. Also, this paper will
investigate the effect of initial cell concentration of S. cerevisiae on the production of ethanol with the extended
and improved current mathematical model.

MATERIALS AND METHODS

The experimental data of the fermentation process for the bioethanol production from OPT sap by
Saccharomyces cerevisiae was used in this work. We relied on available data from (Halim, 2016).
Saccharomyces cerevisiae Kyokai no.7 was used as the microorganism for the production of bioethanol from
OPT sap. The strain was collected from the Faculty of Engineering Technology, Universiti Malaysia Pahang,
Malaysia.
OPT sap was collected in a big container and mixed well before divided into 5 L bottle-shaped reactor. It was
kept under -20◦C for storage. OPT sap was mixed well in a container before storage since different part of OPT
gain different compositions of sugar. OPT sap media was filtered with 9.0 m filter prior to use. The composition
of OPT sap at a different part of trunks was also checked. In the present study, the fermentation was managed at
three different temperatures at 25, 30, and 35 ◦C in static condition. The initial inoculum size 1.25 g/L and initial
sugar 86.63 g/L were used respectively for all temperatures. Shake flask was purged with nitrogen to remove
oxygen to create an anaerobic condition. Samples were withdrawn periodically at predetermined time intervals
for analysis.
During the batch fermentation, 2 ml sample was taken at each time interval using a disposable syringe (5ml) and
centrifuged at 10000 rpm/ 11963 g force for 5 mins to separate the supernatant from the pellet using
microcentrifuge (Biofuge Pico, Heraeus). After the supernatant was removed, and the pellet was dissolved in
distilled water for rinsing, cell growth was monitored by measuring the optical density of the dissolved pellet at
600 nm (OD 600) using a UV-VIS spectrophotometer (Hitachi U-1800 Spectrophotometer). For cell, dry weight
(DCW) determination, the pellets were dried in a 60◦C oven until they reached constant weight. All DCW
determination was repeated for consistency. To determine ethanol concentration, GC (Gas chromatography)
technique was employed where a flame ionization detector was used with an HPINNOWax Polyethylene Glycol
(30 m x 250 m x 0.25 m nominal) column. The oven temperature was maintained for the Initial temperature,
maximum temperature, and temperature rate at 50 ◦C, 170◦C and 20◦C/min, respectively. Helium was used as
carrier gas at a flow rate of 45 ml/min. The n-propanol was used for washing and dilution. One part of the
sample was mixed with 9 part of n-propanol and was filtered through a 0.2-micron nylon membrane (Fisher
ScientificTM).

MATHEMATICAL MODELLING
Kinetic model
In order to effectively analyze the kinetics of the fermentation process, we described the ethanol production
route and the phenomenon to express in terms of mathematical equations. This proposed model is the
modification from Phisalaphong model (Phisalaphong et al., 2006) and Leudeking-Piret relationship (Luedeking
& Piret, 2000) for the solution of substrate consumption and product formation. The model was modified by the
following assumption. i) Limitation of cell growth due to substrate deficiency; ii) Cell growth inhibition by
ethanol and substrate; iii) Growth and non-growth associated product formation; iv) Existence of cell death or
inactivation; v) Temperature dependence on cell growth.

*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]

Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
281
 Modified kinetic model for ethanol fermentation from oil palm trunk sap

The modified kinetic model for the batch fermentation from oil palm trunk sap for cell concentration, X,
substrate concentration, S and ethanol concentration, P is as follows.
(1)
(2)

(3)
where a is growth associated specific productivity coefficient, b is the non-growth associated specific
productivity coefficient, c is the coefficient of cell maintenance, YXS represents the yield coefficient for the cell
on the substrate. According to our model, the ethanol production rate depends on instantaneous biomass (cell)
concentration, X.
A mathematical model of ethanol fermentation by S. cerevisiae Kyokai no.7 was customized from the Monod
kinetics (Monod, 1949). Monod equation defines the relation between the growth rate and the substrate
concentrations. In our model, the specific growth rate of the microorganism was described by the modified
Monod equation that includes substrate inhibition and product inhibition as suggested from (Oliveira et al.,
2016) i.e.

(4)
where µ is the maximum specific growth rate, Ks is the substrate affinity coefficient, Ki is the inhibition
parameter for sugar, and Pm is the inhibition parameter for ethanol.
The ordinary differential equations of the presented models were solved numerically by the 4th order
RungeKutta (RK4) scheme implemented in MATLAB (version 8.4). In our analysis, the initial values for
(X,P,S) were taken from population distributions derived from the available experimental data.

Parameter Estimation
The estimation of fermentation parameters is a crucial part of the authentication and consequential use of a
mathematical model (Wang & Sheu, 2000). There are two elements involved in estimating model parameters
from data. First, we constructed an error function that measures the difference between a model with a specific
parameter and the data. Second, we need an optimization method that iteratively finds the value of the parameter
that minimizes the error.
In this approach, we can systematically vary the parameters so that we can get the minimum difference between
the solution of differential equations and the data. We started the model by adjusting the parameter values
manually to obtain a good fit to the experimental data. The best fit value was estimated by using the least square
method to minimize the objective function, as shown below:

(5)
where f(xi) is the calculated data and yi is the experimental data.
When measurement errors are independently normally distributed, the error function will be the logarithm of the
likelihood of the data (log(likelihood)). Minimizing the error function is equivalent to maximum likelihood
estimation of the parameters. We implemented a Runge-Kutta algorithm in Matlab and a Nelder-Mead simplex
method as the optimization routine for our unknown parameter estimation. We used fminsearch, which is
Matlabs builtin command using the Nelder-Mead algorithm.

RESULTS AND DISCUSSION

Model prediction and validation
To estimate the unknown parameters for the proposed model, equations (1), (2) and (3) are used and for the
initial conditions, the initial sugar and cell concentration are maintained at S(0) = 86.63g/L,X(0) = 1.25g/L.
Figure 1-3 explain the experimental results, as well as the modelled values of the fermentation profile at
different temperature with the same initial cell and sugar concentration of 1.25 g/L and 86.63 g/L respectively.
According to these figures, substrate concentration was reduced monotonically until its full reduction, whereas
cell and product concentration was increased monotonically before reaching the stationary phase. As illustrated
in those figures, the predicted models finely fitted to the experimental data for the cell, ethanol and substrate
concentration at 25◦C and 30◦C from the beginning up to the stationary phase. But for the 35 ◦C, model fitting
was slightly deviated for ethanol due to the inconsistency of the cell growth at high temperature. To increase the
accuracy of the model, the parameter function should depend on temperature (Rivera et al., 2016).

*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]

Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
282
 Jour of Adv Research in Dynamical & Control Systems, Vol. 11, Special Issue-12, 2019

Figure 1. Experimental data and model predictions of batch fermentation at 25 ◦C temperature

Effect of initial cell concentration

For all the results presented in this section, the initial concentration of sugar was fixed to 69 g/L. The sensitivity
of the model to the initial cell concentration is considered. To understand the effect of cell concentration on the
fermentation behaviour, we investigated the simulated conversion rate with respect to time. Figure 4 shows the
concentration of ethanol for different ratios of cell: sugar, which is 1:69, 2:69, and 3:69. The ethanol production
was increased with the increase of cell concentration.
Figure 5 shows the concentration of ethanol for ratios of cell: sugar of 10:69, 15:69, and 20:69 at temperature
30◦C. At early times before 30 hours, initially, the concen- tration of ethanol is increasing rapidly but at longer
times (after 30 hours), the ethanol concentration is depleted. In the following discussion, for ratios of cell: sugar
of 10:69, 15:69, and 20:69 will be referred to as the high cell concentration case, and 1:69, 2:69, and 3:69 as the
low cell concentration case.

Figure 2. Experimental data and model predictions of batch fer- mentation at 30◦ C temperature

*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]

Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
283
 Modified kinetic model for ethanol fermentation from oil palm trunk sap

Figure 3. Experimental data and model predictions of batch fer- mentation at 35◦ C temperature

Figure 4. Sugar to ethanol conversion at different initial cell con- centration: 1, 2 and 3 g/L

Figure 5. Sugar to ethanol conversion for 30◦ C at different initial cell

Two distinct patterns of conversion can be identified. For low cell concentration, the conversion rate is almost
constant. In contrast, for high cell concentration, the conversion grows at a constant rate up to 30 hours as
indicated in Figure 5. Then, the conversion rate is depleted. In comparison to Figure 4, at the initial stage, the
conversion rate from sugar to ethanol for high cell concentration is higher than low cell concentration. Also, the
conversion initiates earlier at higher cell concentration. It happened due to the higher initial cell concentration
that can increase the rate of sugar utilization and ethanol formation (Matsushika & Sawayama, 2010). However,
since the ethanol production rate was faster, the cell growth was inhibited by ethanol and the conversion rate
sharply dropped after a short duration (within 30-60 hours). Cell inhibition slows the conversion rate by
*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]
Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
284
 Jour of Adv Research in Dynamical & Control Systems, Vol. 11, Special Issue-12, 2019

accessing the amount of ethanol. Thus, the rate of sugar consumption and ethanol production throughout
fermentation depend on the concentration of cells present in the inoculum.

Model performance
The parameter values were obtained by using a nonlinear regression technique that minimizes the sum-ofsquares
deviation between the model predictions and experimental data. Table 1 records the Relative Root Mean
Squared Error (rRMSE) between the model and the experimental data at different temperatures. The values are
the means with the corresponding confidence intervals (95%). Most of the rRMSE values of all profiles range
from 2.8% to 10% with very few exceptions. It implies that these models can finely describe all temperature
profiles of the ethanol fermentation.

Table 1. Accuracy of the proposed model considering (rRMSE)

Temperature 25◦C 30◦C 35◦C
Cell, X 4.9665 6.7714 4.5725
Sugar, S 4.4063 5.0231 2.8238
Ethanol, P 10.6951 7.5309 20.8075

This result has been predicted with the consideration of the cell death rate and Leudeking-Piret equation.
Therefore, it could be concluded that our extended model has a very significant influence on the mathematical
study of ethanol production from OPT sap through the fermentation process. Hence, to summarize, the
mathematical models were sufficiently reliable for the experimental data for all temperatures with few
deviations.

CONCLUSION
Though it is difficult to understand the step by step transformation of the actual phenomenon of the
fermentation, a valid mathematical model was modified for ethanol production. Our modified model is a
combination of Phisalaphong and LuedekingPiret, that provides an acceptable estimate of cell growth and
ethanol production. A Runge-Kutta algorithm was applied in Matlab and a Nelder-Mead simplex method as the
optimization routine for our unknown parameter estimation. The rRMSE values of all profiles were less than
15%, which indicates the model can finely describe all concentration profiles. From this study, the initial cell
concentration gives a significant effect on cell growth and ethanol production. In this extended model, the
incorporation of the most important factors (substrate limitation, substrate inhibition, ethanol inhibition, growth
and non-growth associated product formation and cell death) for cell growth, cell death rate and ethanol
production reflected approximately the real situation in the fermentation process. This model can be used as a
reference to further optimize the ethanol fermentation process.

ACKNOWLEDGEMENTS
The author gratefully acknowledged the financial supports received from FRGS (RDU 160116) (Ref:
FRGS/1/2016/STG06/UMP/02/3).

REFERENCES
[1] Aiba, S., Shoda, M. 1969. Reassessment of product inhibition in alcohol fermentation. Journal of
Fermentation Technology, 47(12), 790.
[2] Carrau, F., Medina, K., Faria, L., Boido, E., Dellacassa, E. 2010. Effect of Saccharomyces cerevisiae
inoculum size on wine fermentation aroma compounds and its relation with assimilable nitrogen content.
International journal of food microbiology, 143(1-2), 81-85.
[3] Carrau, F.M., Medina, K., Farina, L., Boido, E., Henschke, P.A., Dellacassa, E. 2008. Production of
fermentation aroma compounds by Saccharomyces cerevisiae wine yeasts: effects of yeast assimilable
nitrogen on two model strains. FEMS yeast research, 8(7), 1196-1207.
[4] Eker, S., Sarp, M. 2017. Hydrogen gas production from waste paper by dark fermentation: Effects of initial
sub-strate and biomass concentrations. International Journal of Hydrogen Energy, 42(4), 2562-2568.
[5] Ghaly, A., El-Taweel, A. 1994. Kinetics of batch produc-tion of ethanol from cheese whey. Biomass and
Bioenergy, 6(6), 465-478.
[6] Halim, N.B.A. 2016. Optimization of bioethanol production from oil palm trunk sap, Universiti Malaysia
Pahang.
[7] Hinshelwood, C.N. 1946. The chemical kinetics of the bacterial cell. Oxford: The Clarendon Press, UK.
[8] Hoppe, G.K., Hansford, G.S. 1982. Ethanol inhibition of continuous anaerobic yeast growth. Biotechnology
Letters, 4(1), 39-44.
[9] Jamil, N.M., Wang, Q. 2016. The Nondimensionalization of Equations Describing Enzymatic Cellulose
Hydrolysis. World Applied Sciences Journal, 34(2), 158-163.

*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]

Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
285
 Modified kinetic model for ethanol fermentation from oil palm trunk sap

[10] Jin, H., Liu, R., He, Y. 2012. Kinetics of batch fermentations for ethanol production with immobilized
Saccharomyces cerevisiae growing on sweet sorghum stalk juice. Procedia Environmental Sciences, 12, 137-
145.
[11] Kelkar, S., Dolan, K. 2012. Modeling the effects of initial nitrogen content and temperature on
fermentation kinetics of hard cider. Journal of Food Engineering, 109(3), 588-596.
[12] Liu, Z., Li, X. 2014. The kinetics of ethanol fermentation based on adsorption processes. Kemija u
Industriji, 63(7-
[13] 8).
[14] Luedeking, R., Piret, E.L. 2000. A kinetic study of the lac-tic acid fermentation. Batch process at
controlled pH. Bio-technology and bioengineering, 67(6), 636-644.
[15] Mateo, J., Jimnez, M., Pastor, A., Huerta, T. 2001. Yeast starter cultures affecting wine fermentation
and volatiles. Food Research International, 34(4), 307-314.
[16] Matsushika, A., Sawayama, S. 2010. Effect of initial cell concentration on ethanol production by
flocculent Saccha-romyces cerevisiae with xylose-fermenting ability. Ap-plied biochemistry and
biotechnology, 162(7), 19521960.
[17] Mohamad, N.L., Kamal, S.M.M., Mokhtar, M.N., Husain, S.A., Abdullah, N. 2016. Dynamic
mathematical model-ling of reaction kinetics for xylitol fermentation using Candida tropicalis. Biochemical
Engineering Journal, 111, 10-17.
[18] Monod, J. 1949. The growth of bacterial cultures. Annual Reviews in Microbiology, 3(1), 371-394.
[19] Oliveira, S.C., Oliveira, R.C., Tacin, M.V., Gatts, E.A. 2016. Kinetic modeling and optimization of a
batch ethanol fermentation process. Journal of Bioprocessing & Biotechniques, 6, 266.
[20] Papagianni, M., Moo-Young, M. 2002. Protease secretion in glucoamylase producer Aspergillus niger
cultures: fungal morphology and inoculum effects. Process Biochemistry, 37(11), 1271-1278.
[21] Phisalaphong, M., Srirattana, N., Tanthapanichakoon, W. 2006. Mathematical modeling to investigate
temperature effect on kinetic parameters of ethanol fermentation. Bio-chemical Engineering Journal, 28(1),
36-43.
[22] Rao, R.S., Prakasham, R., Prasad, K.K., Rajesham, S., Sarma, P., Rao, L.V. 2004. Xylitol production
by Candida sp.: parameter optimization using Taguchi approach. Pro-cess Biochemistry, 39(8), 951-956.
[23] Rivera, E.C., Yamakawa, C.K., Saad, M.B., Atala, D.I., Ambrosio, W.B., Bonomi, A., Junior, J.N.,
Rossell, C.E. 2016. Effect of temperature on sugarcane ethanol fermentation: Kinetic modeling and
validation under very-highgravity fermentation conditions. Biochemical Engineering Journal.
[24] Srimachai, T., Nuithitikul, K., Sompong, O., Kongjan, P., Panpong, K. 2015. Optimization and kinetic
modeling of ethanol production from oil palm frond juice in batch fermentation. Energy Procedia, 79, 111-
118.
[25] Tijani, H., Yuzir, A., Dagang, W.R.Z.W., Zamyadi, A., Abdullah, N. 2018. Multi-parametric modelling
and kinet-ic sensitivity of microalgal cells. Algal research, 32, 259-269.
[26] Wang, F.-S., Sheu, J.-W. 2000. Multiobjective parameter estimation problems of fermentation
processes using a high ethanol tolerance yeast. Chemical Engineering Science, 55(18), 3685-3695.
[27] Richa tyagi, gaurav sharma, nakuleshwar dut jasuja, ekta menghani (2016) indian medicinal plants as
an effective antimicrobial agent. Journal of Critical Reviews, 3 (2), 69-71.

*Corresponding Author: Norazaliza Mohd Jamil, Email: [email protected]

Article History: Received: July 02, 2019, Accepted: Sep 27, 2019
286