You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267765677

Multiple-effect evaporation in a sugar factory. A measured variables study

Article in Latin American applied research Pesquisa aplicada latino americana = Investigación aplicada latinoamericana · October 2001

CITATIONS READS

4 7,351

5 authors, including:

Humberto Heluane Mauricio Colombo


National University of Tucuman National University of Tucuman
20 PUBLICATIONS 257 CITATIONS 23 PUBLICATIONS 101 CITATIONS

SEE PROFILE SEE PROFILE

Alejandra Ingaramo Maríaonga Rosa Hernández


National University of Tucuman National University of Tucuman
4 PUBLICATIONS 72 CITATIONS 13 PUBLICATIONS 48 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Process Ingineering View project

All content following this page was uploaded by Alejandra Ingaramo on 26 March 2015.

The user has requested enhancement of the downloaded file.


Latin American Applied Research 31:519-524

MULTIPLE-EFFECT EVAPORATION IN A SUGAR FACTORY.


A MEASURED VARIABLES STUDY
H. HELUANE, M. COLOMBO, A. INGARAMO, M. R. HERNÁNDEZ and M. CESCA

Departamento de Química- Facultad de Ciencias Exactas y Tecnología


Universidad Nacional de Tucumán
Av. Independencia 1800 (T4002BLS) Tucumán – Argentina
54-381-4364093 Ext. 201- e-mail : rhernandez@herrera.unt.edu.ar

Key words: Global heat-transfer coefficient - Multiple evaporation – Sugar industry.

Abstract:
With the aim to increase the efficiency of a sugar factory at Tucumán (Argentina), the evaporation stage is
analyzed due to its importance in the global efficiency of the process. The assays done to acquire measures of
mass flows, temperatures and pressures, as well as the periodic sampling of streams for the analysis of solid
concentration allowed the calculation of vapor flows and their variability. The contribution of those measured
variables to the steam flow variance is also estimated. The global coefficient of heat transfer in each effect was
also calculated under different operating conditions.

1. Introduction The heating vapor is separated from the boiling


The integration of Argentina to Mercosur implies liquid by a tubular surface. The exchange area of each
that cane sugar industry will go into a market of high effect is 1500 m2 except the forth effect which has an
complexity with product prices well below usual ones. area of 1350 m2. The heating steam is fed in the first
The payment of raw material has always been "the effect at about 110 °C.
slack variable" to hide inefficiencies of the productive The level of the boiling liquid in each effect is
process. With the introduction of a new concept of the controlled manipulating the flow of the exiting liquid.
sugar factory as a lender of raw material The level transmitter charts showed that boiling liquid
transformation services together with economic levels in the effects have narrow variabilities.
stability, looking inside the productive process is the Therefore, mass and energy balances in steady state
only way of cut off costs and maintain profitability could be used. The temperature in the barometric
Important advances on cost reduction can be condenser is controlled so that the disturbances of the
achieved just adjusting and correcting the detectable vacuum in the last effect are reduced.
inefficiencies of the productive process and through a The flow of juice fed to the first effect was
better organization, coordination and control of the measured with a segmental orifice plate and flow of
different involved operations that assure the maximum vapor bleeded from the third effect with an insertion
use of the sugar formed in the field. vortex meter. For each effect, temperatures of the
Permanent control on relevant variables of the boiling liquid and vapor were measured with
process is needed in order to achieve such objectives. thermoresistances Pt-100. All variables were difitally
In a previous work, the evaporation stage emerged as acquired. Samples of feed and concentrated liquor
the key process of the global mass and energy balances (product) were taken every 5 minutes. These samples
(Colombo et al.,1999). were analyzed with a refractometer to evaluate their
In this work, a study of variables and parameters concentrations expressed as degrees Brix. Most of the
for the multiple evaporation stage is presented. Several studied variables are actually measured at the plant on
variables such as mass flows, temperatures, pressures a regular basis but some of them were acquired
and specific gravity, were measured with a digital exclusively for this work.
system. Since this work was done at a real plant, the The number of measured variables are 14 (2 flows,
obtained results have potential practical application on 10 temperatures, and 2 compositions) while the
heat integration studies. Global heat transfer unknowns among flows, compositions, and parameters
coefficients were estimated for static and dynamic are 16. The equations in each effect are 4: one of heat
modeling puposes in process control synthesis to transfer, 2 of mass balances, and 1 of energy balance.
reduce global steam consumption of the plant. All non-measured variables are observables (Mah,
1990).
2. Studied system
The work was caried out in a sugar plant at 3. Experimental essays
Tucumán (Argentina). A scheme of the studied four Several runs in steady state were done. Main
effects system is shown in Fig. 1. features are listed bellow,

519
Latin American Applied Research 31:519-524

· Waiting until the variables were stable on a avoided. This is not always possible due to the
time basis. complex structure of the whole evaporation section.
· Sampling to analyse the concentrations of the For instance, the input variables sometimes suffer
feed and product. The sampling duration was severe changes caused by the action of other
about 2 hours. equipment or by the operators.
· Verifing the operation of the level control Data of three assays corresponding to different
loops to assure steady state conditions. days of the equipment’s working cycle are presented.
Each test is associated to a degree of fouling on the
During the test, any interference that could have heat exchange surface.
affected the regular performance of the equipment was

RAW JUICE

VAPOR BLEED CONDENSER

FRESH
E3 V3 Tv3
V1 Tv1 V2 Tv2 V4 Tv4 WA
WATER

FT FC
TC
1 2 3 4
F0 B0 T0 LT LT LT LT Tc TT

V0 Tv0 LC LC LC LC

F1 B1 T1 F2 B2 T2 F3 B3 T3 F4 B4 T4

STEAM HEADER

Fig. 1: Flowsheet of the studied evaporation system

3.1 Test A third effect occurred. This determined an erratic


This assay was done when the equipment was on behavior of the end concentration of dissolved solids.
the third day of operation of a 7 days working cycle, Elapsed time of Test B was about 150 min and the
which corresponds to a moderate fouling on the steady state was observed towards the last 60 min as
exchange surface. Temperatures did not show trends shown in figs. 5 and 6.
but presented great variability. Means, standard
deviations, and observed upper and lower values are Table 1: Statistics of the stationary variables (Test A)
shown in Table 1. The transients of measured variables Variable Mean Standard Minimu Maximum
during the assay are presented in Figs. 2, 3 and 4. Deviation m value value
As seen in Fig. 2, the flow of blended vapor T0 111.8 1.5 108.9 114.7
bleeded showed a moderate trend (not significant at a Tv0 110.2 3.5 106.4 119.2
level of confidence of 95%). During the test the T1 98.3 1.0 96.3 101.0
operator drastically increased the flow of pre- Tv1 96.8 1.2 94.3 99.8
evaporated juice fed to the first evaporator (see Fig. 3). T2 88.4 1.3 85.5 90.9
This caused a drop of solid concentration on the
Tv2 85.6 1.6 82.1 89.1
product as shown in Fig. 4 where steady states before
and after the disturbance are identified. T3 77.5 1.7 74.2 81.2
The statistical information of solid concentration Tv3 74.2 2.0 70.0 79.4
on the product (B4) and juice fed to the first effect (F0) T4 57.4 1.1 53.3 60.3
is presented in Table 2. Heat-transfer coefficients were Tv4 54.3 1.0 52.8 56.1
evaluated at the two steady states (see Table 6) Tc 36.0 1.1 33.2 38.4
E3 12.9 1.5 10.3 17.5
3.2 Tests B and C B0 22.3 1.1 20.3 24.0
Test B was carried out with the equipment just
cleaned. Therefore, the evaporator were analyzed with The working cycle of the equipment from a
highest heat exchange capacities. At the first stage of cleaning until the following one is usually of 7 days.
the test, some changes on the solid concentration of the Test C (2½ hours assay) corresponds to the sixth day
feed stream and on the flow of vapor bleeded from the of work, when the evaporator has an important amount

520
H. HELUANE, M. COLOMBO, A. INGARAMO, M. R. HERNÁNDEZ and M. CESCA

of fouling on the exchange surface. The results of conventional flowmeters. For that reason, other
Tests B and C are shown in Table 3. variable measures are required to calculate vapor
flows.
The resolution of mass and energy balances of the
40
quadruple effect leads to the estimation of different
vapor flows from the mean values of the measured
Temper ature of the water at the
30
barométric condenser(°C)
variables. The mass and energy balance, the equations
Solid conc entration in the feed of heat transfer, and the physical-chemical correlations
(° Brix)
20
and thermodynamic properties of the streams are
described with the following vectorial equation:
10
Vapor Bleed f(x,y) = 0 (1)
(tn/h)

0 where, x: input vector of variables and parameters, and


11:55 12:10 12:24 12:38 12:53 13:07 13:22 y: output vector of variables and parameters being
time
calculated.
Fig. 2: Transient of solid concentration in the feed
(B0), vapor bleed (E3), and water temperature at the Table 2: Statistics of the non-stationary variables (Test
barometric condenser (Tc) (Test A). A)
Initial Steady Final Steady
180 State State
Feed to the
first effect Variable Mean Standard Mean Standard
160
(m3/h) deviation deviation
F0 (m3/h) 130.8 5.3 165.1 4.5
B4 (°Brix) 69.4 1.2 61.8 0.1
140

The variability of the measured variables will


120
affect the estimation of the outputs. Covariance matrix
has strong diagonal dominance that implies a weak
100 correlation between the inputs. Under such condition,
11:45 12:00 12:14 12:28 12:43 12:57 13:12 13:26 13:40
the variance of the variable yi can be estimated as
time
follows (Feyo of Azevedo and Wardle, 1989):
Fig. 3: Juice flow fed to the first effect transient (Test
A). 2
æ ¶y ö
var( y i ) = å ç i ÷ var( x j ) (2)
70 ç ÷
j è ¶x j ø

Solid conc etration in the


68
produc t(Br ix) Equation 2 express a measure of the uncertainty in
66
the output estimate. The fractional contribution of each
Initial Steady State
input xj to the variability of the output yi can be
64
calculated as follows:

2
62
Final Steady State æ ¶y i ö
ç ÷ var( x j )
60 ç ¶x j ÷
w ij = è ø
11:45 12:00 12:14 12:28 12:43 12:57 13:12 13:26 13:40
(3)
tim e var( y i )
Fig. 4: Transient of the concentration of solids on the
The analysis of the experimental data allowed
concentrated liquor (Test A).
estimating the flows of vapor and its variances. The
derivatives in Eqn. 2 were evaluated numerically with
4. Estimation of the vapor flows
STATEVA (static simulator under Matlab
Steam consumption is a key variable in energy
environment that solves mass and energy balances for
balance of sugar factories, hence the estimation of
the evaporation section of sugar industries) while the
measurement of that variable has remarkable
variances were calculated from the time series of
impontance. In the studied system, no vapor flow
mensures. The estimated values of the flows and their
could be measured due to the big pipe diameters and to
standard deviations are presented in Table 4 for the
the lack of straight pipe length that allowed using
following extreme conditions:

521
Latin American Applied Research 31:519-524

· Test B: clean exchange surface The estimation of the steam flow into the first
· Test C: maximum fouling. effect is important for its influence on global energy
balance. On the other hand, the estimation of the vapor
Values of standard deviations oscillate between 1.2 leaving the last effect it is also important because this
and 2.1. flow is linked to water consumption in the barometric
condenser.
The flows of vapor generated in each unit are a
measure of the evaporation capacity. Table 4 shows
that the amount of vapor generated in the effects under
clean conditions is bigger, with a statistical
significance of 95%, than that generated under the
maximum fouling degree. Note that in both cases the
flow of processed pre-evaporated juice is similar while
the end solid concentration in that flow is higher for
clean evaporators (see Table 3).
The fractional contribution of the measured
variables to the variance of the steam flow V0 is
summarized in Table 5. The main contribution is given
by the fed juice composition.
Fig. 5: Transient of feed solid concentration (B0) and Note that the feed flow does not have high
bleeded vapor (E3) (Test B) significance in spite of its important fluctuations. It is
also seen that fractional contribution of the measured
temperatures T0, T1 and T2 are bigger in test C than in
test B as a consequence of the higher variances
observed in test C.

5. Global Heat-Transfer Coefficients Estimation


Global heat-transfer coefficients can be evaluated
from the estimated vapor flows. Values for limit
operating situations, clean surface and maximum
fouling, are presented in Table 7. As a reference,
coefficients estimated with the Swedish technologist’s
formula (see Eqn. 4) are shown in the same table. This
Fig. 6: Transient of juice flow fed to the first effect correlation is well known in the sugar industry
(Test B) although its precision is poor (de Armas et al., 1994).
For the effect i:

Table 3: Statistics of the measured variables Ti


Us i = 432 (4)
Test B Test C Bi
Variable Mean Standard Mean Standard
deviation deviation where Usi: global heat transfer coefficient for clean
T0 113.6 1.5 107.0 2.0 exchange surface (Kcal/h m2 °C); Ti: liquid
Tv0 113.9 1.5 110.8 2.8 temperature (°C); and Bi: solid concentration (° Brix).
T1 99.3 1.2 99.3 1.9
Tv1 97.7 1.2 97.1 2.1 Table 4: Estimated values and standard deviations of
T2 89.9 1.1 88.2 2.5 the vapor flows for tests B and C.
Tv2 86.8 1.5 85.0 2.7 Flow Test B Test C
(Clean equipment) (maximum fouling)
T3 78.6 1.3 78.1 2.1
(tn/h) Mean s Mean. s
Tv3 81.9 1.3 75.4 1.2
V0 26.1 1.6 21.5 1.5
T4 59.4 1.7 57.8 2.0
V1 29.5 1.5 23.2 1.3
Tv4 56.1 0.8 57.8 2.0
V2 31.0 1.5 25.3 1.2
Tc 33.8 0.4 39.1 1.4 V3 18.9 2.1 16.0 1.5
E3 12.4 1.8 10.8 1.3 V4 20.1 2.1 18.1 1.5
B0 19.6 0.3 29.5 1.1
F0 151.8 5.1 155.0 4.5 The value of U obtained for the first effect is low;
B4 64.6 0.9 62.9 0.9 less than half of the predictions of the formula of the
Swedish (see Table 7, Test B). When that situation

522
H. HELUANE, M. COLOMBO, A. INGARAMO, M. R. HERNÁNDEZ and M. CESCA

was commented to the plant personnel, they high variability in the transmitted signals was detected.
investigated and found that the chest was partially Therefore, in order to obtain more accurate estimations
flooded with condensed steam and, therefore, the unit of U coefficientes, temperatuere transmitters should be
was operating with a diminished evaporation capacity properly placed (e.g. far from the interface).
The values obtained for the second and fourth
effects are around the expected ones. The calculated Table 8: Temperatures fractinal contribution ( & i ) to
coefficiente for the third effect is bigger than expected the variability of U coefficients
but this is justified by the vapor bleed in that unit (Leal Test B Test C
et al., 1984) (see Table 7) U Tvi-1 Ti Tvi-1 Ti
U1 0.48 0.37 0.55 0.36
Table 5: Most significant fractional contribution (w ij)
U2 0.50 0.47 0.44 0.52
to the variability of V0
U3 0.52 0.44 0.56 0.43
Variable Test B Test C
F0 0.22 0.13 U4 0.23 0.28 0.25 0.39
B0 0.49 0.32
E3 0.11 0.05 6. Conclusions
T0 0.06 0.13 Test A showed the practical difficulty of analyzing
T1 0.05 0.17 the steady state in equipment with strong interaction
T2 0.02 0.11 with other equipment.
For a more precise calculation of the consumed
When fouling increases, a decrease of the U steam and vapor flows generated in each effect,
coefficient is expected. In this case the calculated U accurate measures of flow and concentration of the
coefficients remain practically constant (statistical feed to the first effect are needed.
significance at 95% of confidence) because the high The values of the heat transfer coefficients in the
coefficient variability masks the fouling influence (see first effect differ notably of the expected ones. This
Table 7). can be due to operating problems in the unit.
The estimated values for the U coefficients present
Table 6: Calculated global heat-transfer coefficients a high dispersion. The accuracy could be increased if
(Kcal/h m2 oC) – Test A temperature transmitters are placed far from the liquid-
Initial Final vapor interface.
Effect Steady State Steady State The chosen variables allowed an acceptable
Mean ) Mean ) estimation of vapor flows as well as steam
1 680 130 760 140 consumption at the first effect
2 1100 220 1260 240
3 1220 280 1420 320 Acknowledgements:
4 380 50 490 60
This work was partially supported by The
European Commission INCO-DC ERB IC18-CT98-
0271 and Consejo de Investigaciones de la
Table 7: Calculated and theoretical global heat-transfer Universidad Nacional de Tucumán (Argentina). The
coefficients (Kcal/h m2 oC) – Tests B and C authors would like to thank Ing. Gustavo Aso and
Test B Test C
factory personnel for their collaboration during the
Effect US
assays.
Mean ) Mean )
1 1960 630 80 630 130
Notation
2 1370 1360 270 910 230 B: Fraction of solids in juice and concentration
3 820 1380 290 1310 430 liquor (° Brix:- % in weight)
4 430 360 40 370 50 E: Flow of vapor extracted from an effect
f: Vector of function
F: Flow of juice and liquor (tn/h)
It is known in industrial experiences that a great T: Temperature of juice and liquor (°C)
dispersion of values is normally obtained (Smith and Tc: Temperature of the barometric condenser
Taylor, 1981). Note from Table 8 that the main Tv: Temperature of vapor (°C)
contributions to the variability of the heat-transfer U: Calculated global heat-transfer coefficient
coefficients are given by vapor and liquid temperatures (Kcal/h m2 °C)
in each effect. Such contributions are high because U Us: Global heat-transfer coefficient from Eq. (4)
coefficients are very sensitivities to vapor and liquid (Kcal/h m2 °C)
temperatures, and also because high individual V: Vapor or steam flow
variability was observed. Since the temperature var: Variance
transmitters are placed near vapor-liquid interface, a x: Input vector of variables and parameters

523
Latin American Applied Research 31:519-524

y: Output vector of variables and parameters Estación de Evaporación-Calefacción. Su Cálculo


Greek letters Mediante Ajuste de Datos de Fábrica”, Int. Sugar
s: Standard deviation J., 96 (1146S), 221-224 (1994).
w: Fractional contribution to the global variability Feyo de Azevedo, S. and A. P. Wardle, “Sensitivity
Analysis Concerning the Design and Operation of a
Subscripts Tubular Fixed-Bed Catalytic Reactor”, Chem.
i: Effect or generic component of a vector Engng. Sci., 44 (10), 2311-2322 (1989).
j: Generic component of a vector Leal, D., P. Friedman and A. Valdés, “Influencia del
0: Feed to the first effect Uso de Extracciones sobre el Coeficiente de
Transferencia de Calor Global en Evaporadores”,
References Cuba-Azúcar, (1), 3-9 (1984).
Mah, R.,. Chemical Process Structures and
Colombo, M, H. Heluane, M.R. Hernández and M. Information Flows, Butterworths, Boston, UK.
Cesca, “Logro de Objetivos a través de (1990)
Reconciliación de Datos”. submited to IV Smith, I. and L. Taylor, “Some Data on Heat Transfer
Congreso Interamericano de Computación in Multiple Effect Evaporators”, Proc. Of The
Aplicada a la Industria de Procesos, San José de South African Sugar Tech. Association, 51-55
Costa Rica (1999). (1981).
De Armas, C., M. Ribas, J. Mesa and D. Leal, “Valor
de los Coeficientes de Transferencia de Calor en la

Received September 2, 1999.


Accepted for publication March 14, 2001.
Recommended by A. Bandoni.
524

View publication stats

You might also like