
Research Article


Fermentation Process Optimization

Bibhu Prasad Panda ,
Mohd. Ali
and
Saleem Javed


ABSTRACT

Many optimization techniques are available for optimization of fermentation medium and fermentation process conditions such as borrowing, component swapping, biological mimicry, onefactoratatime, factorial design, Plackett and Burman design, central composite design, response surface methodology, evolutionary operation, evolutionary operation factorial design, artificial neural network, fuzzy logic and genetic algorithms. Each optimization technique has its own advantages and disadvantages. 




INTRODUCTION For an industrial fermentation process fermentation medium and fermentation process condition plays an critical role because they effect the formation, concentration and yield of a particular fermentation end product thus effecting the overall process economics therefore it is important to consider the optimization of fermentation medium and process conditions in order to maximize the profits from fermentation process (Schmidt, 2005). There are many challenges associated with optimization of fermentation process, it is laborious, expensive, open ended and time consuming process involving many experiments. In bioprocess industry it is often needs to conduct optimization experiments because new mutants and strains are continuously being introduced. In fermentation process optimization different combinations and sequence of process conditions and medium components are needs to be investigated to determine the growth condition that produces the biomass with the physiological state best constituted for product formation (Stanbury et al., 1997). The present review explores the different wellknown and newer optimization method applied in fermentation process. OPEN AND CLOSE ENDED SYSTEMS FOR PROCESS OPTIMIZATION
In closeended system, a fixed number and type of component parameters are
analyzed for optimization, this is the simplest strategy but many different
possible components/parameters which are not considered, could be beneficial
in the medium. In openended system any number and type of components/parameters
are analyzed for optimization of fermentation process. The advantage of this
system is that it makes no assumption of which components/parameters are best
for fermentation process. The ideal method would be to start with an openended
system, select the best components/parameters for optimization of fermentation
process then move to the closeended system (Kennedy and Krouse, 1999).
DIFFERENT METHODS FOR FERMENTATION PROCESS OPTIMIZATION
Borrowing
This is an openended system for process optimization. The medium components
and process conditions are obtained from the literatures and what other workers
were used to grow the same genus, species or strains are analyzed. The problem
with this method is that there are too many options for a given fermentation
process. Therefore short listing is necessary and advantage of this method is
that it is simple, easy and requires no mathematical skill (Kennedy and Krouse,
1999).
Component Replacing
This is an openended system for process optimization and only used to compare
the component of one type in a fermentation medium (Nandi and Mukherjee, 1988).
In this method, one of component of the medium was replaced by a new one at
same incorporation level. However this method does not consider the components
interactions. But this method can useful for screening different carbon, nitrogen
and other source for improving the medium utilization (Kennedy and Krouse, 1999;
Jatinder et al., 2006; Tavares et al., 2005). Screening of suitable
carbon source for mevastatin and citric acid production by solidstate fermentation
was carried out by component replacing techniques (Ahamad et al., 2006;
Kumar et al., 2003).
Biological Mimicry
Biological mimicry is a closeended system for fermentation process optimization.
This method is useful for optimization of various components of fermentation
media and based on concept that cell grow well in a medium that contains every
things it needs in right proportion (mass balance strategy). The medium is optimized
based on elemental composition of microorganisms and growth yield. The limitation
of this method is measuring elemental composition of microorganisms is expensive,
laborious and time consuming moreover it does not consider the component interaction
however this method gives an idea about different micro and macro elements level
require in the media for optimal growth of microorganisms (Kennedy and Krouse,
1999).
Onefactoratatime
Onefactoratatime is a closeended system for fermentation process optimization.
This method can be applied for optimization of medium components as well as
for process condition and it is based on the classical method of changing one
independent variable while fixing all other at a certain level (Ahamad et
al., 2006; Alexeeva et al., 2002; Patidar et al., 2005). This
strategy has the advantage that it is simple, easy and the individual effects
of medium components and process condition can be seen on graphs (Kar et
al., 1999; Kumar et al., 2003) but the limitations of this method
are interaction between the components are ignored, extremely time consuming,
expensive for large number of variable as it involves a relatively large number
of experiments. Because of its easy and convenience onefactoratatime method
has been the most popular method for improving the fermentation medium and process
condition.
Factorial Design
Factorial design is a closeended system for process optimization. In this
method, level of factors/parameters are independently varied, each factor at
two or more levels. This effects that can be attributed to the factors and their
interactions are assessed with maximum efficiency in factorial design more over
it allow for the estimation of the effects of each factor and interaction.
The optimization procedure is facilitated by construction of an equation that describes the experimental results as a function of the factor level. A polynomial equation can be constructed in the case of a factorial design where the coefficient in the equation are related to the effects and interactions of the factors. In a full factorial (complete factorial) design every combination of factor level was tested. Typical factors are microbial strain, medium components, temperature, humidity, initial pH and inoculum volume. The most commonly used full factorials in medium improvement experiments are two factorial designs (denoted by 2^{n} when there are n factors). These designs are the smallest capable of providing detailed information on factor interaction (i.e., antagonistic or synergistic effects) (Xie et al., 2003). A partial factorial design provides a compromise when the number of runs required in full factorials is impracticable. These are usually twolevel factorial design. Two level fractional factorial are denoted by 2^{nk}, where n is the number of factors and ½ k is the fraction of the complete factorial used. This notation gives an immediate idea of the number of runs required. For example, 2^{61} is a half fraction of complete factorial 2^{5} and requires 32 (i.e., 2^{5}) runs per replicate (Rajendhran et al., 2002). In most case factorial design were in combined with other different optimization techniques such as central composite design (Park et al., 2005) and evolutionary operation (Tunga et al., 1999) to optimize fermentation process.
Plackett and Burman’s Design
Plackett and Burman’ s design may be useful to find out the important
variable in a system this design is suitable when more then five independent
variables are to be investigated. Plackett and Burman’ s design are useful
to screen out important factor, which influence the fermentation process (Naveena
et al., 2005). Which are optimized by response surface methodology in
further studies (Sayyad et al., 2006; Singh and Satyanarayana, 2006).
This technique allows for evaluation of n variables by n+1 experiments. n+1
must be multiple of 4 e.g., 8, 12, 16, 24, etc. therefore the number of independent
variables which can be investigated by this method are 7, 11, 15, 19, 23, etc.
Any factors not assigned to a variable can be designated as a dummy variable.
The incorporation of dummy variable into an experiment makes it possible to
estimate the variance of effects (Plackett and Burman, 1946).
Central Composite Design
Central composite design (CCD) was introduced by Box And Wilson; CCDs are
formed from two level factorials by addition of just enough points to estimate
curvature and interaction effects. The design can be viewed as partial factorials
with factors at five levels. The number of runs in CCD increases exponentially
with number of factors. Optimization of media components for compaction production
in complex and chemically defined production medium using CCD has been reported
(Kennedy and Krouse, 1999).
CCD can be combined with response surface methodology, in which experiments were designed by CCD and thereafter optimized by response surface methodology (Chakravarti and Sahai, 2002; Dahiya et al., 2005).
Response Surface Methodology
Statistical experiment design is a powerful method for accumulating information
about a process rapidly and efficiently from a small number of experiments,
thereby minimizing experimental costs. Box and Wilson introduced Response Surface
Methodology (RSM). RSM seeks to identify and optimize significant factors with
the purpose of determining what levels of factors maximize the response (Sayyad
et al., 2006; Singh and Satyanarayana, 2006). RSM uses statistical experimental
design such as Central Composite Design (Chakravarti and Sahai, 2002; Dahiya
et al., 2005), BoxBehnken Design (Sayyad et al., 2006) etc.
in order to develop empirical models that relate a response and mathematically
describes the relationships existing between the independent and dependent variables
of the process under consideration.
The contours of a response surface optimization plot show lines of identical
response. Response means the results of an experiment carried out at particular
values of the variables being investigated. The axes are the contour plots are
the experimental variable and the area with in the axes is termed the response
surface. To construct a contour plot, the results (response) of a series of
experiments employing different combination of variable are inserted on the
surface of the plot at the points delineated by the experimental conditions,
points giving the same results (equal response) are joined together to make
a contour line (Kumar et al., 2004).
The purpose of response surface methodology was to obtain a predicted model and this model can be useful for optimizing the fermentation media formulation or for optimization of fermentation process condition, to carry out simulation with model equation and for better understanding the fermentation process.
Evolutionary Operation
Evolutionary operation employs factorial design sequentially to improve yield.
The changes made to variable from one cycle to the next are restricted and can
only be made when the estimated improvements are greater then the estimated
experimental error. Optimization of production of protease by Rhizopus oryzae
using Evolutionary operation has been reported (Banerjee and Bhattachaaryya,
1993).
Evolutionary Operation Factorial Design
The evolutionary operation (EVOP) factorial design methodology was a hybrid
of evolutionary operation and factorial design technique here, experiments are
designed based on factorial technique and results are analyzed by EVOP. This
methodology is considered to be a multi variable sequential search technique,
in which the effects of n variable factors are studied and response analyzed
statistically. The decisionmaking procedure is easy and clearcut it directs
the change of variable to wards the objective maximum or minimum values. Evolutionary
operation factorial design technique combines the advantage of factorial technique
for designing experiments with n parameters and that of evolutionary operation
methodology for systematic analysis of experimental results and facilitate the
selection of optimum condition or direct the change desired for individual parameters
for design of subsequent experiments. For a study of a five variable system,
the total number of new experiments to be conducted is 2^{5}, apart
from two control experiments (search level regions). The parameters for the
above experiments are arranged in both higher level (+) and lower level ()
compared to search level regions (0), the parameters and the total number of
experiments are represented in a [5 X (2^{5}+2)] matrix Which has been
divided in two blocks i.e., overall negative blocks and overall positive blocks.
All experiments were replicated for two cycles. The magnitude of effects, change
in mean effects, standard deviation and error limits (of average, of effects
and of change in mean effects), analyzed as per the decision making procedure
of Evop to arrive at the optimum. When the experimental results of the first
set did not satisfy the optimum conditions, a second set of experiments was
planed selecting the best condition of the first set as the new search level
for second set. This procedure was repeated till the optimum condition was obtained
(Tunga et al., 1999; Panda, 2001).
Optimization of protease enzyme production under solidstate fermentation by Rhizopus oryzae and optimization of gallic acid production under solidstate fermentation using evolutionary operation and factorial design technique has been reported (Tunga et al., 1999; Kar et al., 2002; Mukherjee and Banerjee, 2004).
Artificial Neural Network
Artificial neural network is the model and trained on a given set of data
and then used to predict new data point and provide a mathematical alternative
to quadratic polynomial for representing data derived from statistically designed
experiments. Artificial neural network’s strong points are that they work
well with large amount of data and handles them easily with out requiring no
mechanistic description of system, this make artificial neural network particularly
well suited to medium optimization (Kennedy and Krouse, 1999).
First data generated by conducting a series of experiments and a network is constructed and getting the network to learn on these data set, once trained, the network is given new data points (media composition or fermentation process condition) and the out put (microbial performance or product formation) predicted. Artificial neural networks are well suitable for predicting the outcome from the fermentation process there by saving time and efforts (Patnaik, 2005). However artificial neural networks are simply a modeling tool and does not work properly when input data sequence are missing neural network s confused when different data are generated for same set of experiments but averaging the data can solve the problems.
Fuzzy Logic
Fuzzy logic utilizes and executes a series of rules using Fuzzy membership
functions. At first the Fuzzy memberships are defined. This defines what should
be the level of the components in a fermentation media whether it is in low
or high. Then next sets of experiments are defined based on results obtained
from the first set of experiment (Ulhaq and Mukhtar, 2006). When a new medium
composition is entered in Fuzzy logic programme, it predicts the result or the
out come (microbial performance or product formation) (Anderson and Jayaraman,
2005; Kennedy and Krouse, 1999).
Genetic Algorithms
In recent years nonstatistical optimization techniques such as genetic
algorithms are used in fermentation technology. This is a powerful stochastic
search and optimization technique, this technique can be used to optimize fermentation
process with out need of statistical designs and empirical models and based
on the principle that after a continuous process of mutation only best individual
exist. These individuals strive for survivals. After some number of generations
only the best individual hopefully represents the optimum solution. In fermentation
media or fermentation process optimization rules of genetic algorithms can be
applied successfully where the set of one experiment i.e. medium composition
are coded in one chromosome and each medium constituent level represents one
gene after completing the first generation of experiments chromosome with highest
productivity are selected and replicated proportionally to the productivity
then cross over of chromosome and mutation of some randomly chosen genes are
performed. In such a way, new generations of experiments are obtained. But main
disadvantage of genetic algorithms is that it does not store the information
generated at each stage of the optimization process (Zuzek et al., 1996).
A hybrid of genetic algorithms and artificial neural network approach was realized
to optimize fermentation process. This technique based on principle that after
a satisfactory neural network model and input space which is generated over
the range of independent parameters, can be optimized using genetic algorithms
the advantage of this technique is that neural network provide better fits to
experimental data then quadratic polynomial equation and model optimized by
genetic algorithms approach which provide a better alternative to the conventional
RSM approach to optimize fermentation process (Nagata and Chu, 2003).
CONCLUSIONS Designing a fermentation media or optimization of fermentation process can be never ending task and every optimization techniques have their own advantages and disadvantages (Table 1). Historically onefactoratatime used mostly fallowed by full factorial technique and response surface methodology but Plackett and Burman’s design and component replacing can be useful for screening medium components. Recently nural networks fuzzy logics, genetic algorithms and different hybrid techniques such as CCDRSM, Plackett and BurmanRSM, factorial designRSM, evolutionary operationfactorial design and genetic algorithmsartificial neural network techniques used efficiently to optimize fermentation medium and fermentation process parameters.
Table 1: 
Type of fermentation optimization techniques and their advantages
and disadvantages 


REFERENCES 
Ahamad, M.Z., B.P. Panda, S. Javed and M. Ali, 2006. Production of mevastatin by solidstate fermentation using wheat bran as substrate. Res. J. Microbiol., 1: 443447. CrossRef  Direct Link 
Alexeeva, Y.V., E.P. Ivanova, I.Y. Bakunina, T.N. Zvaygintseva and V.V. Mikhailov, 2002. Optimization of glycosidases production by Pseudoalteromonas issachenkonii KMM 3549T. Lett. Applied Microbiol., 35: 343346. Direct Link 
Anderson, R.K.I. and K. Jayaraman, 2005. Impact of balanced substrate flux on the metabolic process employing fuzzy logic during the cultivation of Bacillus thuringiensis var. Galleriae. World J. Microbiol. Biotechnol., 21: 127133. Direct Link 
Banerjee, R. and B.C. Bhattachaaryya, 1993. Evolutionary operation (EVOP) to optimize three dimensional biological experiments. Biotechnol. Bioeng., 41: 6771. Direct Link 
Chakravarti, R. and V. Sahai, 2002. Optimization of compaction production in chemically defined production medium by Penicilium citrinum using statistical methods. Process Biochem., 38: 481486.
Dahiya, N., R. Tewari, R.P. Tiwari and G.S. Hoondal, 2005. Chitinase production in solidstate fermentation by Enterobacter sp. NRG4 using statistical experimental design. Curr. Microbiol., 51: 222228. CrossRef  Direct Link 
Jatinder, K., B.S. Chadha and H.S. Saini, 2006. Optimization of medium components for production of cellulases by Melanocarpus sp. MTCC 3922 under solidstate fermentation. World J. Microbiol. Biotechnol., 22: 1522. CrossRef  Direct Link 
Kar, B., R. Banerjee and B.C. Bhattachaaryya, 1999. Microbial production of gallic acid by modified solidstate fermentation. J. Ind. Microbiol. Biotechnol., 23: 173177. CrossRef  Direct Link 
Kar, B., R. Banerjee and B.C. Bhattachaaryya, 2002. Optimization of physicochemical parameters for gallic acid production by evolutionary operationfactorial design techniques. Process Biochem., 37: 13951401. CrossRef 
Kennedy, M. and D. Krouse, 1999. Strategies for improving fermentation medium performance: A review. J. Ind. Microbiol. Biotechnol., 23: 456475. CrossRef  Direct Link 
Kumar, D., V.K. Jain, G. Shanker and A. Srivastava, 2003. Citric acid production by solid state fermentation using sugar cane bagasse. Process Biochem., 38: 17311738. Direct Link 
Kumar, S. and T. Satyanarayana, 2004. Statistical optimization of a thermostable and neutral glucoamylase production by a thermophilic mold Thermomucor indicaeseudaticae in solidstate fermentation. World J. Microbiol. Biotechnol., 20: 895902. Direct Link 
Mukherjee, G. and R. Banerjee, 2004. Evolutionary operationfactorial design technique for optimization of conversion of mixed agro products into gallic acid. Applied Biochem. Biotechnol., 118: 3346. CrossRef  Direct Link 
Nagata, Y. and K.H. Chu, 2003. Optimization of a fermentation medium using neural networks and genetic algorithms. Biotechnol. Lett., 25: 18371842. CrossRef  Direct Link 
Nandi, R. and S. Mukherjee, 1988. Medium optimization for the fermentative production of glucoamylase by an isolated strain Penicillum italicum. J. Microb. Biotechnol., 3: 1521. Direct Link 
Naveena, B.J., M. Altaf, K. Bhadriah and G. Reddy, 2005. Selection of medium components by PlackettBurman design for production of L(+) lactic acid by Lactobacillus amylophilus GV6 in SSF using wheat bran. Bioresour. Technol., 96: 485490. CrossRef 
Panda, B.P., 2001. Optimization of Bioconversion of Myrobalan Tannin to Gallic Acid Under Modified SolidState Fermentation by Rhizopus oryzae NRRL 21498. Dissertation, Birla Institute of Technol., India.
Park, P.K., D.H. Cho, E.Y. Kim and K.H. Chu, 2005. Optimization of carotenoid production by Rhodotorula glutinis using statistical experimental design. World J. Microbiol. Biotechnol., 21: 429434. CrossRef  Direct Link 
Patidar, P., D. Agrawal, T. Banerjee and S. Patil, 2005. Chitinase production by Beauveria felina RD 101: Optimization of parameters under solid substrate fermentation conditions. World J. Microbiol. Biotechnol., 21: 9395. Direct Link 
Patnaik, P.R., 2005. Neural network designs for polybhydroxybutyrate production optimization under simulated industrial conditions. Biotechnol. Lett., 27: 409415. CrossRef  Direct Link 
Plackett, R.L. and J.P. Burman, 1946. The design of optimum multifactorial experiments. Biometrika, 33: 305325. Direct Link 
Rajendhran, J., V. Krishnakumar and P. Gunasekaran, 2002. Optimization of a fermentation medium for the production of Penicillin G acylase from Bacillus sp. Lett. Applied Microbiol., 35: 523527. Direct Link 
Sayyad, S.A., B.P. Panda, S. Javed and M. Ali, 2007. Optimization of nutrient parameters for lovastatin production by Monascus purpureus MTCC 369 under submerged fermentation using response surface methodology. Applied Microbiol. Biotechnol., 73: 10541058. CrossRef  PubMed 
Schmidt, F.R., 2005. Optimization and scale up of industrial fermentation processes. Applied Microbiol. Biotechnol., 68: 425435. CrossRef  Direct Link 
Singh, B. and T. Satyanarayana, 2006. A marked enhancement in phytase production by a thermophilic mould Sporotrichum thermophile using statistical designs in a costeffective cane molasses medium. J. Applied Microbiol., 101: 344352. Direct Link 
Stanbury, P.F., A. Whitakar and S.J. Hall, 1997. Principles of Fermentation Technology. Elsevier, London, UK.
Tavares, A.P.M., M.S.M. Agapito, M.A.Z. Coelho, J.A. Lopes da Silva, A. BarrosTimmons, J.A.P. Coutinho and A.M.R.B. Xavier, 2005. Selection and optimization of culture medium for exopolysaccharide production by Coriolus (Trametes) versicolor. World J. Microbiol. Biotechnol., 21: 14991507. Direct Link 
Tunga, R., R. Banerjee and B.C. Bhattachaaryya, 1999. Optimization of n variable biological experiments by evolutionary operationfactorial design techniques. J. Biosci. Bioeng., 87: 125131.
Ulhaq, I. and H. Mukhtar, 2006. Fuzzy logic control of bioreactor for enhanced biosynthesis of alkaline protease by an alkalophilic strain of Bacillus subtilis. Curr. Microbiol., 52: 224230. Direct Link 
Xiao, J.H., D.X. Chen, J.W. Liu, Z.L. Liu, W.H. Wan, N. Fang, Y. Xiao, Y. Qi and Z.Q. Liang, 2004. Optimization of submerged culture requirements for the production of mycelia growth and exopolysaccharides by Cordyceps jiangxiensis JXPJ 0109. J. Applied Microbiol., 96: 11051116. CrossRef  PubMed  Direct Link 
Xie, L., D. Hall, M.A. Eiteman and E. Altman, 2003. Optimization of recombinant aminolevulinate synthase production in Escherichia coli using factorial design. Applied Microbiol. Biotechnol., 63: 267273. CrossRef  Direct Link 
Zuzek, M., J. Friedrich, B. Cestnik, A. Karalic and A. Cimerman, 1996. Optimization of fermentation medium by modified method of genetic algorithms. Biotechnol. Tech., 10: 991996. CrossRef  Direct Link 



