-
Computers and Chemical Engineering 29 (2005) 14911505
Dynamic modelling and simulation oftallitos a,e Miis, 1045
Estar
2005
Abstract
The aim o rocessatmospheric conditions on the system. For this
main goal it was built a dynamic model, which includes the
cogeneration system, the plateheat exchangers and the salt
production unit.
The cogeneration system was modelled and analysed in GateCycle
5.34.0.r. and the interface variables were used as input of the
dynamicmodel of the remaining integrated process. This model was
developed and exploited through gPROMS 2.3.
Some parsystem.
The best seach situatioto optimisehaving a tu
The efficiHowever, thof 7080%, 2005 Else
Keywords: D
1. Introdu
Processies as a newthe efficienprocess anoptimal strplex
plantsmaterials asystematic
Up to 1were consi
CorresponE-mail ad
0098-1354/$doi:10.1016/jticular issues (start-up, scheduling and
atmospheric conditions) were investigated to forecast the
performance of the integrated
tart-up conditions were established. Several atmospheric
conditions were studied and the minimum number of ponds required
forn was calculated. The scheduling of the evaporation ponds in
operation was also investigated to enhance the salt production
and
the salt harvesting. The process simulation indicated that it is
better to work with the corresponding minimum number of ponds,rbo
pond that receives a larger quantity of heated brine.ency of the
cogeneration system (thermal plus electric power divided by natural
gas consumption) was approximately 92%.e global process efficiency
(accounting for energy losses in the evaporation step of the salt
production process) was in the rangedepending on the atmospheric
and operational conditions considered.vier Ltd. All rights
reserved.
ynamic modelling; Optimisation; Process integration; Industrial
case study; Cogeneration; gPROMS; Crystallization process
ction
integration (PI) emerged in the decade of eight-area in Chemical
Engineering with emphasis on
t use of primary energy by analysing the wholed not the stand
alone units to find the best andeams heat exchanger network.
Currently the com-are characterized by the existence of recycles
of
nd energy making necessary their integration in aand rational
way (Linnhoff et al., 1982).990, Process Synthesis and Process
Integrationdered separately although complementary activi-
ding author.dress: [email protected] (H.A. Matos).
ties. The original definition in 1995 from International En-ergy
Agency (IEA) states Process Integration as: System-atic and General
Methods for Designing Integrated Pro-duction Systems, ranging from
Individual Processes to To-tal Sites, with special emphasis on
theEfficient Use ofEnergy and reducing Environmental Effects
(Gundersen,2002).
However, in recent years the borderline between these
twoactivities has practically disappeared (Smith, 1995).
There-fore, Process Integration appears now as an optimal
integra-tion of different units in process system and this goal is
effi-ciently achieved using process simulation that was a
formermain tool for analysis and synthesis. The power of
modernsimulation techniques enables the study of complex
processesclose to the real situation.
see front matter 2005 Elsevier Ltd. All rights
reserved..compchemeng.2005.02.015integrated with a salt
recrysRaquel Durana Moita a, Henrique A. Ma
Clemente Pedro Nunes a, Jorga Instituto Superior TecnicoDEQ, Av.
Rovisco Pa
b Quimigal, Quinta da Industria, 3864-75Available online 9
March
f this study is to optimise a Portuguese industrial three
integrated pa cogeneration systemzation process, Cristina Fernandes
a,
guel Prior b
9-001 Lisboa, Portugalreja, Portugal
system, by studying the effect of some operational and
-
1492 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
Nomenclature
A area (m2)Cp heat capacity (MJ/(kg C))Cp,air air heat capacity
(J/(kg K))Econv energy loss by convection (MJ/(m2 h))Eevap enthalpy
of the vapor water stream (MJ/(m2 h))ENaClpp enthalpy of the salt
precipitated stream
(MJ/(m2 h))Erad energy loss by radiation (MJ/(m2 h))Esolar net
solar energy that is effectively absorbed
through the brine (MJ/(m2 h))Ethermal effective thermal power
(MJ/h)fEntP,i fraction of the brine flow rate QHE that is sent
to each pond ifExitP,i fraction of the brine flow rate
(Qchannel) that
leaves each pond iFM salt mass fraction (%)hwater 1 enthalpy of
the inlet water stream of the
economizer at temperature T1 (kJ/kg)hwater 2 enthalpy of the
outlet water stream of the
economizer at temperature T2 (kJ/kg)H enthalpy of the liquid
brine stream (MJ/h)HL,Tap,Ei enthalpy of the liquid brine stream
entering
in Ei through the tap, per unit length of theelement (MJ/(h
m))
HT total enthalpy of the accumulated brine(MJ/m2)
Kair air conductivity (W/(m K))Lc characteristic length (m)Mevap
water evaporation rate (kg/(m2 h))MNaclpp salt precipitation rate
(kg/(m2 h))MNaclpp,total total mass of salt precipitated (kg)Mwater
mass flow rate of the water circulating in the
economizer (kg/h)n number of moles (moles)P water vapour partial
pressure at temperature T
(Pa)Patm atmospheric pressure (Pa)PGateCycle thermal power
determined through
GateCycle (MJ/h)Q brine volumetric flow rate (m3/h)QL,Tap,Ei
brine volumetric flow rate entering in Ei
through the pond tap, per unit length of theelement (m3/(h
m))
S salinity (%)Sol NaCl salt solubility in the water
(g NaCl/100g H2O)T brine temperature (C)T1 temperature of the
cold water entering in the
economizer (C)T2 temperature of the heated water leaving
from
the economizer (C)
Tair Tdew-poinTskyUHE
VwindWXXmZ
Greek syHdiss
Tlm
evapair
0
air
Subscripconv
EiEntP,ievapExitP,iFHEradsatSolarTap
DomainsLt
AbbreviaCSEGPEPHE
Severalused in theactivities toof the indu2002).air temperature
( C)t dew-point temperature (C)sky temperature (C)service overall
heat transfer coefficient of theplate heat exchanger (MJ/(m2 C
h))wind velocity (m/s)
pond width (m)salt concentration in the brine solution
(kg/m3)mole fraction of the waterbrine level (m)
mbolsheat of dissolution of the salt in the water(MJ/kg
NaCl)logarithmic mean temperature difference (C)surface
emissivityair relative humiditywater latent heat of vaporization
(MJ/kg)air viscosity (N/(s m2))brine density (kg/m3)auxiliary
variable used in brine densitycalculations (kg/m3)air density
(kg/m3)StefanBoltzmann constant (W/(m2 K4))
ts and superscriptsconvectioninfinitesimal element i in each
pondbrine stream entering in each pond ievaporationbrine stream
exiting from each pond ifresh treated brineplate heat
exchangerradiationsaturationsolar energypond tap for the heated
brine entrance
axial (m)time (h)
tionscogeneration system efficiencyglobal process
efficiencyplate heat exchanger
powerful systematic methodologies have beenlast two decades to
support Process Integrationachieve a parametric and structural
optimisation
strial sites (Relvas, Fernandes, Matos, & Nunes,
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1493
The main goal of this study is to build a model of anindustrial
integrated system including three processes, andthrough simulation
identify the best operational conditionsto maximize its global
energy efficiency and to minimize theenvironmental impact, reducing
the primary energy supplyand the raw materials usage. This
industrial site is located atCarrico, Pombal (Portugal) and
includes: a salt cavern construction process to build a natural
gas
reservoir (owned by Transgas); a gas turbine cogeneration system
(owned by Galp Power); a salt recrystallization process (owned by
Renoeste).
The three processes considered as separated units are nei-ther
efficient nor feasible. The integration of these three inde-pendent
units, as shown in Fig. 1, improves the global systemefficiency.
The leaching programme to construct the cavernsfor a future natural
gas storage generates the brine needed tofeed the salt
recrystallization ponds, minimizing the environ-mental damage.
Those caverns will allow a strategic storagefor the Portuguese
Natural Gas supply system, providing abuffer for eventual future
supply fluctuations on the nationalsupply and consumption. The
cogeneration system providesthe electricport
towardassociatedprocess thrface betweeefficiency.Renoeste
iindustrial smitted for p
2. Framew
A dynamthe cogene
rated sy
developed to determine the best conditions to maximize theglobal
energy efficiency. This framework allows a very widerange of
studies such as:
Analyse the behaviour and feasibility of the integrated
pro-cesses, by verifying if crucial variables values (such asstream
temperatures) are within the advisable operationalintervals, and
determine the global energy efficiency.
Study by simulation several scenarios of operating condi-tions
such as start up or ponds working scheduling.
Study the effect of different atmospheric cases on the
per-formance of the integrated process.
To obtain a global description of this integrated system, itwas
considered mainly the following procedure and tools:
1. modelling and simulation of the cogeneration systemthrough
GateCycle 5.34.0.r of the GE Enter Software;
2. determination of a thermal power correlation,
throughTableCurve 3D from SPSS Inc., using data from cogen-eration
system;
3. design of a dynamic distributed model of the whole
in-tegrated system (cogeneration, plate heat exchangers and
lt prod
sing dvariabg otheoncen
al effi
nalysi
e cogsts ofts assour se.5 m2,al power to satisfy the processes
needs and to ex-s the regional net the remaining production.
Its
thermal energy is used in the salt recrystallizationough a set
of plate heat exchangers as an inter-n the two processes, improving
the global energyThe pure salt (NaCl) produced in the ponds bys the
main raw material to Quimigals chemicalite at Estarreja (Aveiro)
(Moita et al., 2004, sub-ublication).
ork developed
ic model of the whole process, which includesration system and
the salt production unit, was
Fig. 1. The three processes integ
sa
Ustateamon
and ctherm
3. A
Thconsiand iinto f2302stem.
uction process) via gPROMS 2.3 of the PSE Ltd.ifferent
atmospheric and operational conditions asles it is possible to
obtain as output of the model,r variables, the pond level, the
brine temperature
tration profiles, the salt production and the processciency.
s of the cogeneration system
eneration system installed at the industrial sitea RollsRoyce
natural gas turbine (RB211T DLE)ociated economizer. The economizer
is dividedrial set of tubes, each having a superficial area ofin
which circulates the water that will be heated
-
1494 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
eCycle
with the eximately 50
The cogGateCycleallows to meral energyeration sysgas turbine
Fig. 2 shtem in this svalues for t
This moables, such
rculatrmancg. 3 illrmanc
umptiog. 4 hir enteable aom thhe air
FFig. 2. Model structure of the cogeneration system in Gat
haust turbine gases (leaving the turbine at approx-0
C).eneration system is modelled in the simulator5.34.0.r of the GE
Enter Software. This softwareodel, simulate and predict the
performance of sev-systems, such as combined cycle plants,
cogen-
tems, combined heat-and-power plants, advanceds cycles, etc.
(GateCycle Software Website, 2004).ows the model structure of the
cogeneration sys-imulator, the general required input and the
output
ter ciperfo
Fiperfocons
Fiwateavail
Frthat the nominal operating conditions.del allows the study of
the influence of some vari-as the air and natural gas temperatures
or the wa-
system respcauses a re
around 3%
ig. 3. Variation of net electric power, natural gas consumption,
exhaust gases temp5.34.0.r (nominal conditions).
ion conditions, on the gas turbine and economizere.ustrates the
effect of air temperature on gas turbinee, namely the net electric
power, natural gas heatn, exhaust gases temperature and mass flow
rate.ghlights the influence of air temperature and coldring into
the economizer on the thermal powernd on the exit gases
temperature.e analysis of these figures it is possible to
concludetemperature strongly influences the cogeneration
onse: an increase of 10 C in Tair (above 15 C)
duction of almost 10% in the electric power andin the thermal
power. The thermal power avail-
erature and mass flow rate with air temperature.
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1495
Fig. 4. Variation of thermal power and exit gases temperature
with air temperature and with cold water temperature.
able is also very dependent on the temperature of the coldwater
entering in the economizer (T1): an increase of 10 Cin T1
leadsFurthermoraround 15
Fig. 5 shthe economwater tempadvisable owater in th
Throughthat air husystem andincreases that the site (
It was apower as atemperaturwhich wasin the next
4. Dynamic model of the integrated system
e integrated system includes the cogeneration system,our plunit
ins, a fe. This
tions ue glos all th
icalcn of mhermofer, etal mod
Cogen
e cogr corrto a reduction of around 2% in the thermal power.e,
the maximum thermal power is reached at TairC, with the smallest
working T1 value.ows the temperature of the heated water
leavingizer (T2) as a function of Tair and of the cold
erature T1. In this figure it is also represented theperational
temperature interval for the circulatione economizer: T2 = 90 5 C
and T1 = 65 5 C.
simulation in GateCycle it was also concludedmidity does not
have a significant effect on the
that the pre-heating of the natural gas slightlye turbine
efficiency in the temperature range used35 C).lso possible to
obtain a correlation of thermalfunction of cold water temperature
(T1) and air
e (Tair), using TableCurve 3D from SPSS Inc.,included in the
integrated process model discussedsection.
Ththe fThispondFig. 6equa
Thcludephysvatioand ttransment
4.1.
ThpoweFig. 5. Variation of heated water temperature with air
temperature aate heat exchangers and the salt production
unit.volves, mainly, a maximum of six recrystallizationed tank and
a collecting channel, as illustrated insection presents all the
algebraic and differentialsed to model each one of these physical
units.bal dynamic model of the integrated process in-ese
mathematical equations. It is based on known
hemical relationships, which includes the conser-ass and energy,
as well as transport phenomenadynamics relationships (phase
equilibrium, heat
c.), and therefore it can be classified as a funda-el (Bequette,
1998).
eration system
eneration system is modelled via the thermalelation obtained
through the simulated values ofnd with cold water temperature.
-
1496 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
xchang
the cogenedue to genepends on thentering inther heat na
pseudo-sentering wa
PGateCycle =
EThermal =The con
Table 1.
4.2. Plate
The planecting settallization pequation, hterial
balanaccumulati
The PHin a countechanger sin
qual:
rmal =
e the souling(AHE)rence
= (T
Table 1Constants a, b
a
45 078Fig. 6. Integrated process: cogeneration system, plate
e
ration systemEq. (1), which was reduced in 3%ral energy losses
in the pathwayEq. (2). It de-e temperature of the air (Tair) and of
the cold waterthe economizer (T1). In this system there is nei-
or mass accumulation, that is, it is treated as if inteady-state
since there is a time variation on theter stream and atmospheric
conditions:[a+ bT1 +
8i=1
ciTiair
]3.6 (1)
0.97P (2)
are e
EThe4
wherder farea
diffe
TlmGateCycle
stant values for a, b and c1 to c8, are given in
heat exchangers set
te heat exchangers (PHE) are the physical con-between the
cogeneration system and the recrys-rocess. The PHE set was included
using its design
eat balances at both water and brine sides and ma-ces, assuming
there is neither mass nor energyon inside the exchangers.E design
equation Eq. (3) (Kakac & Liu, 2002),r-current flow
arrangement, is applied to each ex-ce it is considered that all
four heat exchangers
T1, T2, Ttaheated wattank and th
The PHbased on thues in Gatand hwater 2the triple-pfor the
diff500 kPa; ou
EThermal =hwater1 = 4hwater 2 = 4
and c1 to c8, for the thermal power correlation (Eq. (1))b c1 c2
c3 c4 c588.73 910.5 244.6 40.49 3.56 0.17ers and salt production
unit.
UHEAHE Tlm (3)
ervice overall heat transfer coefficient (UHE), un-conditions,
is 24.69 MJ/(m2 C h), the exchangeris 78.2 m2 and the logarithmic
mean temperatureTlm is a function given by:
2 THE) (T1 Ttank)ln[T2THET1Ttank
] (4)
nk and THE are the temperatures of the cold ander in the
economizer and of the exit streams of thee PHE, respectively.E heat
balance on the water side (Eq. (5)) ise linear correlations
obtained via simulated val-eCycle for the water stream enthalpies
(hwater 1). The zero-point of enthalpy is liquid water atoint: 0 C
(GateCycle, 2001). It is also accountederent pressures values in
the water streams (inlet:tlet: 490 kPa):Mwater[hwater2 hwater1]103
(5).1835T1 + 0.02714 (6).2040T2 1.5812 (7)
c6 c7 c8
0.00457 6.356 105 3.577 107
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1497
: solid
Mwater is ththe econom
Eq. (8) rin which Hinlet stream
EThermal =
The enthresponds to25 C (thatthalpy of
acentrationdeterminedHoughen, W
H = Q[(+X
where Cwpues calculand Green0.8712 1lution Hdrelation basWagman
et
Hdiss =[
where n recomponent
Thus, ththrough eqquate value
The brintemperaturtion was obdifferent tein Perry
anMathemati
FM(F
0.5
1001
e Qalculalt (E
pitati
tank
Xtank
Splitt
fter lis di. Thi
red topond
P,i =
PondFig. 7. Phase thermodynamics equilibrium
e mass flow rate of the water stream circulating inizer and is
equal to 1.47 106 kg/h.epresents the PHE heat balance on the brine
side,HE and Htank are the enthalpies of the outlet ands in the
exchangers set, respectively:
HHE Htank (8)
alpy reference state considered in all streams cor-the
components in their state of aggregation atis, liquid water and
solid NaCl). Thus, the en-
liquid brine stream with a flow rate (Q), salt con-(X),
temperature (T) and brine density () will be
through Eq. (9) (Coulson & Richardson, 1989;atson, &
Ragatz, 1972):
X)Cwaterp (T 25)+XCNaClp (T 25)Hdiss] (9)
ater and CNaClp are mean heat capacity val-ated using the
expressions given by Perry
(1997), and are equal to 4.189 103 and03 MJ/(kg C),
respectively. The heat of disso-iss (Eq. (10)) is determined
through a linear cor-ed on the enthalpies of formation values given
byal. (1982), for X> 150 kg/m3:(
nwater) ]
3
nate
=
0 =
Thare c
the sapreci
Qtank
Qtank
4.3.
AbrineFig. 6requieach
QEnt
4.4.
1.3712
nNaCl+ 19.4063 10 (10)
presents the number of moles of the respective.e enthalpies
values HHE and Htank are determineduations Eqs. (9) and (10), by
considering the ade-s for Q, X, T and .e density is related with
brine concentration and
e values through Eqs. (11) and (12). This correla-tained using
the tabulated values of the density atmperatures and salt mass
fractions (FM) presentedd Green (1997). The expression was modified
inca 4.1.0.9 of the Wolfram Research Inc. to elimi-
As it isinvolves afor modell
The dyand differelibrium th(see Fig. 7in the variathe expectthe
heatedtaps in eacwell as co, liquid and gas.
M = 100X/):
0 +20 + 400X(7.7780 0.0063176T )
(11)
.23 0.22715T 0.0020480T 2 (12)
and X values of the brine stream leaving the PHEted via the
material balances for the brine and for
qs. (13) and (14)). It is assumed that there is no salton within
the exchangers:
= QHEHE (13)
= QHEXHE (14)
er unit
eaving the plate heat exchangers set, the heatedvided into the
working ponds, as it can be seen ins is an isothermal process, and
therefore it is onlydetermine the brine flow rate fraction entering
in
:
fEntP,iQHE, i = 1, . . . , 6 (15)
unitillustrated in Fig. 6, the salt production processmaximum of
six recrystallization ponds, which
ing purposes are considered to be equal.namic model proposed is
built through algebraicntial equations taking into account the
phase equi-ermodynamics: solid, liquid and gas equilibrium). It is
a distributed model, since there is a changebles within both time
and axial domain, providing
ed profile inside the ponds and allowing modellingbrine entrance
through one or both the two existingh pond. It involves material
and heat balances, as
nstitutive relationships (Bequette, 1998).
-
1498 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
4.4.1. Material balancesConsider a parallelepiped volume element
V (in the ax-
ial domain L) and a time element t, in which there is
acontinuouselement, wthe law of min the elemtion,
result(Bequette,
EiW(Z)t
where W,infinitesim
QL,Tap,Etap, per unipond, locatis either equtotal flow eflow
fractioelement lenof element
The watconcentrati(T) and ofhumidity mixed lami
Mevap =
[
Lc is the cdirection isgiven by:
Xm = 1+
and the salinity S:
S = 100 X
(19)
rine anir temtion Eand a
exp[
+ 4.1
e saltinfini
alt preumedn of tvaluelpp,Ei
(Eq. (2. If thee balauatio
.05
(XEL
uatio
lpp,Ei
XEiZ
t
e brigh eq
=Sol
e the s26) (L
ity sa:
35.51
e toClpp,to
NaClpp,tbrine flow, a brine entrance through the tap in theater
evaporation and salt precipitation. Applyingass conservation for
the brine solution contained
ent considered, and after mathematical manipula-s into the brine
material balance equation Eq. (16)1998):
= (EiQEi)L
Mevap,EiW MNaClpp,EiW+QL,Tap,EiTap,Ei (16)
the pond width, is 54.8 m and Ei represents theal element i.i is
the brine flow rate entering in Ei through thet length of the
element. There are two taps in eached in specific axial positions.
Therefore, QL,Tap,Eial to 0 (no tap exists in the element) or equal
to the
ntering the pond multiplied by the brine entrancen in the tap
(between 0 and 1) and divided by thegth (pond length182.5 m divided
by the number
s considered).
er evaporation rate (Mevap) is a function of the salton (X), of
the temperature of the brine solutionthe atmospheric conditions
(air temperature Tair,and wind velocity Vwind), and corresponds to
thenar-turbulent flow regime (Sartori, 1991, 2000):
0.00407V 0.8windL0.2c 0.01107L1c ] [XmPbrine Pair]3600
Patm(17)
haracteristic length (equal to 54.8 m when windnorth), and the
mole fraction of the water Xm is
10.621 S100S
(18)
Pband aequabrine
P =
Thsame
the sis asstratiotionMNaCtion(21))by th
EqXEi = Xsat,Ei + 0
XEiW (Z)t
=
EqMNaC
W (
Ththrou
Xsat
wherEq. (densvalue
Sol =
Th(MNa(MXdPair are the water vapour partial pressure at
brineperature, respectively, and are determined throughq. (20)
(Perry & Green, 1997) by considering their temperatures:
73.649 7258.2T + 273.15 7.3037 ln[T + 273.15]
653 106(T + 273.15)2]
(20)
material balance is determined by applying to thetesimal element
the law of mass conservation forsented in the brine solution (Eqs.
(21)(24)). Itthat precipitation only occurs when the concen-
he brine solution is 0.05 kg/m3 above its satura-(XEi Xsat,Ei
0.05). The salt precipitation rateis determined through the
material balance equa-2)) and the salt concentration value is fixed
(Eq.solution is not supersaturated,Xwill be computednce equation
(Eq. (24)), with MNaClpp,Ei = 0:ns used if XEi Xsat,Ei 0.05:
(21)
iQEi) MNaClpp,EiW +QL,Tap,EiXTap,Ei (22)
ns used if XEi Xsat,Ei < 0.05= 0 (23)) = (XEiQEi)
L+QL,Tap,EiXTap,Ei (24)
ne saturation concentration Xsat is calculateduation Eq.
(25):Sol+ 100sat (25)
alt solubility in the water Sol is given by equationanger &
Offermann, 1982), and the saturation
t by equations Eqs. (11) and (12), using the Xsat
49 0.23125T0.0069163T
(26)
tal salt mass obtained in the whole pondtal) is determined
by:
total) = L=182.5L=0
MNaClpp,EiW dL (27)
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1499
4.4.2. Energy balancesThe same parallelepiped volume element V
and time
element t were considered. In this case, it will also be
ac-counted fotion, as wethe brine sosulting ene
W(HT,Ei)
t
The enthalare comput)Ei and (Q
The totthrough eqstate as be(Hdiss,Ei)and X value
HT = Z[(+X
The ententhalpy re
Eevap = Mwith the waand the latebased on th
evap = [25Thus, th
through Eqelement Ei
The radsurface (Er(IncroperaErad = [
where isStefanBolTsky is the1996):
Tsky = [Tai
The dewperature an
sented by Perry and Green (1997), and is computed by:Pdew-point
= Pair (34)
ew-poipointned thecauseare e
ndent(air teminede (Inc= [d
constarties:
0.037
871K
e they (airs, ass giveair te0.30.e enthelem
lpp,Ei
e exthere,e time(10 y), at
tion thion anum o
eWitt,o-Diesuns
A valuis refleasses
nergybed ben (19depth
. Bous the min thefor th
entratir the energy losses due to radiation and convec-ll as the
solar energy, which is absorbed throughlution contained in the
defined element. The re-
rgy balance is given by:
= (HEi)L
Eevap,EiW Erad,EiWEconv,EiW + ENaclpp,EiW + Esolar,EiW+HL,Tap,Ei
(28)
py of the liquid brine streams HEi and HL,Tap,Eied via equations
Eqs. (9) and (10), using (Q, X, T,L, X, T, )Tap,Ei values,
respectively.
al enthalpy in the element HT,Ei is calculateduation Eq. (29),
for the same enthalpy referencefore, and the heat of dissolution in
the elementdetermined by Eq. (10), with the corresponding Ts of
Ei:
X)Cwaterp (T 25)+XCNaClp (T 25)Hdiss] (29)
halpy of a vapor water stream, according to theference state
defined, is determined by:
evap[evap + Cwaterp (T 25)] (30)ter evaporation rate Mevap
determined by Eq. (17)nt heat of vaporizationevap by a linear
correlatione values given by Daubert (1985):03.0 2.432T ]103 (31)e
enthalpy stream value Eevap,Ei is computeds. (30) and (31) at the
brine temperature in the
.iation energy losses through the horizontal brinead) depends on
the sky and brine temperatures& DeWitt, 2001):(T + 273.15)4
(Tsky + 273.15)4]3.6 103
(32)the surface emissivity (equal to 0.95), is thetzmann
constant (5.67 108 W/(m2 K4)) andsky temperature calculated by Eq.
(33) (Sartori,
r + 273.15][
0.8+ Tdew-point250
]1/4 273.15 (33)
-point temperature is correlated with the air tem-d humidity
through the psychometric charts pre-
Pddew-termi
Btheredepetionsdeterregim
Econv
Theprope
d1 =
d2 =whercositlationvaluemean
d2 = 2Th
in the
ENaC
Thmospon thvalue(2004radiasorptis a s&
DCastrusingNASfacerest pthe eabsorNielsbrine
4.4.3A
tionsfinedconcnt and Pair are the water vapour partial pressure
atand dry air temperature, respectively, and are de-ough equation
Eq. (20), by replacing those values.of the air movement above the
liquid brine surface
nergy losses due to forced convection. These areon both brine
temperature and atmospheric condi-mperature Tair and wind velocity
Vwind), and areby considering the mixed laminar-turbulent
flowropera & DeWitt, 2001):
1V0.8windL
0.2c d2L1c ][T Tair]3.6 103 (35)
nts values d1 and d2 depend on the air physical
K2/3air C
1/3p,air
7/15air
0.8air (36)
2/3air C
1/3p,air
1/3air (37)
air conductivity (Kair), heat capacity (Cp,air), vis-) and
density (air) are calculated through corre-a function of
temperature, based on the tabulatedn by Perry and Green (1997).
Thus, using themperature value of 25 C gives: d1 = 6.006 and
alpy of the salt (NaCl solid) precipitated stream,ent Ei, is
computed by:
= CNaClp (TEi 25)MNaClpp,Ei (38)raterrestrial solar radiation,
at the top of the at-depends on the geographic coordinates as well
asof the day, month and year. Its monthly averageears data) can be
retrieved from NASA Websitelatitude: 39.983 and longitude: 8.8. The
solarat reaches the earths surface is lower due to ab-d scattering
by the atmospheric constituents, andf the direct and diffuse
contributions (Incropera2001). These values are estimated through
the
z, Aladaos-Arboledas, and Jimenez (1989) work,hine data
collected at the site and the retrievedes. Part of the radiation
that reaches the brine sur-cted (Weinberger, 1964), part is
absorbed and thethrough. The net solar energy value considered
inbalance equation Eq. (28),Esolar,Ei, is the radiationy the brine,
which is calculated using the Rabl and75) work, assuming a light
path of two times the.
ndary and initial conditionsodel includes first order partial
differential equa-spatial domain boundary conditions must be
de-
e distributed state variables, namely, for the brineon (Eq.
(39)), temperature (Eq. (40)), flow rate
-
1500 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
(Eq. (41)) and salt precipitation rate (Eq. (42)):X
L
L=0
= 0 (39)
T
L
L=0
=
Q|L=0 = 0MNaClpp
L
It is alsovariables, tlevel (Z), beach eleme(MNaCl,total
4.5. Chann
The brintank througnel will bestate condievaporationand
therefoered to be athe tank (EQExitP,i =
The temthe channein the exit s
Xchannel =
Tchannel =
4.6. Tank u
The tanproductionbe seen in Flected in ththen sent topurge. It
isequations,ances and istate variab
Equatiorial balanceinto accouning the smastirring due
there is no salt precipitation:
tankAtank(Ztank) = QFF +Qchannelchannel
rine dEqs. (ms. Thtank, repute
tank vf 4 m
ual toe brin
minedequatydroded tos. This brineeen th
=
e brinassoci
k =
n a stamputetion anlem, dtion Ederiva48): Ve corq. (4(Xta
e tanto then and
idered
(HT0 (40)
(41)
=0 = 0 (42)necessary to define the initial values of the
state
hat is, at time zero. Initial values for the brinerine
concentration (XEi) and temperature (TEi) innt Ei, as well as for
the total salt mass precipitated), are given.
el unit
e leaving from each working pond flows to theh the channel. For
modelling purposes, the chan-
considered as a collecting unit only. It is in steady-tions,
without neither salt precipitation nor water. All hydrodynamics
phenomenon are ignored,re the brine flow rate leaving each pond is
consid-fraction of the total brine flow rate that goes into
q. (43)):fExitP,iQchannel, i = 1, . . . , 6 (43)perature and
salt concentration of the brine leavingl is the weighted mean of
its corresponding valuestreams from the ponds (Eqs. (44) and
(45)):6
i=1XExitP,iQExitP,i6i=1QExitP,i
(44)
6i=1TExitP,iQExitP,i6
i=1QExitP,i(45)
nit
k is the physical connecting unit between the saltprocess and
the plate heat exchangers set, as it canig. 6. Through this unit it
is received the brine col-e channel, mixed with the fresh treated
brine, and
the heat exchangers set, discounting the systemmodeled using
algebraic and ordinary differentialwhich consist in the brine and
salt material bal-ts energy balance. It is a lumped model, since
theles change only with time (Bequette, 1998).n Eq. (46) represents
the macroscopic brine mate-, assuming a perfectly mixed system, and
takingt for all entering and exiting streams. Consider-ll
dimensions of the tank and its permanent fluidto the strong pumps
suction, it is assumed that
The btionsstreaandTis comand X(Lc) ois eq
ThdeterextraAll hsiderpondpondbetw
Ztank
Thalso
Ztan
t
Wheto coequaprobequatimeand (
Thtion E
Atank
ThDuediatiocons
Atankt
Qpurgetank QtanktankMevap,tankAtank (46)
ensity value of each stream is determined by equa-11) and (12),
usingX and T values on its respectivee purge stream X and T values
are equal to Xtankspectively. The water evaporation rate
(Mevap,tank)
d by equation Eq. (17) replacing T andXwith Ttankalues, and by
considering a characteristic length, if wind direction is north.
Tank area value (Atank)40 m2.e flow rate collected by the channel
(Qchannel) isthrough Eq. (46), and therefore it is necessary an
ion to calculate the brine level in the tankZtank.ynamics
phenomenon are ignored, so Ztank is con-be associated with the
brine level inside all thes value is obtained by the weighted mean
of the
levels (Zi, i= 1, . . ., 6) plus the 50 cm differencee tank and
ponds floor (Eq. (47)):6i=1ZiQExitP,i
6i=1QExitP,i
+ 0.5 (47)
e level time variation inside the tank (Ztank/t) isated with its
variation inside the ponds:
6i=1
[Zit
]QExitP,i6
i=1QExitP,i(48)
te variable is specified in a DAE system in ordera term on the
right hand side of a differentialindex problem could arise. To
avoid this index
ue to the calculation of the state variable Ztank byq. (47), it
was used an auxiliary variable for thetive of the tank brine level
in equations Eqs. (46)AR = (Ztank)/t.responding salt material
balance is given by equa-9):nkZtank)t
= QFXF +QchannelXchannelQpurgeXtank QtankXtank (49)
k energy balance equation is given by Eq. (50).tank small
surface area the energy losses by ra-convection, as well as the
solar energy, were not
:
,tank)t
= HF +Hchannel Hpurge HtankEevap,tankAtank (50)
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1501
The total enthalpy inside the tankHT,tank is calculated
throughequation Eq. (29), withHdiss,tank value obtained by Eq.
(10),using Ttank and Xtank values.
The enthalpy of the liquid brine streams HF, Hchannel,Hpurge and
Htank are computed via equations Eqs. (9) and(10), using its (Q, X,
T, ) respective values.
The enthalpy value Eevap,tank is determined by Eqs. (30)and
(31), using Ttank.
The brine concentration and temperature initial values, attime
zero, are given.
5. Dynamic simulations
The whole integrated system was modelled through
thegeneral-purpose modelling, simulation and optimisation
toolgPROMS2.3, of the Process System Enterprise Ltd. This soft-ware
allows to adequately handling process discontinuities,lumped and
distributed systems and many different types ofoperating procedures
(gPROMS, 2004).
Fig. 8 is an information flow diagram illustrating the dy-namic
model structure implemented in this software. Thereare five subical
unit, inneeded to dthe brine flodensity ()
The gloas the Recrthe plate hepower (TP)FlowSeparindependen. .
., 6) all m(such as: eqcharacterizsalt concen
into account the brine received from each working pond al-lowing
to characterize the brine entering in the feed tank(model Tank).
Through this model Tank it is included the in-formation on the
fresh treated brine fed to the recrystallizationunit and the purge
of the system.
The external atmospheric conditions are included in themodels
Heating and Pond i.
The model described in Section 4.4 is axially distributed.For
this kind of systems the selection of an appropriate
dis-cretisation method is crucial (gPROMS, 1998). For
purelyconvective problems the finite method is suggested with
ageneral rule: discretisation method opposite to the directionof
the flow. As the axial domain goes from top to bottom itwas chosen
the backward finite difference method.
The total number of model variables depends on the num-ber of
discretisation intervals used in the axial distributed do-main. For
20 intervals its value is approximately 5800, whilefor 60 intervals
is around 15,000. An increase of the numberof intervals requires a
greater computational effort.
Among all the model variables, it is important to identifythe
decision and the state variables. The decision variables arethe
ones that allow verifying if the system response is within
gion onal inodel derature colle
2). Thderabrefullyvariabic conitions,rate of2 sum
sing ththroug
rated p-models, each corresponding to an industrial phys-cluding
the algebraic and differential equationsescribe each system. Some
variables referring tow rate (Q), concentration (X), temperature
(T) andare used as connecting sub-models information.
bal model that includes all these units is labelledystallization
Unit model. Model Heating refers toat exchangers equation set using
also the thermalcorrelation determined previously. In the
modelation the brine flow rate fraction that is sent to eacht pond
is determined. In the model Pond i (i= 1,ass and heat balances
equations and other relationsuilibrium, heat and mass transfer) are
included toe the brine profiles (for instance: temperature
andtration) inside the ponds. Model Channel takes
its reeratiothe mtempin thand Tconsibe
castatesphercondflowTable
Ucess,
Fig. 8. Structure of the dynamic model of the integf working
feasibility. In this case, to respect the op-tervals of the
temperature of the circulation water,ecision variables would be
either the heated watereT2 (withT2 = 90 5 C) or the brine
temperaturecting channel (since it can be correlated with T1e state
variables are independent variables with a
le influence in the system behaviour, and shouldanalysed. So,
for the integrated process the main
les are: the number of ponds in service, the atmo-ditions, the
fresh brine flow rate, the initial brinethe number of plate
exchangers working and thethe brine pumped into the plate heat
exchangers.marizes the model mathematical structure.
e developed dynamic model of the integrated pro-h simulation in
gPROMS, it is possible to analyse
rocess in gPROMS 2.3.
-
1502 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
Table 2Mathematical structure the dynamic model
Main input variables Main output variables Typical performance
statistics for 60 elements
Atmospheric conditions (d(t)) Distributed Number of
variablesTair, Vwind, , Esolar T Differential: 1114
X Algebraic: 14 538Operational conditions (u(t)) MNaClpp
Number of working ponds Mevap Number of equations: 15 652Taps
used in each pondNumber of exchangers working Others Process time
horizon: 1 monthBrine flow rate Qtank Z gPROMS execution time: 150
saFraction fEntP for each pond EthermalFresh treated brine: QF, XF,
TF T1 and T2Initial conditions to the ponds and the tank: Z(0),
T(0), X(0), MNaCl(0) Process efficiencies: CSE and GPE
u(t): input variables that can be manipulated to optimise the
system performance.d(t): Input variables that correspond to
external disturbances.
a Pentium 4, 2.0 GHz, 524 MB RAM.
the effect of some of these state variables on the system andto
study so
5.1. Start-
Since itup the recrydeterminemum numbbrine flowatmospherisystem
resrate. Amonconcludedthermal enthree pondstemperatursponse in
tconditions.dependent
5.2. Scena
To addrescenarios oThus, four
nario there are four ponds in service receiving the thermalr
throd scerst pof heating pocenariing siFig. 9): Tair =energr all
secon
It wasworkirios 1f heatrio. T8 Culingarting2 C avaporonlys 3
athirdme scenarios in some special conditions.
up conditions
was necessary to have an operational plan to start-stallization
process, several studies were made to
the best start-up conditions: function of the mini-er of ponds,
the initial brine levels and the fresh
rate. The simulations allowed concluding that thec conditions
have a very strong influence on theponse since they modify the
water evaporationg several possible atmospheric conditions, it
wasthat it is not possible to start the reception of theergy from
the cogeneration system with less than
in service, to respect the operational intervals ofes. The brine
level only influences the system re-he beginning, since it led to
the same steady stateThe flow rate of fresh brine allowed is also
very
on the evaporation rate.
rios of scheduling working ponds
ss one of the main objectives of this work severalf scheduling
working ponds have been analysed.different scenarios were
evaluated. In the first sce-
powesecon
the fitity oworklast swork(seewere
solarFo
in thevals.threescena
tity oscena
up tosched
StT= 2the esaltnarioTheFig. 9. Fraction of heated brine entering
in each pond for sevugh an identical flow rate of heated brine. In
thenario there are also four ponds working, howevernd is a turbo
because it receives a larger quan-ed brine. The third scenario
considers only threends, with one of them as a turbo pond. In theo
there are four ponds involved but only three aremultaneously, and
the first pond is also a turbo. The average atmospheric conditions
considered12 C, Vwind = 4.5 m/s, humidity = 82% and net
y = 3.7 MW.cenarios, the temperatures of the water
circulatingomizer are within their defined operational
inter-observed that in the scenarios 3 and 4, with onlyng ponds,
water evaporates up to 6% more than inand 2. Having a pond that
receives a larger quan-
ed brine than the others is even a more favourablehe brine
temperature in the turbo pond increases. Thus, amongst all
scenarios the more efficientprocedure was the third one.from the
initial status (no solid, level = 1.5 m,nd X= 215 kg/m3) the salt
production follows
ation profile, therefore the precipitation of theoccurs when
three ponds were utilized (sce-nd 4), for the period time interval
assumed.scenario was also the more adequate, since iteral
scheduling scenarios.
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1503
Fig. 10. Variation of the circulation water temperatures for two
atmospheric cases, with the number of ponds indicated (( ) feasible
working region).
enhances and accelerates the salt production (3.6
timesmore).
5.3. Effect
Fig. 10 scirculatingatmospherimidity = 41has the sama wind
velservice (rethen five pofinally all thsible to anathe
systemspheric con
four ponds. However, when the wind velocity is smaller,
andtherefore the water evaporation rate is reduced, it
becomesnecessary to use at least five ponds, to obey the
operational
ed tem
Inueitions
nce thur it ipatterveloc
g profihich th
(dur) to 50ergy v
Fig.of atmospheric conditions
hows the variation of the temperature of the waterin the
economizer (T1 and T2) for two differentc cases. In the first case
study Tair = 20.7 C, hu-.7% and Vwind = 1.5 m/s. The second case
studye values for the air temperature and humidity, but
ocity of 2.0 m/s. In both cases, four ponds are inceiving heated
brine) while the fifth is filled up,nds are in service while the
sixth is filled up, ande six ponds are working. From this figure it
is pos-lyse the strong influence of the wind velocity onand to
conclude that for the most favourable atmo-ditions (Vwind = 2 m/s)
it is only required the use of
defin
5.4.cond
Sihavionightwindsprinin w20 Cnightlar en11. Effect of the day
and night patterns on the water circulation temperatures
withperature intervals.
nce of day and night patterns of atmospheric
e atmospheric conditions influence the system be-s essential to
analyse how the different day andns of air temperature, humidity,
solar energy andity values affect the system response. A typicalle
for the atmospheric conditions was assumed,e air temperature varies
from 5 C (at night) to
ing the day). Air humidity varies from 85% (at% (at daylight).
The wind velocity and the net so-ary from 0.5 to 3.5 m/s and 0 to
3.5 kWh/(m2 day),3, 4 and 3 + 1 ponds (( ) feasible working
region).
-
1504 R.D. Moita et al. / Computers and Chemical Engineering 29
(2005) 14911505
respectivelit would beto respect tas illustratehas
confirmcirculation
6. Determ
The coggrated procover naturaon the atmimately 92energy lossfrom
the oFig. 12. Energy distribution in the recrystallization
ponds
y. From this study it was possible to conclude thatadvisable to
use an extra pond during the night
he mentioned operational temperatures intervals,d in Fig. 11.
The industrial experience at the siteed that the highest
temperatures values for the
water were observed in the morning.
ination of the process energy efciencies
eneration system efficiency (CSE) of the inte-ess is the ratio
of thermal and electrical powerl gas heat consumption. This value
is dependentospheric conditions and in average it is approx-%.
During the recrystallization process there arees due for instance
to radiation and convectionpen-air ponds. So, the global process
efficiency
(GPE) of thThe GPE eoration
powconsumptiooperationalSimulationthe range o
7. Conclu
A dynaof an integtem and thsome variastudied usiof the whofor
two example cases.
e integrated process is inferior to its CSE value.fficiency is
calculated by the ratio between evap-er plus electrical power and
the natural gas heatn. It is strongly dependent on the atmospheric
andconditions considered, as illustrated in Fig. 12.
s showed that the global process efficiency is inf 7080%.
sions and future work
mic model was built to simulate the behaviourrated process that
includes the cogeneration sys-e salt recrystalisation process. The
influence ofbles on the cogeneration system performance wasng
GateCycle 5.34.0.r. Then, a dynamic modelle integrated process was
developed and simu-
-
R.D. Moita et al. / Computers and Chemical Engineering 29 (2005)
14911505 1505
lated in gPROMS 2.3. Different scenarios were explored withthe
purpose of maximizing the global process efficiency, byanalysing
the influence of several atmospheric and opera-tional conditions on
the integrated system.
The advantage of a model that can simulate the main pro-cesses
is to minimize the negative influence caused by someadverse
atmospheric conditions to achieve the highest possi-ble global
efficiencies. Furthermore, the better understandingof the
integrated system acquired by simulating the processunder different
possible scenarios, allows the definition of animproved set of
operational conditions to obtain a long-termprofitable
business.
Future work will include the study of the spray systemeffect on
the salt production and long-term time simulations,accounting for
day and night patterns of atmospheric condi-tions. Moreover this
will allow to obtain in advance the saltharvesting
Acknowled
The autfrom the P(Grupo Nahas been oProcess In
Reference
Bequette, W.ulation. N
Castro-Diez,for climatation on a
Coulson, J. MLisboa: F
Daubert, T. EMcGraw-
GateCycle. (LLC.
GateCycle Soproducts/p
gPROMS. (1don: Cent
gPROMS. (2004). gPROMS introductory user guide. London:
ProcessSystems Enterprise Ltd.
Gundersen, T. (2002). A process integration primer, SINTEF
En-ergy Research, available on the web site of IEA
(http://www.iea-pi.org/primer.html).
Houghen, O. H., Watson, K. M., & Ragatz, R. A. (1972).
Princpiosde Processos QumicosI Parte: Balancos Materiais e
Energeticos.Porto: Lopes da Silva Editora (in Portuguese).
Incropera, F. P., & DeWitt, D. P. (2001). Fundamentals of
heat and masstransfer. New York: John Wiley & Sons.
Kakac, S., & Liu, H. (2002). Heat exchangers: Selection,
rating andthermal design. Florida: CRC Press.
Langer, H., & Offermann, H. (1982). On the solubility of
sodium chloridein water. Journal of Crystal Growth, 60(2), 389.
Linnhoff, B., et al. (1982). User guide on process integration
for theefcient use of energy. Rugby, UK: Inst. Chem. Engrs.
Moita, R. D., Matos, H. A., Fernandes, C., Nunes, C. P., Prior,
J., & San-tos, D. A. (2004). Process integration of a dynamic
industrial system.In A. Barbosa-Povoa & H. Matos (Eds.), ESCAPE
14, computer-
ed chesevier., R. D.Santos,egratedd Techn
WebsiR. H.,ok. NewA., & Nergy, 1
s, S., Feacao deAmbienrtuguesi, E. (1ntration47235i, E. (e
study9.i, E. (20ion of t(1), 77., R. (ll.an, D.iley, S.mic
pro(Suppl.erger, H.activity scheduling.
gements
hors gratefully acknowledge financial supportortuguese National
Team on Process Integrationcional de Integracao de Processos,
GNIP), whichperating in Portugal, within the framework of
thetegration Implementing Agreement.
s
B. (1998). Process dynamics: Modeling, analysis and
sim-ew-Jersey: Prentice Hall.
Y., Aladaos-Arboledas, L., & Jimenez, J. I. (1989). A
modelological estimations of global, diffuse and direct solar
radi-
horizontal surface. Solar Energy, 42(5), 417.., &
Richardson, J. F. (1989). Tecnologia Qumica, Vol. VI.
undacao Calouste Gulbenkian (in Portuguese).. (1985). Chemical
engineering thermodynamics. New-York:Hill, Prentice Hall.2001).
GateCycle manual. California: GE Enter Software
ftware Website. (2004). http://www.gepower.com/prod serv/lant
perf software/en/gatecycle/index.htm.
998). Training course I: An introduction to gPROMS. Lon-er for
Process Systems Engineering.
aidEl
Moita&intan
NASAPerry,
boRabl,
EnRelva
gre
PoSartor
ce
23Sartor
tiv19
Sartorlat68
SmithHi
WagmBana
11Weinb
45mical engineering: Vol. 18 (pp. 445450). Amsterdam:
, Fernandes, C., Matos, H. A., Nunes, C. P., Prior, J. M.,D. A.,
et al. Dynamic modelling of a cogeneration systemwith a salt
recrystallization process. Chemical Engineeringology, submitted for
publication.te. (2004). http://eosweb.larc.nasa.gov/sse/.&
Green, D. W. (1997). Perrys chemical engineers hand-
York: McGraw-Hill.ielsen, C. E. (1975). Solar ponds for space
heating. Solar
7, 1.rnandes, M. C., Matos, H. A., & Nunes, C. P. (2002).
Inte-ProcessosUma Metodologia de Optimizacao Energetica
tal. Lisboa: Programa Operacional de Economia do ME (ine).991).
Evaporation from a free water surface with salt con-. In
Proceedings of the ISES solar world congress (pp.1).1996). Solar
still versus solar evaporator: A compara-
between their thermal behaviors. Solar Energy, 56(2),
00). A critical review on equations employed for the calcu-he
evaporation rate from free water surfaces. Solar Energy,
1995). Chemical process design. New York: McGraw-
D., Evans, W. H., Parker, V. B., Schumm, R. H., Halow, I.,M., et
al. (1982). The NBS tables of chemical thermody-
perties. Journal of Physical and Chemical Reference Data,2),
2301.. (1964). The physics of the solar pond. Solar Energy,
8(2),
Dynamic modelling and simulation of a cogeneration system
integrated with a salt recrystallization
processIntroductionFramework developedAnalysis of the cogeneration
systemDynamic model of the integrated systemCogeneration
systemPlate heat exchangers setSplitter unitPond unitMaterial
balancesEnergy balancesBoundary and initial conditions
Channel unitTank unit
Dynamic simulationsStart-up conditionsScenarios of scheduling
working pondsEffect of atmospheric conditionsInfluence of day and
night patterns of atmospheric conditions
Determination of the process energy efficienciesConclusions and
future workAcknowledgementsReferences