Production of Fumaric Acid in Solid-State Fermentation (SSF) with ...

sciencediscussionAI and Robotics

Oct 20, 2013 (3 years and 9 months ago)

100 views

Production of Fumaric Acid in Solid
-
State Fermentation
(SSF) with
Rhizopus arrhizus
using
C
assava
B
agasse as
the
S
ole Carbon Source. Optimization with Neural
Network


Rogério A. Strapasson
1
; Adenise L. Woiciechowski
1
; Carlos. I. Yamamoto
1
; Ashok
Pandey
2
; C
arlos R. Soccol*
1


1
Federal University of Paraná; Dep. of Chemical Engineering; P. O. 19011


81531
-
990. Paraná. Brazil.

2
Biotechnology Division; Regional Research Lab. CSIR. Trivandrum
-

695
-
019. India
.


ABSTRACT



SSF technology has been known for cent
uries; at more or less 2600 BC, the Egyptians
have used it for making bread. Cassava bagasse is a starch
-
rich lignocellulosic residue
with 60
-
70% of residual starch, which was not extract at the industrial process of the
cassava roots. Thousands of this re
sidue is disposed daily at the environment. This solid
raw residue can be used as the sole carbon source to produce fumaric acid by fermentation
using Rhizopus arrhizus at the process called solid
-
state fermentation. This production is
largely confirmed to

the organisms of the order Mucorales, mainly to the genus Rhizopus.
Cassava bagasse was milled, added nitrogen source and other salts. Temperature, pH,
humidity and time of fermentation, were optimized using the neural network tool. The
proposal of this p
roject was to optimize and evaluate the production of fumaric acid,
through neural net work, an important tool to optimize and simulate processes.


2

INTRODUCTION



Solid State Fermentation (
SSF
) may be compared to a tri phase system, solid, gas and
liquid
. Inside the fermentation media, water may be founded in three distinct ways, the
water linked (composition water), water weakly linked (solvatation water), and water
adsorbed (free water), but in SSF, free water is not found.


The grown support and the
water that carry the nutrients for the growth of the
microorganism constitute the solid phase. The carbon source may be the own support, for
example starch, or not when glucose is adsorbed in polyurethane, an inert support.


Cassava is a root (
Mannihot e
sculenta Crantz
) used as a meal or processed to obtain
cassava flour. Cassava Bagasse is a solid waste composed by the fibrous material of roots,
containing part of the starch not extracted in the process. Fumaric acid is an organic acid,
used industrially

as an intermediate in chemical synthesis at sterifications reactions. It is
used as food additive, anti
-
oxidant, acidulante in food industries, pH corrector; used in
pharmaceutical industries to prepare medicines due to its properties of low toxicity and
low
water absorption.


Neural Network is a class of computers programs based in the working of the brains of
superior mammalians, in an attempt to imitate the intelligence (YAMAMOTO, 1998). The
utilization of Neural Network like a tool of optimizing pro
cess is crescent since 1943 when
McCullough & Pitts first discuss the software imitation of biologic systems for data and
information processing (BEZDEK, 1993). After the input of some data, the result of the
problem is given in the form of a response of t
he variable dependent (target variable) in the
conditions proposed, different of the conditions input to the program. It is based on the fact
that each independent variable affects the value of the response variable (dependent or
target variable). The prog
ram simulates the real process and projects the value of the
dependent variable at those particular conditions, based on the input data. Normally a group
of conditions with the dependent variables are tested, and the bigger value of the variable
response i
s found if this response is given by one of the conditions tested. The
advantageous of this program is that it is not necessary a regular interval between the
values of the dependent variables.


3

MATERIAL AND METHODS


Substrate



In this work it is utiliz
ed cassava bagasse given from Agroindustrial de Polvilho Ltda


Paranavaí
-

Paraná.

Cassava bagasse is milled to mesh 0.84 at 2.00 mm and utilized as the
sole carbon source by the microorganism.

Saline Solution Used in the Medium Optimized 1



The saline

solution used to complement the nutritional needs of the microorganism is
presented at Table 1.


Table 1. Nutrients utilized to compose the saline solution used at the optimized
fermentation medium.

Nutrient

Amount

CaCO
3

5 g

B
iotin 0.0002% Solution

1.25

mL

ZnSO
4
.7H
2
O

0.01 g

MgSO
4

0.0625 g

KH
2
PO
4

0.0375 g

KNO
3

3.76 g

Water

Up to: 250 mL


Fermentations


To optimize the fermentation physic conditions, the fermentation was carried on
erlenmeyers flasks of 250 mL, with 5 g of the dry substrate (cassav
a bagasse),
complemented with the saline solution shown at table 1 (1.5 mL/5 g substrate)

Fermentation Kinetic



The fermentation was carried on columns (Raimbault Columns), with samples being
collected each day, in duplicate. (Two columns each day). Bes
ides, the gas from the
columns are analyzed in a Gas chromatography, on order to evaluate the content of CO
2

(related to the microorganism growth), N
2

and O
2
. This is a test of respirometry (data not
shown).

Biomass Analysis


The protein formation (grow
th) was measured by the Kjeldahl method

(ADOLFO
-
LUTZ,
1985).

Fumaric Acid


Fumaric Acid produced was measured by HPLC, High Performance Liquid
Chromatography.

Neural Network


It was made on a feedforward back
-
propagation neural network system that dete
rmines
the best fermentation conditions for the
T
emperature, inoculation rate, pH, humidity, and
fermentation time (days of fermentation). These are called independent variables, which
variation affects the value of the dependent or response variable, the
fumaric acid produced
in each block of dependent variable.


4

RESULTS AND DISCUSSION



The best fumaric acid producer is the strain
Rhizopus arrhizus

NRRL 2582 among the
several strains of
Rhizopus

studied (screening data not shown). With the
Rhizopus arrhi
zus
NRRL 2582 strain, it was performed the optimization of the physical fermentation
conditions, using the Neural Network tool.

The data obtained from the Neural Network Program Simulation is presented at Table 2. It
shows the response in fumaric acid to e
ach block of tested conditions. Each block of these
conditions are the hypothesis to be tested at the program simulation.


Table 2. Results from neural network simulation for the test conditions proposed,
generating objective values for fumaric acid.

Fumar
ic Acid

(g/kg cassava
bagasse)

Temperature
(
o
C)

Inoculation Rate

(spores / g dry
substrate)


pH

Initial
Humidity
(%)

T
ime

(days)

21.3703

32.00

1.00 x 10
8

8.60

70.00

11.00

57.8272

30.00

1.00 x 10
7

7.20

72.50

8.00

37.1695

30.00

1.00 x 10
8

8.00

72.00

13.
00

59.9384

30.00

1.00 x 10
7

7.20

72.50

13.00



The Neural Network System determined that the best fermentation conditions among the
range conditions tested are: the maximum inoculation rate tested (10
8

spores/g) and the
maximum humidity possible (72.5
%). pH = 7.2, temperature = 30
o

C and best fermentation
time (13 days) as shown in table 2. It is very clear the importance of the humidity.
Decreasing 0.5% the initial humidity, the fumaric acid produced decreased very hard. The
network accuracy may be v
iewed in the Figures 1 and 2.



Figure 1. Training
-
experiment values, circles shows experimental values and lines
plots the curves traced on synaptic weights obtained from the network.



The curves are calculated on synaptic weights to each point determi
ned by the network.


5






Figure 2. Test
-
experiment values, circles are experimental values and lines are the
curve traced for the network.




The first graphic (Figure 1) is the relationship between the fumaric acid production in the
training experiment

(real values fed to the program, and these data are used by the network
as a reference) and the objective value determined by the network. The second graphic
(Figure 2) is the relationship between values of fumaric acid production in the test
experiments
(real values fed to the program, but they are used as a check up for values
validation) and the objective value determined by the network. In other words, in these
graphics, the points assigned with circles are the experimental values determined for
fumari
c acid production in each group of variable (experimental conditions tested named
training experiments), when computer learns attributing different synaptic weights for each
variable, and named experiments of test when the computer plots the error curve fo
r
personal verification. As near as the curve is to the experimental points, the most accurate
is the analysis showed at Figure 2. According to Figure 2, the analysis done has a kind of
imprecision, but is satisfactory for our process.


6



Figure 3. SSF ki
netic for the production of Fumaric
A
cid and
B
iomass.



According to the Neural Network, the best fermentation time is 13 days. Even being the
best time of fermentation determined in 13 days by the Neural Networks, a simple kinetic
(Figure 3.) determine
that with 8 day of fermentation, the large amount of fumaric acid is
still formed. So fermentation with 8 days is economically the best time viable in an industry
in some practical fermentations processes. Greater fermentation times will not furnish a
corr
esponding increment at the fumaric acid production.



These experiments demonstrated that the Neural Networks really is useful tool to
determine the optimized fermentation conditions, and offers a way to verify theorically the
fumaric acid produced in an
y conditions desired, but it has to be used very carefully,
because an pure analysis without examine the process particularities may incur in error.


Process Scale
-
up


When one made a scale up of a fermentation system from 5g of dry cassava bagasse to a
tray system, through the column system it is critical the control of humidity. In column
system and in the tray system, the humidity must be 70 % at maximum, because 72.5 % is
too much water that drains from the support. This can cause eluation of the nutr
ients
inoculum and the risk of contamination at the bottom of the column and the trays by the
water that impregnate the cotton barrier to microorganisms that supports the weight of the
cassava bagasse structure in the columns. In tray system humidity may e
vaporate very fast
because of the large area and any kind of way must be used to keep the humidity, such as
sterilized water must be spilled to the system every day or twice per day, or the
environmental humidity must be assured with water surface, and eve
n so the production
may decrease for the lost of water.



7


CONCLUSION


The use of cassava bagasse showed to be a suitable substrate for fermentative processes,
such as the production of fumaric acid in solid state fermentation (SSF), with molds of the
gen
us
Rhizopus,

when complemented with some salts and a nitrogen source. The important
step of optimization of the fermentation conditions can be done using the Neural Network
System, taking care with the peculiarities of the process. When working with the pr
ocess
scale up, the humidity is an important factor that must be kept in order to keep the mold
activity.





REFERENCES


ADOLFO LUTZ,
Normas Analíticas do Instituto Adolfo Lutz
, vol.
I pp. 32
-
33. 1985.


BARRIOS
-
GONZÁLEZ, J. and MEJÍA, A. Production
of secondary metabolites by solid
-
state fermentation.
Biotechnol. Annu. Rev.

2, 85
-
121. 1996.


BEZDEK, J. C. A review of probabilistic, fuzzy, and neural models for pattern recognition

Journal of
intelligent and Fuzzy Systems
,

1(1):1
-
25. 1993.


CARTA, F.
S.
Hidrólise Enzimática dos resíduos de fecularias e produção de ácido fumárico por
fermentação de fungos do gênero Rhizopus.
Curitiba, 1999.
88f. Dissertação (Mestrado em Tecnologia
Química)
-

Setor de Tecnologia, Universidade Federal do Paraná.


LAROCHE,
C. & GROS, J.B.
Special transformation processes using fungal spores and immobilized cells.
Adv. Biochem. Eng. Biotechnol
.

55, 179
-
220. 1997.


PANDEY, A. Recent process developments in solid
-
state fermentation.
Process Biochem.

27, 109
-
117.
1992.


YAMAMO
TO, C. I.
Modelagem matemática e otimização do processo industrial de síntese da Amônia
Utilizando Redes Neurais.
São Paulo, 1998. 209f Tese (Doutorado em Engenharia)
-

Escola Politécnica,
Universidade de São Paulo.