Vous êtes sur la page 1sur 18

Multicriteria-oriented preliminary design of a ash evaporation

process for cooling in the wine-making process


Vanessa Ho Kon Tiat
*
, Patrick Sebastian, Jean-Pierre Nadeau
Laboratoire Inter-etablissements Transferts Ecoulements Fluides Energetique, TR.E.FL.E ENSAM, UMR CNRS 8508,
Esplanade des Arts et Metiers, 33405 Talence, France
Received 19 March 2007; received in revised form 10 August 2007; accepted 10 August 2007
Available online 31 August 2007
Abstract
The ash evaporation process uses pressure dierences to cause instantaneous uid cooling and concentration. It is thus used for
shortened cooling time in the wine-making process. Although promising since it allows grape cooling and concentration in addition
to improving wine quality with one single device, the development of this process remains limited because of the poor eciency of
the thermal exchanges and the important size of the installations. The preliminary design stage is one of the rst steps of the design pro-
cess. Often neglected, this stage is nevertheless important since decisions taken at this point, like gauging and choosing components, are
of major importance in relation to global development costs.
In order to improve the system compactness and thermal eciency and, to support the decision-making process, a multicriteria and
multiphysics-oriented preliminary design of an innovative two-stage ash evaporation process for cooling in wine-making process was
performed. Using a specic formalisation methodology that uses Constraint Satisfaction Problem techniques to enable taking into
consideration dierent types of data, constraints related to the physical behaviour of the system and to manufacturing, encumbrance,
eco-design and cost criteria are simultaneously included in the design model. The entire physical model, considering multiple coupled
phenomena, was validated through an experimental analysis carried out on an instrumented pilot unit. Results are presented and dis-
cussed. Global results, considering physics, gauging, mass, costs and eco-design constraints are also presented and discussed. A detailed
design ash evaporator specically for industrial use is briey described.
2007 Elsevier Ltd. All rights reserved.
Keywords: Grape cooling; Flash evaporation; Multicriteria; Preliminary design; Decision support; Constraint satisfaction problem; Ecodesign
1. Introduction
The Wine-making process consists in the conversion of
grapes into wine. It is a long succession of operations
that vary according to the type of wine. Recently, some
scientic works have highlighted the advantage of using
ash evaporation for the cooling process in wine-making,
to improve wine quality (Ageron, Escudier, Abbal, &
Moutonet, 1995; Escudier et al., 1995; Escudier,
Mikolajczak, & Moutounet, 1998). However, since grapes
are biological products, their processing involves specic
constraints that tend to lower the cooling eciency of
the ash evaporation process. Therefore, to ensure an
optimized cooling eciency, volumes and costs are some-
times inadequate with regards to industrial needs. Thus,
there is a need to develop tools to improve the relation
between sizes, eciency and costs in the ash evaporation
process.
This article deals with the development of a knowledge
base in order to facilitate the preliminary design of a ash
evaporation process dedicated to the wine-producing eld.
This base was built from experimental results translating
the physical behaviour of the components. These results
were obtained with an instrumented pilot unit described
below. Technical, economical and environmental criteria
were considered in the knowledge base.
0260-8774/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2007.08.015
*
Corresponding author. Tel.: +33 0 556 845 400; fax: +33 0 556 845
436.
E-mail address: v.hokontiat@gmail.com (V. Ho Kon Tiat).
www.elsevier.com/locate/jfoodeng
Available online at www.sciencedirect.com
Journal of Food Engineering 85 (2008) 491508
2. Flash evaporation in the wine-making process
The wine-making process is a lengthy succession of
operations, from grape harvesting to bottling, that results
in the transformation of grapes into wine.
After the harvest, the destemming process separates the
grapes from the stems. Then, grapes are pressed during the
masking step in order to reach a liquid state mixed with
solid particles (skin, pulp and seeds) that we shall call mix-
ture in this work. This mixture is then stored for macera-
tion and alcoholic fermentation, for the improvement of
wine colouring and tannic richness. The wine thus obtained
is then pressed to reach to the clarication and chillproof-
ing steps, until the nal bottling step (Vinihor web site).
In some traditional post-harvest processes, heating
before storage is used to improve some wine gustative
properties (Celotti & Rebecca, 1998). But the mixture has
to be quickly and homogeneously cooled to a suitable tem-
perature so that the yeasts are not destroyed and that fer-
mentation can occur. Many processes are used for this
task, such as tubular heat exchangers, cryogenic refrigera-
tion by injections of liquid CO
2
, immersion and ash evap-
oration process.
Flash evaporation is the phenomenon observed when a
liquid uid is suddenly exposed to a pressure loss under
its saturation pressure. Because of to this sudden pressure
loss, the uid becomes superheated. As this energy surplus
cannot be contained in the uid as sensible heat, it is con-
verted into latent heat of vaporization. As a result, the
most energizing molecules of liquid are extracted and reach
a vapour state, leading to the instantaneous cooling of the
remaining liquid that reaches a state of thermodynamic
equilibrium.
Vintage cooling by ash evaporation is thus caused by
the vaporization of part of the mixture in a low pressure
system. This vaporization generates vapour that must be
condensed to ensure thermal eciency. As a consequence,
a ash evaporator is usually composed of a separation
chamber where the uid is being cooled, a condensation
unit made of shell-and-tube condensers for vapour treat-
ment, and a vacuum unit that enables the pressure drop
in the entire apparatus.
Wine constituents that have evaporated inside the sepa-
ration chamber during the cooling process are water, vola-
tile compounds and primary aroma (Sebastian & Nadeau,
2002). The ash evaporation process thus allows wine con-
centration in addition to grape cooling. Furthermore, the
sudden pressure drop at the separation chamber inlet causes
mechanical constraints that disorganize the grapes skin tis-
sues. Skin tissues cells are made up of aromatic compounds
like tannins that give the wine its specicity and their disor-
ganization intensies the diusion phenomena observed
during the maceration step. This is conrmed by the studies
of Ageron et al. (1995) and Escudier et al. (1995, 1998) on
Nomenclature
A heat transfer area, m
2
C cost,
Cp heat capacity, J kg
1
K
1
d diameter, m
e thickness, m
g gravity acceleration, m s
2
G mass ux, kg m
2
s
1
h local heat transfer coecient,
W m
2
K
1
Dh latent heat of evaporation, J kg
1
k global heat transfer coecient,
W m
2
K
1
m mass, kg
NTU number of transfer units
P pressure, Pa
Po power, kW
Pr Prandtl number
DP pressure dierence, Pa
q mass ow rate, kg s
1
Q volumic ow rate, m
3
s
1
Re Reynolds number
S section, m
2
T temperature, K
V volume, m
3
X strength
Greek symbols
e heat eectiveness
l dynamic viscosity, kg m
1
s
1
q density, kg m
3
k thermal conductivity, W m
1
K
1
r surface tension, N m
1
Subscripts
cd condensation
cdsr condenser
cl coolant
cp condensate pump
ext exterior
f uid to be treated
i inlet
int interior
l liquid, condensates
o outlet
ref reference
sat saturation
sc separation chamber
v vapour
w plate wall inside the condensers
1 high pressure stage
2 low pressure stage
1 ?2 intermediate state between 1 and 2
492 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
the inuence of ash evaporation on wine quality. These
studies have also shown that wines treated with ash evap-
oration are coloured and that their colouring has a great sta-
bility. In addition, the eect of vacuum varies according to
wine types. It is thus possible to conclude that ash evapo-
ration preserves and accentuates wine specicity without
involving a standardization of its quality.
Therefore, ash evaporation proves to be of great inter-
est in the winemaking process, as it performs three dierent
operations in one device: grape cooling, wine concentration
and improvement of wine quality. One can wonder why this
process is not more widespread in wine-producing elds.
3. Toward the optimisation of the design of the ash
evaporation process
The weak development of the ash evaporation process
in wine-producing elds is mainly due to limiting factors
that reduce the thermal and size eciency of the process.
Indeed, in the ash evaporation process, the pressure
drop has to be sucient enough to ensure that the grapes
skin tissues are destroyed. Flash evaporators used in wine
producing elds are thus mono-staged since the pressure
drop occurring is very abrupt in such systems. Condensers
used in those devices have a large exchange area. This leads
to bulky systems compared to the vintage quantities they
are able to treat, with a low thermal eciency.
Moreover, the abrupt pressure drop leads to the gener-
ation of a spray at the inlet of the separation chamber.
Some droplets containing sugar and pulp are carried along
with the vapour that goes to the condensation unit. In this
condition, the process may not be able to concentrate the
mixture, since other constituents than water might be con-
tained in the condensates and these additional constituents
may clog up the condensers and reduce their thermal e-
ciency. To avoid these undesirable eects, droplets consti-
tution is eliminated both by using a large-sized separation
chamber and by inserting a mist eliminator between the
separation chamber and the condensation unit. But mist
eliminators generate pressure drops that tend to lower the
cooling eciency of such process.
Thus, the design of such systems for vintage treatment
has to be optimized.
3.1. Design considerations
According to Pahl and Beitz (1996), the design process
can be divided into four stages:
The Planning and Clarifying the Task stage, for the for-
malisation of requirements. This leads to the denition of
specication lists of functions to be fullled by the system.
The Conceptual Design stage, where creativity pro-
cesses are used to evaluate dierent principles of solu-
tions in order to end up with concepts that meet the
requirements. This stage corresponds to the denition
of the problem.
The Embodiment Design stage, where the main char-
acteristics of the system are selected and where the glo-
bal performances of the concepts retained are estimated.
This leads to choices of architecture, structure, materials
and dimensions.
The Detailed Design stage, which corresponds to the
denition of complete les for the system. The selected
solution is tested to see if all requirements are satised.
The conceptual design stage and the embodiment design
stage are gathered under the name of Preliminary Design
stage. The functional structure of the system, its compo-
nents and geometry are chosen during this stage. It is there-
fore of major importance for the complete decision-making
process and life cycle of the system. According to the study
of Berliner and Brimson (1988), the preliminary design
stage might engage up to 70% of the project development
cost, whereas it represents only 5% of this cost.
The optimization of the design of a process thus passes
through the selection of design constraints in the prelimin-
ary design stage. In this study, we focused on the develop-
ment of preliminary design considerations for a ash
evaporation process, in order to optimise the system in
relation to design considerations.
3.2. Towards a two-stage ash evaporation process
A previous study has been carried out on the thermody-
namic analysis of falling jet ash evaporator performance
for vintage treatment (Sebastian & Nadeau, 2002) which
led to the understanding of physical phenomena occurring
during the ash evaporation mono-stage process. This
study also displayed interest to reach a two-stage system
to increase thermal eciency and compactness.
More recently, Bouchama, Sebastian, and Nadeau
(2003) proved the great thermal eciency and compactness
of a two-stage process compared to the mono-stage pro-
cess. They reached a denition of design constraints related
to the physical behaviour of this two-stage process consid-
ering shell-and-tube condensers that can be used for preli-
minary design modelling.
Using these works as a starting point, an industrialized
two-stage ash evaporation process was dened with sepa-
ration, condensation and vacuum units. The separation
unit puts together a buer tank provided to regulate uid
supply, a two-stage separation chamber and an extraction
pump, for uid recovery. The separation chamber is
divided into the high pressure stage and the low pressure
stage. These stages are separated by a oat that regulates
the ow between them. The uid being treated accumulates
in the bottom of the high pressure stage until it reaches a
sucient level to activate the oat. It then drops into the
low pressure stage because of gravity.
In order to improve the compactness of the apparatus,
the shell-and-tube condensers at each stage of the conden-
sation unit were replaced by vertical welded-plate and shell
condensers. These condensers show a great exchange area
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 493
for a smaller global size. To avoid droplets transportation
towards the condensation unit, a mist eliminator was envis-
aged between the separation chamber and the condenser
for the high pressure stage.
The vacuum unit is composed of a liquid-ring pump
coupled with an air ejector to improve uid cooling tem-
perature (Sebastian & Nadeau, 2002).
3.3. Formalization of the design problem
This industrialized process has been entirely dened but
has not been validated. The purpose of this work is thus
devoted to the entire denition and validation of con-
straints related to the physical behaviour of this industrial-
ized version of the two-stage ash evaporation process.
Moreover, we sought to complete the preliminary design
of this process by integrating cost and manufacturing
requirements and ecocriteria in addition to the laws of
physics describing the physical behaviour.
To make it possible to treat simultaneously all those var-
ious-typed requirements from the preliminary design stage,
we adjusted a specic formalization methodology that
leads to the translation of the design problem, composed
of all requirements, into a specic formalism that enables
the processing of dierent types of data. This formalism
is called Constraint Satisfaction Problem (CSP).
3.4. Constraint satisfaction problem
A Constraint Satisfaction Problem (Fischer, Nadeau, &
Sebastian, 2002; Scaravetti, Nadeau, & Sebastian, 2003;
Thornton, 1996) is dened as a problem described in terms
of a set of relationships called constraints C, variables
V and domain values D for these variables. They are
based on the notion of constraint, which converts knowl-
edge into the form of conditions of compatibility between
variables of a design problem. Constraints are relations
which restrict the variable domains. Values assigned to
the variables must belong to their respective domains and
satisfy every constraints of the problem.
They are usually expressed as explicit algebraic rela-
tions, and can be divided into three categories:
Equal constraints (X = Y) which usually represent
laws of physics or denitions of performance criteria
(see Eq. (1) for example).
Unequal constraints (X < Y) which usually refer to
economic constraints (cost), required space, etc. (see
Eq. (11) for example).
Logical constraints (X and Y, X or Y, etc.) which
represent conditional constraints such as technical skill
rules, selection of components from catalogues, etc.
Constraint Satisfaction problem are treated with digital
CSP solvers. The one used in this work is a mixed digital
one called Constraint Explorer

(Zimmer, Anglada,
Christie, & Granvilliers, 2004). This software has been
developed in the context of project CO
2
, a French Project
coordinated by the aeronautics company DASSAULT-
Aviation, the computer science departments of LINA
(Nantes) and LIP6 (Paris VI) and the TREFLE (Bor-
deaux) and ESTIALIPSI (Bayonne) industrial design
laboratories.
The Constraint Explorer

solving algorithm is based on


interval analysis (Benhamou & Older, 1997; Jaulin, Kieer,
Didrit, & Walter, 2001). Constraint Explorer

processes
the knowledge base in a two-phase iterative and sequential
alternance which gradually reduces and partitions the
domains assigned to the variables until they satisfy all the
constraints of the problem.
Solutions of a design problem expressed as a CSP are
restrictions of the denition domains of the variables that
satisfy all the constraints, and all the existing solutions to
the problem are included in these restrictions. Therefore,
by using constraint expression, all relevant parameters
can be incorporated into this knowledge database.
4. Preliminary design of the two-stage ash evaporation
process
4.1. Formalization methodology
As seen previously, the translation of the design problem
as a constraint satisfaction problem is achieved through a
formalisation methodology starting from the denition of
a concept of solution to the selection of suitable architec-
tures in the entirely explored solution domain. On this pre-
mise, this methodology provides decision support, since it
is possible to deal with all possible solutions of the design
problem and select the most appropriate solution accord-
ing to the needs. Furthermore, models resulting from this
methodology are qualied in terms of decision support
through the four criteria of parsimony, accuracy, precision
and specialisation.
Parsimony is the ability of a model to describe correctly
the behaviour of a system with a minimum number of vari-
ables and relations.
Accuracy is the distance measurement between the solu-
tions space of the model considered and reference
behaviour.
Precision is the extended eld of possible values for a
variable.
Specialisation is composed of assumptions and informa-
tion that tend to restrict the applicability of a model.
A model will be great for decision support if its parsi-
mony, accuracy and precision are high and its specialisa-
tion is low. However, such a model does not exist, and a
trade-o has to worked out in order to reach to a suitable
model.
The formalisation methodology, presented in Fig. 1,
consists of three steps. Analysis and structuring, rst,
are devoted to the denition of the design problem and
the decomposition of the system into a set of components
and interactions in order to simplify the design. This is
494 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
carried out by translating these requirements into techno-
logical, economical, environmental, qualication or mar-
keting criteria and determining relevant characteristics
(variables, life cycle situations, functions, etc.) for the
design problem. Then, the formalisation and qualication
step is dedicated to validating the models before the pro-
cessing step.
More details on the global methodology can be found in
Scaravetti, Pailhe`s, Nadeau, and Sebastian (2004), in Sca-
ravetti, Nadeau, Pailhe`s, and Sebastian (2005) and in Ver-
nat, Fischer, Nadeau, and Sebastian (2004) works.
For the two-stage ash evaporation process structuring,
we envisaged the following life cycle situations: production,
exploitation, switching o, cleaning operation and trans-
portation. In this paper, the only life cycle situation pre-
sented is the exploitation phase. The identied functions
used to describe requirements and expected performances
in this life cycle situation are as follows:
cooling and concentrating the uid,
treating the formed vapour,
operation in vacuum conditions,
regulating uid supply.
Fig. 2 shows the realization mode of these functions,
expressed in the Function Analysis System Technique
(FAST) formalism. This formalism emphasizes essential
functioning units. This formulation of the problem allows
the identication of the relevant components of the system.
With a thorough analysis of the components interactions,
through organization charts, Functional Blocks Diagram,
Fields and Material graphs (Nadeau & Pailhes, 2006;
Nadeau, Pailhes, & Scaravetti, 2006; Vernat et al., 2004),
it is possible to select the relevant parameters that have
to be taken into account in the design model.
4.2. Experimental study
An experimental study, provided with an instrumented
pilot unit, has been used to validate the modelling of the
physical behaviour of components. As seen previously,
the two-stage ash evaporation system is composed of a
separation unit where mixture is being cooled and concen-
trated, a condensation unit for vapour treatment and a vac-
uum unit to satisfy the pressure drop inside the apparatus.
The physical behaviour of all components has to be mod-
elled for the process design.
Fig. 1. Operations of the embodiment design stage according to the design formalisation approach.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 495
Facing the problem of vintage availability for experi-
ments, we used water as the processed uid. This choice
was made since water is almost exclusively the only uid
evaporated during vintage cooling. At this stage of the
wine-making process, the liquid phase of the mixture is
only composed of water and sugar. The high sugar concen-
tration brings the density of this liquid phase of about ten
per cent higher than pure water density. Its capacity is
approximately fteen per cent lower than pure water. Tak-
ing into account these dierences, it is possible to transpose
easily the results obtained for water as processed uid to
vintage, using the model described below. This is conrmed
by the works of Sebastian and Nadeau (2002).
The two-stage ash evaporation instrumented pilot unit
used for the experiments, displayed on Fig. 3, is composed
of a two-stage separation chamber separated by a oat.
Downstream from this chamber, is a tank for uid recov-
ery. Upstream from the separation chamber, an additional
heating tank is devoted to uid heating and supply. This
tank is provided with a mixer to maintain uid temperature
homogeneity. The condensation unit is provided with two
vertical welded-plate and shell condensers, one for each
separation stage. A liquid ring vacuum pump coupled with
an air ejector constitutes the vacuum unit.
During the operation, the entire apparatus is put under
vacuum conditions. When starting one experiment, vintage
coming from the supply tank at a temperature between
343.15 K and 363.15 K, is sucked up the high pressure
stage of the separation chamber at a pressure of approxi-
mately 10
4
Pa. Due to the pressure loss, part of the liquid
phase is violently vaporized, thus cooling the remaining
part to around 313.15 K. The height of the cooled uid
increases in the high pressure stage and activates the oat.
It is then once again partially vaporized in the low pressure
stage at a pressure close to 4 10
3
Pa, and reaches a tem-
perature of 303.15 K.
During the ash, droplets are produced. If the chamber
is not big enough, compared to inlet and outlet tempera-
ture dierences, these droplets can be carried along with
the vapour and clog up the condensation unit. This will
lead to a decrease in the system performance. To prevent
this from happening, a mist eliminator has to be placed
between the separation chamber and the condensation
unit. In the low pressure stage, the dierence between the
inlet and outlet temperatures (about 10 K) leads to an
acceptable tank diameter (see encumbrance and mass mod-
els paragraph) that ensures that droplets will not be carried
along with the vapour. On the other hand, the temperature
dierence in the high pressure stage reaches a tank diame-
Fig. 2. Realisation mode of functions related to requirements, expressed
in the Function Analysis System Technique (FAST) formalism.
Fig. 3. Front and side views of the experimental two-staged pilot unit.
496 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
ter that is too big to guarantee agreement with encum-
brance requirements, and a mist eliminator was placed
between the high pressure separation stage and the
condenser.
Various tests were carried out taking into account tem-
perature, pressure and ows measurements at operation
points in the system:
Temperature measurements with T type thermocouples,
with an accuracy of 0.5 C and a measuring range
between 45 and 125 C.
Pressure measurements with passive pressure transduc-
ers with piezoelectric room, with a measuring range of
[0200] mbar for the high pressure stage and of [0
100] mbar for the low pressure stage. Their accuracy is
about 1% of the upper limit of measuring range.
Flow measurements using a vapour owmeter, with a
measuring range from 36 to 360 m
3
/h and an accuracy
of 1.6% of the maximum capacity. All these sensors
are connected to a numerical acquisition system made
up of a computer (CPU: 250 MHz, RAM: 64 Mo) and
of a National Instrument SCXI acquisition station with
32 channels.According to these sensors accuracies, mea-
surements uncertainties are about:
0.5 C for temperature measurements,
1 mbar for low pressure measurements,
2 mbar for high pressure measurements,
about 5.8 m
3
/h for vapour mass ow measurements
(with the owmeter),
1 s for time measurement,
10
3
m for heights measurements,
1 mL for volume measurements,
about 3 L/min for coolant mass ow measurements.
For each test carried out, the temperatures and pres-
sures at the various operation stages were recorded, as well
as mist eliminator droplets level and condensates levels in
the condensers. The dierent operation stages are pre-
sented in Fig. 4, displaying a scheme of the ash evapora-
tion pilot unit.
Fig. 5 shows the evolution of temperature, pressure and
vapour ow measurement during one experiment. We can
see that, when starting an experiment, the system initially
passes through a transient state before reaching its station-
ary state. This transient state corresponds to the thermal
equilibrium of the system. Moreover, the dierence
between the high pressure stage and the low pressure stage
is easily visible.
Fig. 4. Scheme of the high performance two-stage ash evaporation experimental pilot unit specifying sensors placement.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 497
A trial run was made in order to validate the modelling
of the physical behaviour for the relevant component pre-
viously identied.
4.3. Physical models
The main assumption on which all the physical models
are founded is that of saturated vapour. Indeed, we
assumed that, due to the separation chamber geometry,
vapour temperatures are homogeneous. Furthermore, the
chamber geometry guarantees that the vapour overheating
never exceeds 1.5 C. This overheating can thus be consid-
ered as negligible and the vapour temperature in the sepa-
ration chamber is assumed to be close to the saturated
temperature. This assumption was validated through the
experimental study, by comparing the vapour temperatures
recorded by the thermocouples with those obtained by cal-
culating the saturation temperatures corresponding to
pressure sensors measurements.
In these conditions, Sebastian and Nadeau (2002) have
shown that the evaporation phenomenon can be modelled
by a heat and mass transfer balance between the inlet and
the outlet of the separation chamber. The heat balance
enables us to evaluate the vapour mass ow rate in the high
pressure and low pressure stages, as shown in Eqs. (1) and
(2).
q
v1

q
fi
Cp
fi
T
fi
Cp
f1
T
v1

Dh
v1
1
q
v2

q
f1!2
Cp
f1!2
T
f1!2
Cp
f2
T
v2

Dh
v2
2
The mass balance led to the evaluation of the mass ow
rates necessary to the calculation of the vapour mass ow
rates, through Eqs. (3) and (4).
q
fi
q
v1
q
f1!2
0 3
q
f1!2
q
v2
q
fo
0 4
Furthermore, as the vapour overheating is negligible,
Eqs. (5) and (6) are useful to estimate the vapour and uid
temperatures through the separation chamber.
T
f1!2
T
v1
T
v:sat1
5
T
fo
T
v2
T
v:sat2
6
For the modelling of the condensers in the condensation
unit, an NTU-e model was used, as in the work of Boucha-
ma et al. (2003), to evaluate the experimental condensates
side heat transfer coecients of the condensers, on the
assumption that all the energy released during vapour con-
densation is transmitted to the coolant. This assumption
was validated through the comparison between the vapour
side and coolant side powers, estimated through experi-
mental results.
Fig. 6 shows the experimental validation of assumptions
of saturated vapour and energy conservation inside
condensers.
Using coolant and vapour temperature measurements
on the experimental pilot unit, the coolant heat eective-
ness was calculated for the high pressure and low pressure
condensers, as shown in Eqs. (7) and (8), where T
clo
and
T
cli
are the coolant outlet and inlet temperatures and T
v.sat1
and T
v.sat2
are the vapour temperatures at the condensers
inlet. If pressure losses are negligible, these vapour temper-
atures are similar to the vapour saturation temperatures in
the separation chamber. This is conrmed by experimental
measurements.
e
cl1

T
clo
T
cli
T
v:sat1
T
cli
7
0
20
40
60
80
100
120
1 52 103 154 205
time [s]
T [C]
P [mbar]
Qv [m /h]
3
T_HP
P_HP
T_LP
P_LP
Qv
Opening of the
fluid supply valve
Transient
state
Stationary
state
Fig. 5. Appearance of temperature, pressure and vapour ow measurement during an experiment.
498 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
e
cl2

T
clo
T
cli
T
v:sat2
T
cli
8
The evaluation of the coolant eectiveness values leads
to the calculation of the number of transfer unit NTU
through the energy balance relation for counter current
heat exchangers presented in Eq. (9).
ln
1
1 e
cl
_ _
NTU
cl
9
The NTU translates the exchange capacity of the con-
densers, and its denition expression is given in Eq. (10).
Eq. (9) thus translates, in a roundabout way, a relation
between coolant heat eectiveness and average global heat
transfer coecient k.
NTU
cl

k A
q
cl
Cp
cl
10
Thus, knowing the values of the heat exchange areas, the
coolant mass ow rates and heat capacities and the num-
bers of transfer unit, it is possible to calculate the global
heat transfer coecient k.
Moreover, the link between this global heat transfer
coecient and the local coolant side and condensates side
heat transfer coecients is translated in Eq. (11), where
the clogging resistances related to the plates walls are
negligible.
1
k

1
h
cl

1
h
cd
11
Vertical welded-plate condensers are not usually used in
ash evaporation processes. Thus, we came across some
diculties in nding correlations adapted to the constraint
satisfaction problem formalism that can be directly
adapted to the two-stage ash evaporation process
described. To overcome this problem, we chose to carry
out an experimental modelling of these condensers, in
order to nd correlations of local heat transfer coecients
that would t their real behaviour.
The coolant side heat transfer coecients h
cl
in the high
pressure and low pressure condensers have been modelled
through correlations obtained starting from data provided
by the condenser manufacturer, KAPP FRANCE (Kapp
France web site). These correlations are the result of an
interpolation of values of the heat transfer coecient for
given values of coolant mass ow rates. The correlation
translates the inuence of Reynolds number, mass ux,
heat capacity and Prandtl number of the coolant on the
local heat transfer coecient.
Eq. (12) gives the expression of the coolant side heat
transfer coecient for the high pressure stage condenser.
h
cl1
0; 031 Re
0;2
cl1
G
cl1
Cp
cl1
Pr
2=3
cl1
12
For the low pressure stage condenser, the coolant side
heat transfer coecient is expressed in Eq. (13).
h
cl2
0; 034 Re
0;2
cl2
G
cl2
Cp
cl2
Pr
2=3
cl2
13
These two expressions lead to an estimation of the cool-
ant side heat transfer coecient with an uncertainty of
approximately 10%.
0
20000
40000
60000
80000
100000
120000
140000
40.8 60.34 70.01 80.29 89.72
P
o
c
l

,

P
o
v
[
W
]
Po_v
Po_cl
0
10000
20000
30000
40000
50000
60000
40.8 60.34 70.01 80.29 89.72
P
o
c
l

,

P
o
v
[
W
]
Po_v
Po_cl
20
25
30
35
40
45
50
40.8 60.34 70.01 80.29 89.72
T [C]
fi
T [C]
fi
T [C]
fi
T [C]
fi
T
s
a
t
(
P
v
i
)
,

T
s
a
t
(
P
v
o
)
,

T
v
i
,

T
v
o


[

C
]
Tsat(Pvi)
Tvi
Tsat(Pvo)
Tvo
20
25
30
35
40
45
40.8 60.34 70.01 80.29 89.72
T
s
a
t
(
P
v
i
)
,
T
s
a
t
(
P
v
o
)
,
T
v
i
,
T
v
o
Tsat(Pvi)
Tvi
Tsat(Pvo)
Tvo
[

C
]
a
b
Fig. 6. Experimental validation of saturated vapour (a) and energy conservation inside condensers (b) assumptions.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 499
From the estimate of the coolant side local heat
transfer coecients and knowing the values of global
heat transfer coecients, as shown previously, the exper-
imental condensates side heat transfer coecients can
thus be estimated, with an uncertainty, linked to mea-
surements and uncertainty of intermediate values, of
about 10%.
In our case, condensation occurs outside the plates. So
we chose to model the experimental condensates side heat
transfer coecient using Nusselts correlation in vertical
plate condensation, putting aside the inuence of plate
corrugations.
Kuhn, Schrock, and Peterson (1997) have experimen-
tally studied condensation of a steam-gas mixture inside
vertical tubes and showed that Nusselts correlation shows
good results in modelling such phenomenon. They tested
three kinds of adaptation of this correlation to experimen-
tal results, and the most suitable was the use of a correction
factor, which is the ratio between the experimental heat
transfer coecient and the Nusselts theoretical heat trans-
fer coecient.
Considering these results, we used a correction factor to
adjust the estimates of Nusselts correlation to our experi-
mental results. The experimentally adapted correlation is
presented in Eq. (14) for the high pressure stage condenser,
where B
1
is the correction factor taking into account the
variations due to the experiments. This relation is appro-
priate for condensate Reynolds numbers from 0 to 150.
Eq. (15) highlights the correction factor B
1
.
These two expressions were obtained from experiments
with vapour temperature T
v1
of about 308 K to 323 K.
The uncertainty associated to the evaluation of this theo-
retical heat transfer coecient, linked with measurements
uncertainties, is about 13%. The details of measurements
uncertainties, particularly the sensors sensitivity, have been
previously presented.
h
cd1
B
1
1; 47 k
l1
Re
1=3
l1

l
2
l1
q
l1
:q
l1
q
v1
:g
_ _
1=3
14
B
1
0; 00071 T
3
v1
0; 67517 T
2
v1
213; 73839 T
v1
22552; 5 15
For the low pressure stage condenser, the theoretical
condensates side heat transfer coecient was estimated
the same way, as seen in Eq. (14). The correction factor
B
2
is obviously adapted to its own behaviour, as shown
in Eq. (15). Eqs. (16) and (17), related to the low pres-
sure stage condenser, were obtained from experiments
with vapour temperature T
v2
between 298 K and
313 K. For this condenser, the uncertainty associated
to the evaluation of the heat transfer coecient is about
30%. This high uncertainty, compared to the one of the
high pressure stage coecient, is due to the vapour mass
ow rates which were very low in some experiments,
leading to increased measurements uncertainties, due to
sensors sensitivity. Additional information is provided
below.
h
cd2
B
2
1; 47 k
l2
Re
1=3
l2

l
2
l2
q
l2
:q
l2
q
v2
:g
_ _
1=3
16
B
2
0; 000845 T
3
v2
0; 78902 T
2
v2
245; 46221 T
v2
25450; 1 17
This experimental modelling has the advantage of
matching closely the real behaviour of the condensers,
but it is limited to this application and thus specialized.
Figs. 7 and 8 show the comparison between experimen-
tal values and theoretical model results for the condensates
side heat transfer coecients in the high and low pressure
condensers. Each point of the curves represents the average
value for one experiment.
For the high pressure condenser (see Fig. 7), the shapes
of the curves for theoretical and experimental results are
quite similar. This expresses a good agreement between
those values. The few dierences observed between the
two series of values are explained by uncertainties caused
by correlation approximations. Several coupled phenom-
ena are included in the energy balances. The modelling of
these energy balances through correlations will thus lead
to some deviation between theoretical and experimental
0
500
1000
1500
2000
2500
3000
3500
4000
311.83 313.26 314.85 315.72 316.26 316.87
T
v.sat
[K]
h
c
d
[
W
/
m
2
.
K
]
hcd.th
hcd.exp
Fig. 7. Comparison of experimental and theoretical average local heat
transfer coecients in high pressure condenser.
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
303.68 307.2 308 308.52 308.96 310. 3
T
v.sat
[K]
h
c
d
[
W
/
m
2
.
K
]
hcd.exp
hcd.th
Fig. 8. Comparison of experimental and theoretical average local heat
transfer coecients in low pressure condenser.
500 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
values. Moreover, uncertainties due to experimental mea-
surements have also to be considered. Nevertheless, these
dierences can be put aside, since they never exceed fteen
per cent. The model accuracy is therefore convenient.
For the low pressure condenser (see Fig. 8), the agree-
ment between the theoretical and experimental results is
less obvious. The shapes of the curves seem to be quite dif-
ferent because of some divergent values that correspond to
specic experimental cases where vapour mass ow rates in
the low pressure stage were low. During the experimental
analysis, we observed that low vapour mass ow rates in
the low pressure stage condenser lead to very irregular tem-
perature measurements, because of the sensors sensitivity.
Thus, the evaluation of the experimental condensates heat
transfer coecient is disturbed.
The vapour mass ow rate is linked to the mass ow rate
of the cooled uid coming from the high pressure stage sep-
aration chamber and to uids temperatures, as shown in
Eq. (2). So, the vintage heating temperature has a great
inuence on the vapour mass ow rate. When the heating
temperature is low, the steam output in the condenser is
very low and the temperature in the sensor enclosure is
no longer homogeneous. The sensors thus record a lower
temperature than the actual one, which leads to a distortion
of the experimental values of the heat transfer coecient.
If these particular values are not taken into account, the
shapes of the curves become suciently similar to reach to
an acceptable accuracy in the low pressure condenser.
However, these observations show that the low pressure
condenser size could be reduced.
As for the condensation unit, droplets mass ow rates in
the mist eliminator were modelled and validated through
the experimental results. Droplets recovery measurements
were taken for each experiment. The correlation obtained
is presented in Eq. (16), and shows the link between droplets
mass ow rate, vapour mass ow rate and vapour tempera-
ture in the high pressure stage separation chamber high-
lighted during experiments. Uncertainties are about 20%
for the experimental values and 5% for theoretical ones.
For the vacuum unit, an air ejector was coupled with the
liquid ring vacuum pump, in order to increase the perfor-
mances of the ash evaporator by eliminating the risk of
cavitation and thus reducing the limit of operation of the
pump.
The model of the ejector was obtained through energy,
mass and linear momentum balances for each main section
considering relevant parameters for the two uids involved
(vapour and air) at these sections (Bouchama et al., 2003).
This model is based on the main assumptions of perfect u-
ids, stationary ows, negligible kinetic energy and adiabatic
conditions.
For the liquid ring pump, the volumetric ow is esti-
mated on the basis of pressures, volumes and temperatures
knowledge, as shown in Eqs. (18) and (19), where Qv
VP
is
the volumetric ow rate of the pump, R is the perfect gas
constant, M
m
is the molar mass of the vapour, q
leak
is the
leakage mass ow rate T
opt
and P
opt
are the temperature
and pressure that the pump is supposed to maintain during
the experiments, V
Sys
is the volume of the entire system,
DP
stop
is the pressure dierence after the pump stops.
The coecient of 0.1 refers to industrial vacuum conditions
for air leakage.
Qv
VP
q
leak

R
M
m
_ _

T
opt
P
opt
18
q
leak
0:1 V
sys

DP
stop
dt
19
In addition to this physical model, other design con-
straints were taken into account in the ash evaporation
process global model. These constraints are presented
below.
4.4. Embodiment models
One of the design objectives for the ash evaporation
system is to reduce its size, compared to existing systems.
Embodiment requirements have thus to be included in
the design model.
The global system encumbrance is mainly due to the size
of the separation chamber. Indeed, each stage of this sepa-
ration chamber must be sized so that it is of minimal diam-
eter to ensure that droplets are not carried outside, and of
minimal thickness to prevent buckling.
The minimal diameter modelling is based on droplets
diameter evaluation, given by Laplace relation for a
liquid-gas interface, as shown in Eq. (20), where d
dr
is the
droplets diameter. Subscript 1 refers to the high pressure
stage. For the low pressure stage, subscript 1 should be
replaced by 2.
d
2
dr1

4:r
1
P
sat
T
fi
P
sat
T
v1

20
Droplets diameter is thus dened as the ratio between sur-
face tension and saturated pressure dierence between uid
and vapour.
Based on this evaluation of droplet diameter, the separa-
tion chamber diameter is determined by applying Newtons
second law to a droplet submitted to earths gravity, buoy-
ancy and friction force. This results in Eq. (21). As for the
droplets diameter denition, the subscript 1 should be
replaced by 2 for the low pressure stage.
d
sc1

72 l
f
q
v1
g q
f
p d
2
dr1
q
v1
q
f1

21
This denition of separation chamber diameters ensures
that droplets are not dragged along the low pressure stage.
For the thickness estimate, the French CODAP (Pres-
sure Equipment Directives) imposes instructions for calcu-
lation. These instructions are expressed in Eq. (22), where
P is the enclosure pressure, D is the enclosure diameter, f
is the calculation design stress and z is the welding coe-
cient. For stainless steel, which is the predominant material
used, z is equal to 0.85 and f is about 207 10
6
N m
2
.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 501
e P
P D
2 f z P
22
The heating tank designed to heat and provide uid in
the instrumented pilot unit is replaced by a buer tank in
the industrialised version of the process. This buer tank,
devoted to transit passage of uid, has to compensate for
any break in uid supply. Thus, its minimum volume
V
BT
has been estimated based on a relation between the ll-
ing time, the break time corresponding to the desired
autonomy of the system, the uid mass ow rates at the
inlet and outlet of the tank and its density, as shown in
Eq. (23).
V
BT
t
break
q
fo

1
q
f
_ _
t
filling
q
fi
q
fo

1
q
f
_ _
23
The tank is designed to be of cylindrical form, with a
conical bottom with a h angle taper. The volume of this
part of the tank is evaluated in Eq. (24), where subscript
BT refers to the buer tank.
V
conical bottom

1
3
p
d
BT
2
_ _
2

d
BT
2 tan
h
2
24
The tank height is estimated on the basis of the tank
diameter, which is chosen according to encumbrance
requirements, in Eq. (25).
H
BT

V
BT
p
d
2
BT
4
25
To guaranty the performances of the system, the inlet
temperature of the mixture has to be as constant as possi-
ble, and the buer tank is subjected to environmental con-
ditions. Moreover, it should be checked that there is not
any danger of getting burnt for the user, since mixture is
supposed to enter the system at a temperature close to
353.15 K. As a consequence, insulation is envisaged to
avoid temperature variations. To perform this insulation,
a rockwool layer will be put in between the buer tank
and an aluminium sheet.
To estimate this insulation thickness, a heat transfer
analysis was carried out, considering that the full tank is
submitted to forced convection and radiation. Eq. (26) dis-
plays the insulation thickness calculation, where subscripts
BT, ins and al respectively refers to the buer tank, the
insulation layer and the aluminium sheet. T
alu
is the desired
temperature for the external aluminium sheet and T
air
is
the air temperature around the buer tank. For design cal-
culations, the air temperature is expected to be of 288.15 K
degrees, with a heat transfer coecient of 35.8 W m
2
K.
e
ins

e
ins
k
ins

e
BT
k
BT

e
alu
k
alu
_ _
p H
BT
T
fi
T
alu


_
h
air

p2:d
BT
2:e
BT
H
BT
2
_ _
r e
p
2
2:d
BT
2:e
BT
H
BT
T
4
alu
T
4
air

_ _

p H
BT
T
fi
T
alu

e
ins
k
ins

e
BT
k
BT

e
alu
k
alu
d
BT
2:e
BT
e
alu

_ __
26
In addition to these gauging constraints, the two-stage
ash evaporation process must respect mass limitation,
since it has to be transportable on a truck. Its mass and sur-
face area should thus be limited by the gross vehicle weight
rating of this truck. Calculations were based on platform
body straight truck limitations. Such trucks are limited to
a 19,00026,000 kg gross weight rating. Their platform
body surface is of 6.4 2.5 = 16 m
2
. The total mass, eval-
uated by adding together the masses of all constituent com-
ponents and the global surface of the apparatus have to be
under the above limitations. These requirements are
expressed in the design model as inequalities.
4.5. Ecodesign criteria
Facing the emergence of eco-design in industry, it seems
obvious to consider the environmental impact of the indus-
trialized process in the design model.
Indeed, the formalisation methodology used allows inte-
grating simultaneously data such as eco-design criteria and
physical parameters, the laws of physics, etc. It is thus pos-
sible to inuence the design of the system before the
detailed design step. This could be of great interest for
deadlines and cost. Moreover, ash evaporators are sys-
tems which use a lot of energy, materials and ows, and
it appears to be important to quantify energy, materials
and ow consumption for the two-stage industrialized
process.
The energy consumption evaluation was calculated
through the electrical power consumption of the system
and through the evaluation of the cooling powers for each
separation stage.
Electrical energy is the only energy consumed by the
system. Only the vacuum pump (VP), the extraction
pump (EP), the condensates pump (CP) and the mixer
(Mix) use electrical energy. Thus, the total electrical con-
sumption calculation for the whole system was calculated
by adding their power consumption Po increased by the
operation time of the system t
ope
, expressed in hours.
Eq. (27) shows the electrical consumption of the system,
expressed in kWh.
Co
EE
t
ope
Po
VP
Po
EP
Po
CP
Po
Mix
27
The cooling powers in the separation chamber corre-
spond to the heat transferred to the vapour during vapori-
sation. Thus, they are evaluated, for each separation stage,
through a heat transfer balance between the inlet and the
outlet. Eqs. (28) and (29) give these cooling powers Po
sc1
and Po
sc2
for the high pressure and the low pressure stage,
respectively.
Po
sc1
q
fi
Cp
fi
T
fi
Cp
f1
T
v1
28
Po
sc2
q
f1!2
Cp
f1
T
v1
Cp
f2
T
v2
29
The presented industrial two-stage ash evaporation
process uses water as coolant in the condensers. In order
to evaluate the global eciency of the system regarding
502 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
ow consumption and production, its ow consumption
was modelled on the ratio between water consumption
and vapour heat powers for each condenser. If pressure
losses are put aside, these vapour heat powers are equal
to those in the separation stages, since the assumption of
saturated vapour is also taken into account.
The ow consumption is thus evaluated as shown in
Eq. (30), and expressed in Litre per hour per thermal
kilowatts.
Co
wa

2
i1
q
cli
q
vi
Dh
v
T
vi
X
vei
X
vsi

2
i1
q
cli
q
fi
Cp
fi
T
fi
Cp
fo
T
f0

30
For the translation of material consumption environ-
mental impact, our choice was made on the EcoIndica-
teur99 methodology (PreConsultants, 2000a, 2000b,
2001), which is a single score methodology.
This method makes it possible to represent the envi-
ronmental impact of a material or operation by quanti-
ed indicators. These indicators are then standardized,
balanced and added in order to reach to a single score
characterizing the global impact. The EcoIndicateur99
methodology thus enables to take into account the envi-
ronmental impact related to the manufacture of a system
in terms of resources (mineral fossils, fuels, ground used),
ecosystem quality (acidication, toxicity) and human
health (climatic changes, reduction of the ozone layer)
by inventorying the consumed emissions and resources,
as seen in Fig. 9.
For the ash evaporation process, material consumption
is mainly due to stainless steel tanks production. As a con-
sequence, we used the EcoIndicator99 methodology to
evaluate the single score linked to this production, and
reached a global single score of 13237.2 points per ton of
stainless steel. In order to reach the process material
impact, expressed in points, this global score is multiplied
by the total mass of stainless steel used for the system pro-
duction in the design model.
For the selection of the obtained solutions, in the fourth
part of the formalisation methodology presented in para-
graph 4.1, a reference value of 1000 points is considered.
The closer material impact will be to this reference value,
the greater the solution will be from an environmental
point of view.
4.6. Costs models
In the case of an industrialised system, economical con-
siderations are of great interest. We chose to include these
considerations in the design model through production cost
and cost-in-use.
The global production cost of the system includes mate-
rial acquisition and forming costs (raw materials) added to
the purchase of standard components and pipes, as shown
in Eq. (31). Material acquisition and forming are for the
buer tank (BT) and the separation chamber (sc). Standard
components are the mist Eliminator (ME), the condensers
(cdsr) and the vacuum, condensates and extraction pumps
(VP, CP, EP). The additional coecient has been consid-
ered based on the work of Marechal (2003) and Bolliger,
Favrat and Marechal (2005).
C
Sys
3:1 C
BT
C
sc
C
ME
C
cdsr1
C
cdsr2
C
Pipes
C
VP
C
CP
C
EP
31
The estimated cost for material acquisition and forming
were calculated according to the required weight of mate-
rial for each component concerned, based on prices accord-
ing to weight data given by industrial suppliers.
For the stainless steel, prices from ALMET supplier
were used, as shown in Eq. (32).
C
Stainless steel
4:7446 Weight 4:7715 32
Eq. (33) displays the buer tank aluminium protection
sheet cost correlation, according to METAPROFIL SUD
manufacturers prices.
C
Aluminium
20:09 :Weight
0:6605
33
For the pipes estimation cost, data from GALVA SER-
VICES was used, linking prices to the required pipes
weight, as shown in Eq. (34).
C
pipes
0:5048 Weight 0:0253 34
The evaluation of the cost of the rockwool insulation
layer was made considering a range of prices to the heat
transfer area, from dierent manufacturers:
11 A
ins
6 C
rockwool
6 18 S
ins
35
The forming cost was also modelled according to mate-
rial weight. This correlation leads to a simplistic approxi-
mation of forming cost, since it does not considerer the
dierent operations (welding, turning, etc.) specically.
Fig. 9. Principle of the EcoIndicateur99 methodology.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 503
However, this correlation, expressed in Eq. (36), was
obtained and based on information given by ENSAM-Bor-
deaux forming workshop.
C
Forming
11:287 Weight 427:53 36
The mixer price and the vacuum, condensates and extrac-
tion pumps prices were supplied by manufacturers.
Therefore, the production cost of a component can be
evaluated by adding together material and forming costs,
relative to its unloaded weight.
As an example, Eq. (37) highlights the production cost
for the buer tank which is calculated by adding together
the purchasing cost of the mixer (Mixer), the material
acquisition and forming costs for the stainless steel buer
tank (BT), the purchase cost of the rockwool insulation
layer (IL) and the acquisition and forming costs of the alu-
minium external sheet (AES).
C
BT
C
Mixer
CBT
Stainless Steel
CIL
Rockwool
CAES
Aluminium
CBT AES
Forming
37
For the condensers and the mist eliminator, relations
between purchasing cost and heat transfer area and inlet
area were used, respectively, as shown in Eqs. (38) and (39).
C
cdsr
3789:48 A
0:4678
cdsr
38
C
ME
3448:11 S
0:7817
ME
39
Cost-in-use was evaluated through expenditure due to
water and electrical consumption, as shown in Eqs. (40)
(42), where C
3
m
is the price of a cubic meter of water and
C
kWh
is the price of a kilowatt of electrical energy con-
sumed in 1 h.
C
in use
C
water
C
EE
40
C
water
C
m
3 q
cl1
q
cl2
t
ope
41
C
EE
C
kWh
Co
EE
C
kWh
t
ope
Po
VP
Po
EP
Po
CP
Po
Mix
42
5. Towards process improvement
The global model of the two-stage ash evaporation
process is thus composed of physical, technical, economical
and environmental constraints. This model, expressed in a
Constraint Satisfaction Problem formalism has been solved
with the operational requirements presented in the Appen-
dix. Several solutions were obtained, corresponding to dif-
ferent system congurations. In order to allow a ranking
between all these solutions, indexes and indicators were
considered.
Quality indexes, dened as the ratio of one criterion
increased by reliability to the global system cost, as
expressed in Eq. (43), suggest a selection among all solu-
tions according to the specic criterion considered. These
quality indexes are used to evaluate the entire solution
space and compare all the solutions.
QI
i

Cr
i
reliability
cost
43
Performance indicators, outlined by the ratio of one cri-
terion to a reference value of the same criterion, as pre-
sented in Eq. (44), propose to evaluate the distance
between one solution and the reference value on the base
of the criterion considered. These indicators thus allow
the selection of the most suitable solution for one criterion.
PI
i

Cr
i
Cr
ref
44
Starting from this denition of quality indexes and per-
formance indicators, a comparison between the solutions
obtained through the two-stage ash evaporation process
solving was made, based on an analysis of cooling power,
water and electrical consumptions and an evaluation of
the eco-design criteria chosen.
A quality index related to the cooling power and perfor-
mance indicators related to water and electrical energy con-
sumption and to the single EcoIndicator99 score obtained
for the process have thus been outlined.
Fig. 10 shows the evolution of the cooling power of the
system towards its quality index, presented in Eq. (45).
QI
coolingpower

Po
sc1
Po
sc2
C
in use
C
sys
45
We can notice that this evolution is quite linear. More-
over, the quality index is multiplied by seven with the
increase of the cooling power. Regarding Eqs. (28) and
(29) that give the denition of cooling powers, we can see
that these parameters are proportional to the temperature
dierence in the separation stage. Thus, the higher this tem-
perature dierence will be, the greater the cooling power,
resulting in a great quality index. This means that it is pref-
erable to initially heat the uid as much as possible to make
the global system thermal eciency protable. Since the
quality index is an expression of the link between the ther-
mal eciency and the global cost of the system, we can also
conclude that a great heating temperature will result in a
better nancial amortization. However, such recommenda-
tion should be used with precautions for biological uids
such as wine.
0
50
100
150
200
250
300
350
400
450
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035
QI
cooling Power
[kW/ ]
C
o
o
l
i
n
g

P
o
w
e
r

[
k
W
]
Fig. 10. Evolution of cooling power compared to quality index for model
results.
504 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
The performance index related to electrical energy con-
sumption is presented in Eq. (46).
PI
Co
EE

Co
EE
Co
EE
ref
46
It shows the links between the electrical consumption of the
system with a reference value. This reference value was
established with a measurement campaign carried out from
February to March 1995 on a ash evaporation system
from the GEYSERS Company, used for the RABOUR-
DIN Industries Company industrial wastewater treatment
(EDF Industry technical reports web site). This reference
value rises to 230 kWh per ton of evaporated water. Since
the electrical energy consumption has been evaluated con-
sidering the components powers which are xed, the previ-
ous performance index remains the same for all solutions.
It is about 0.60.8%. Thus, we can conclude that the
two-stage ash evaporation process is quite ecient with
electrical consumption, compared to competing ash evap-
oration processes.
The water consumption performance index is presented
in Eq. (47). It shows the ratio of the system water consump-
tion to a reference value of water consumption. This refer-
ence value corresponds to the average cooling water
consumption in single passage circuit condensers of indus-
trial cooling systems in an agro-alimentary eld (Bertrand,
2004). This value was estimated at about 1.433 L/
min kW
th
.
PI
Cowa

Co
wa
Co
wa
ref
47
Fig. 11 presents the evolution of this performance index
compared to high and low pressure condenser coolant mass
ows. In both cases, we note a linear evolution. Variations
observed for the high pressure stage are lower than those
for the low pressure stage. Starting from the denition of
the water consumption (see Eq. (30)), which is a ratio
between coolant mass ows and cooling powers in the sep-
aration chamber, we can see that this is because the cooling
power created in the high pressure stage is greater than that
in the low pressure stage. This means that the global water
consumption of the system should be reduced by control-
ling the low pressure condenser coolant ow. Moreover,
we could also assume that the low pressure condenser size
could be reduced. This conclusion is conrmed by experi-
ments observations of vapour ows, seen in Section 4.3.
The performance index related to the EcoIndicator 99 is
dened through Eq. (48), where EI
sys
is the single score
obtained for the ash evaporation process.
PI
EI99

EI
sys
EI
ref
48
The reference value EI
ref
is based on data provided in
the appendix of the methodology (PreConsultants, 2001).
This value is of 1000 points. As seen previously, the most
suitable solutions, from an environmental point of view,
are those with the smaller PI
EI99
performance index.
Fig. 12 shows the evolution of this performance index
compared to vapour temperature in the high pressure
stage. This temperature corresponds to the uid cooling
temperature at the low pressure stage inlet. We observe
that maximum values are obtained for higher vapour tem-
peratures, with a great increase for small temperature
variations. This evolution indicates that for a high inlet
uid temperature, the ash observed will be more violent,
and droplets of improved size may be formed. In the low
pressure stage, diameter calculations were made so that
droplets could not been carried along with the vapour.
This means that for higher vapour temperatures in the
high pressure stage, the diameter of the low pressure stage
will be increased, leading to an increase in material con-

0
50
100
150
200
250
300
350
400
450
500
150 200 250 300 350 400 450 500
Q
cl1
[L/min]
P
I
C
O
w
a
0
50
100
150
200
250
300
350
400
450
500
0 50 100 150 200 250
Q
cl2
[L/min]
P
I
C
O
w
a
Fig. 11. Evolution of water consumption performance Indexes for cooling
powers compared to coolant ows for high pressure and low pressure
condensers.
0
5
10
15
20
25
30
35
40
312 314 316 318 320 322 324
T
vHP
[K]
I
P
E
I
S
y
s
Fig. 12. Evolution of the performance indicator related to the environ-
mental impact compared to vapour temperature in the high pressure stage
of the separation chamber.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 505
sumption, and thus in the global EcoIndicator 99 single
score value.
Regarding those results and observations, we can con-
clude that the ash evaporation process will be optimized
on thermal and economical grounds with the increase of
the heating temperature of the uid. Nevertheless, the cool-
ing temperature in the high pressure stage must stay con-
stant and as low as possible, so that the temperature
dierence in the low pressure stage remains at about
20 C max.
Based on these conclusions, a simulation test was carried
out with the following imposed conditions: T

= 343.15 K,
q

= 2.78 kg s
1
, q
cl1
= 550 L min
1
, q
cl2
= 300 L min
1
,
T
cl1,2
= 283.15 K, T
fo
= 303.15 K. Table 1 gives the encum-
brance, mass, costs and ecocriteria values obtained for this
test.
This results in the validation of the detailed design of the
industrialized two-stage jet falling ash evaporator proto-
type, illustrated in Fig. 13. It can be found, in Ho Ho
Kon Tiat, Sebastian, and Nadeau (2005) a comparison
between this prototype and two other competing cooling
processes regarding power eciency, weight and residence
time criteria.
6. Conclusion
We introduced the ash evaporation process. This pro-
cess is used in the wine producing eld for vintage cooling
after heating before storage.
Although promising because it allows grape cooling,
vintage concentration and improvement of the wine quality
all in one device, the development of this process remains
limited because of the low eciency of the thermal
exchanges and of the large size of the installations.
Table 1
Encumbrance, mass, costs and ecocriteria values obtained for a simulation
test
Operation conditions Outlet values
T

= 343.15 K
q

= 2.78 kg s
1
q
cl1
= 550 L min
1
q
cl2
= 300 L min
1
T
cl1,2
= 283.15 K
T
fo
= 303.15 K
Separation chamber
d
sc1
= 0.65 m e
1
= 0.0029 m
d
sc2
= 1.51 m e
2
= 0.005 m
q
v1
= 0.15 kg s
1
q
v2
= 0.045 kg s
1
Buer tank
V
BT
= 1.36 m
3
d
BT
= 1.3 m
e
ins
= 0.016 m
Condensation unit
k
cd1
= 1307 W m
2
K
k
cd2
= 917 W m
2
K
Entire system
Co
wa
= 0.911 L min
1
kW
th
C
water
= [7; 25] h
1
Co
EE
= 9.92 kWh
C
EE
= 1.05 h
1
C
Sys
= 140810
m
Sys
= 1795.5 kg
EI
Sys
= 22237 pts
Fig. 13. Illustration of the industrial high performance two-stage jet falling ash evaporator prototype with the Catia V.5 software.
506 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508
An innovative two-stage ash evaporation process was
presented that improves the system compactness and ther-
mal eciency, through the use of a new type of condens-
ers. The modelling of an industrialized version of this
process was carried out for the preliminary design stage,
in order to support the decision making process and thus
optimising the design process. This has been achieved
through a formalisation methodology that uses Constraint
Satisfaction Problem solving techniques to deal with sev-
eral types of data. The modelling of the physical behav-
iour of the system was validated through an
experimental analysis carried out on an instrumented pilot
unit. Technical, economical and environmental criteria
were also considered.
Results, which correspond to dierent congurations of
the same process, are presented and compared based on
quality indexes and performance indicators that translate
technical, economical and environmental aspects. The way
to improve the system performances is thus highlighted.
Appendix A. Functional requirements
Inlet parameters:
T
fi
f343:15; 353:15; 363:15g K
q
fi
f1:388; 2:78g kg s
1

q
cl1
f170; 265; 360; 455; 550g L min
1

q
cl2
f40; 105; 170; 235; 300g L min
1

T
cl1i
T
cl2i
f283:15; 293:15; 303:15g K
Outlet parameters:
T
fo
303:15 K
References
Ageron, D., Escudier, J. L., Abbal, Ph., & Moutonet, M. (1995).
Pretraitement des raisins par ash detente sous vide pousse. Revue
Franc aise dnologie, 153, 5054.
Benhamou, F., & Older, W. (1997). Applying interval arithmetic to real,
integer and Boolean constraints. Journal of Logic Programming, 32(1),
124.
Berliner, C., & Brimson, J. (1988). Cost management for todays advanced
manufacturing. Boston: Harvard Business Press.
Bertrand, A. (2004). La production de froid dans le milieu agro-
alimentaire, industriel et tertiaire, Impacts environnementaux et
applications techniques, Les cahiers de la production propre, no. 4,
Centre de Recherche Publique Henri Tudor, Centre de Ressources des
Technologies pour lEnvironnement (CRTE), Luxembourg. http://
www.crte.lu/mmp/online/website/content/clean/guidelines/index_EN.
html or http://www.crte.lu/mmp/online/website/function/documenta-
tion/le_4998/crte_cpp04_2006.pdf.
Bolliger, R., Favrat, D., & Marechal, F. (2005). Advanced power plant
design methodology using process integration and multi-objective
thermo-economic optimisation. In ECOS 2005, 18th international
conference on eciency, cost, optimization, simulation and environ-
mental impact of energy systems (pp. 777784). 2.
Bouchama, A., Sebastian, P., & Nadeau, J.-P. (2003). Flash evaporation:
modelling and constraint formulation. Trans IChemE, Chemical
Engineering Research and Design Part A, 81, 12501258.
Celotti, E., & Rebecca, S. (1998). Experiences recentes de thermo-
maceration des raisins rouges. Revue des nologues, 87, 2328.
EDF Industry technical reports web site. http://www.trs-online.com/
09.php.
Escudier, J. L., Mikolajczak, M., Martin, M., Samson, A., Thibert, D.,
Boulet, J. C. (1995). Interets du. traitement des raisins par unvide pousse,
Rapport contrat TRIAL-ONIVINS-Region. INRA Pech Rouge.
Escudier, J. L., Mikolajczak, M., & Moutounet, M. (1998). Pre-traitement
des raisins par ash detente sous vide et caracteristiques des vins. J. Int.
Sciences de la Vigne et du Vin, hors se rie Traitements Physiques des
Mouts et des Vin, 105110.
Fischer, X., Nadeau, J.-P., Sebastian, P. (2002). Decision support in
integrated mechanical design through qualitative constraint, IDMME
2000 selected paper book. Kluwer Academic Publishers Group (pp.
3542).
Jaulin, L., Kieer, M., Didrit, O., & Walter, E. (2001). Applied interval
analysis. Springer-Verlag.
Kapp France web site. http://www.kappfrance.fr/anglais/Acadre1.html.
Ho Kon Tiat, V., Sebastian, P., Nadeau, J.-P. (2005). Decision support
system for high performance ash evaporators. In Proceedings of
the international conference on heat transfer in components and
systems for sustainable energy technologies (HEATSET), Grenoble,
France.
Kuhn, S. Z., Schrock, V. E., & Peterson, P. F. (1997). An investigation of
condensation from steam-gas mixtures owing downward inside a
vertical tube. Nuclear Engineering and Design, 177, 5369.
Marechal, F. (2003). Denition du proble`me thermo-economique: esti-
mation des investissements et calcul du cou t operatoire pour les
syste`mes industriels, calcul des emissions et cou ts des emissions, notes
de cours, Laboratoire dEnergetique Industrielle, Ecole Polytechnique
Federale de Lausanne, Canada. http://personnes.ep.ch/francois.
marechal.
Nadeau, J.-P., Pailhes, J., (2006). Integration de linnovation et des
sensations utilisateur en conception preliminaire par le biais de
lanalyse fonctionnelle. Ingenierie de la conception et cycle de vie
du produit, Chapitre 2, Traite IC2, ISBN: 2-7462-1214-5, Herme`s,
Paris.
Nadeau, J.-P., Pailhes, J., & Scaravetti, D. (2006). Des outils de lanalyse
fonctionnelle vers la creativite technique. International Journal Of
Design and Innovation Research, 3(34), 87106.
Pahl, G., Beitz, W. (1996). Engineering design A systematic approach
(2nd ed.). Londres: Springer Verlag. http://www.pre.nl/eco-indica-
tor99/ei99-reports.htm.
PreConsultants, b.v., (2000a). The eco indicator 99: A damage oriented
method for life cycle impact assessment, methodology report, Ministry
of Housing, Spatial planning and the Environment, Communications
directorate, The Netherlands.
PreConsultants, b.v., (2000b). Eco Indicator 99: a damage oriented
method for Life Cycle Impact Assessment, report, Manual for
designers, Spatial planning and the Environment, Communications
directorate, The Netherlands.
PreConsultants, b.v., (2001). The Eco Indicator 99: A damage oriented
method for Life Cycle Impact Assessment, Methodology annex, Third
Edition, Ministry of Housing, Spatial planning and the Environment,
Communications directorate, The Netherlands.
Scaravetti, D., Nadeau, J.-P., Sebastian, P. (2003). Structuring
functions and constraints formulation for enhanced embodiment
design. international CIRP Design Seminar, Grenoble, France (12
14 May).
Scaravetti, D., Nadeau, J.-P., Pailhe`s, J., Sebastian, P. (2005). Structuring
of embodiment design problem based on the product lifecycle, Int. J.
Product Development (Vol. 2(1/2), pp. 4770). Gene`ve, Suisse:
Indersciences Ed.
Scaravetti, D., Pailhe`s, J., Nadeau, J.-P., & Sebastian, P. (2004). Aided
decision-making for an embodiment design problem. In Alan Bramley,
Daniel Brissaud, Daniel Coutellier, & Chris McMahon (Eds.),
Advances in integrated design and manufacturing in mechanical
engineering. Kluwer Academic Press.
V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508 507
Sebastian, P., & Nadeau, J.-P. (2002). Experiments and modeling of falling
jet ash evaporators for vintage treatment. International Journal of
Thermal Sciences, 41, 269280.
Thornton, A. (1996). The use of constraint-based design knowledge to
improve the search for feasible designs. Engineering Application of
Articial Intelligence, 9, 393402.
Vernat, Y., Fischer, X., Nadeau, J. P., Sebastian, P. (2004). Strategy for
model formalization in preliminary design. In 8th World multi-
conference on systemics, cybernetics and informatics, SCI 2004, 1821
July 2004, Orlando, Florida, USA.
Vinihor web site. http://www.onivins.fr/Vin/VigneVin/VinifIndex.asp.
Zimmer, L., Anglada, A., Christie, M., Granvilliers, L. (2004).
Constraint explorer: A modelling and sizing tool for engineering
design, invited session on metamodelling and constraint based
problem solving for embodiment design support systems in SCI,
Orlando.
508 V. Ho Kon Tiat et al. / Journal of Food Engineering 85 (2008) 491508

Vous aimerez peut-être aussi