Full Paper

Steady-State Analysis of the Multiple Effect Evaporation

Desalination Process
By Hisham El-Dessouky, Imad Alatiqi, S. Bingulac and Hisham Ettouney*

Mathematical modeling of the multiple effect evaporation (MEE) desalination process has been carried out to determine
the effects of the important design and operating variables on the parameters controlling the cost of producing fresh water.
The model assumes the practical case of constant heat transfer areas for both the evaporators and feed preheaters in all ef-
fects. In addition, the model considered the impact of the vapor leak in the venting system, the variation in thermodynamic
losses from one effect to another, the dependence of the physical properties of water on salinity and temperature, and the
influence of noncondensable gases on the heat transfer coefficients in the evaporators and the feed preheaters. The modi-
fied fixed-point iterative procedure is used to solve the large number of highly nonlinear equations describing the MEE
desalting system. The algorithm consists of 10 calculation blocks and 6 logical blocks. The algorithm is implemented using
L-A-S computer aided language. Results show that the heat transfer coefficients increase with the boiling temperature.
Also, the heat transfer coefficient in the evaporator is always higher than that in the feed preheater at the same boiling
temperature. The plant thermal performance ratio is nearly independent of the top brine temperature and strongly related
to the number of effects. The specific heat transfer area increases by raising the number of effects and reducing the top
brine temperature. The effect of the top brine temperature on the specific heat transfer area is more pronounced with a
larger number of effects. The required specific heat transfer areas at a top brine temperature of 100 °C are 30.3% and
26% of that required at 60 °C when the number of effects are 6 and 12, respectively. The specific flow rate of cooling
water is nearly constant at different values of top brine temperature and tapers off at a high rate as the number of effects
is increased. Two correlations are developed to relate the heat transfer coefficients in the preheater and the evaporator to
the boiling temperature. Design correlations are also developed to describe variations in the plant thermal performance,
the specific heat transfer area, and the specific flow rate of cooling water in terms of the top brine temperature and the
number of effects.

1 Introduction 11000 with a capacity of 18.6 ´ 106 m3/d [1]. This capacity
is expected to double within the next 20 y. Nevertheless,
To live is to use water. Unfortunately, fresh water, like many other countries still suffer from water shortage and
most other natural resources, is unevenly distributed in the cannot afford the high capital required and the running
universe. At the same time, many countries, especially, in costs of desalination plants. Presently, the majority of de-
the Middle East suffer from the increasing demand for fresh salination plants in operation are the multistage flash
water and the limited available natural water resources. This (MSF) desalination systems. However, the MEE desalina-
gap is expected to widen dramatically in the near future, due tion process has recently acquired a potential interest. The
mainly to the high rate of population growth and the im- authors strongly believe that the high temperature MEE
provement in the standard of living, which needs a large process is a very strong candidate for realization in the
quantity of good quality water per capita. Desalination of near future.
sea water or brackish water is the logical or the only avail- The preeminent advantages of the MEE when operates at
able solution to safeguard supplies of water so that the tap high temperature (about 100 °C) compared to the predomi-
does not run dry. nant MSF system are
In the last four decades, the water desalination industry ± high thermal efficiency, with a lower number of effects;
has developed at a very high rate. The number of desali- ± high heat transfer coefficient;
nation plants in 1960 was 5 units with a total production ± relatively low specific investment cost;
rate of 5000 m3/d, while in 1995 the number of units was ± low pumping power;
± high operating flexibility with short start-up period and
capabilities for closely matching production volume with
water demand and energy supply; and
± a distinct advantage offered by the MEE system when
±
[*] H. El-Dessouky (to whom correspondence should be addressed), Che-
compared with the MSF design is that tube leaks, if they
mical Engineering Department; I. Alatiqi, Chemical Engineering De- occur, do not cause distillate contamination. This is be-
partment, S. Bingulac, Electrical & Computer Engineering Department, cause the vapor side has higher pressure than the sea
and Hisham Ettouney, Chemical Engineering Department, Faculty of
Engineering & Petroleum, Kuwait University, P.O. Box 5969, 13060 Sa- water side. Thus, tube leaks result in loss of distillate with-
fat, Kuwait; E-mail Address: [email protected] out contamination of the product water.

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00437
0930-7516/98/0505-0437 $$17.50+.50/0
17.50+.50/0 437
 Full Paper

There are three different possible configurations for the mized processes were as follows:
MEE desalting systems, which differ in the flow directions of ± unit capital cost/gpd: MSF $8.7 (PR at 10), MEE, $7.01
the heating steam and the evaporating brine. The backward- (PR at 8);
feed arrangement is not suitable for application in sea water ± operating and maintenance cost/kgal: MSF, $4.2, MEE,
desalination. The parallel feed layout is by no means the $3.35;
most economical and is efficient only when the feed brine is ± total cost of water/kgal: MSF, $7.05, MEE, $5.65.
nearly saturated to begin boiling inside the effects. The salt On the other hand, Hess and Morin [5] presented results
concentration reaches the maximum permissible value in all of two different studies comparing the cost of water pro-
effects. Therefore, the parallel configuration is very appro- duced from different desalination processes. The study indi-
priate for low top-brine temperature design. In the forward- cated that the water cost at the distribution point is $1.35/m3
feed system, both the brine and the heating steam flow con- for MEE and $1.058/m3 for RO (1997 basis).
currently from the high temperature to the lower tempera- Rautenbach [6] stated that substantially lower primary
ture effect. The concentration of the evaporated brine in- energy consumption could be achieved in thermal desalina-
creases from the first effect to the last. The layout advantage tion processes by the MEE principle. Horizontal tube MEE
is the absence of pumps for moving or rejecting brine from units can be operated economically with a very small driving
the effects. It has the disadvantage that all the feed has to be force and can be designed with a high gain ratio even for low
heated to the boiling temperature before boiling com- top brine temperatures. However, Rautenbach, et al., [7]
mences. In other words, a part of the heating steam when pointed out that the MEE process has the potential of com-
condensed does not accomplish any evaporation in the first peting with MSF with respect to simplicity and with RO with
or consecutive effects; thus, less vapor is generated from the respect to specific energy (or primary energy) consumption.
heating steam, which lowers the performance ratio. Further- However, although large units can be built, it remains an
more, the most concentrated brine is subjected to the coolest open question whether they will be ordered.
temperature, which reduces the heat transfer coefficient be- Michels [8] reported the following outstanding features
cause of the viscosity increase. However, at the same time, for the MEE process when combined with thermal vapor
this reduces the rate of scale formation. All these facts make compression (MEE-TVC). These features include low corro-
the system very suitable for high temperature application. sion and scaling, which is caused by low temperature opera-
Most of the published work on the MEE water desalina- tion (top brine temperature below 60 °C). Other features in-
tion process concentrated mainly on comparisons between it clude low energy consumption, short delivery time, and easy
and the MSF process. Darwish and El-Dessouky [2] cited operation and maintenance with proven reliability in the
that the race for the second generation of sea water desalters Gulf region. The cost of the plant erection, civil work, and
had been settled between reverse osmosis (RO) and MEE. the sea water intake is 35% cheaper than the MSF plants.
The MEE desalination system is more efficient from a ther- A limited number of publications which focus on the mod-
modynamic and heat-transfer point of view than the predo- eling and simulation of MEE desalination process are avail-
minant MSF desalination system. The pumping power and able in the literature. El-Dessouky and Assassa [9] carried
the specific heat transfer area required for the MEE system, out a simple analysis for the process, assuming a constant
in a respective way, are about 20% and 50% of that needed heat transfer area in each effect, constant physical properties
for the MSF plants. Moreover, instead of making partitions of water, and absence of feed heaters. They showed, for the
for 24 stages in the MSF plants, only partitions of 10 effects first time, the advantages of the MEE-TVC system over any
are made in MEE plants, having the same performance as other thermal desalination system. In pioneering work,
that of the MSF plant. Thus, the capital cost of the MEE sys- Darwish and El-Hadik [10] presented the basic characteris-
tem is expected to be about 50% less than that of MSF sys- tics, thermodynamic analysis, and the required heat transfer
tem. Minnich, et al. [3] reported that the MEE process re- area for the MEE desalination system. Their model was
quires lower heat-transfer areas than the MSF plants at high based on the following postulates:
top brine temperatures. ± Equal latent heat of evaporation in all effects.
Morin [4] carried out technical and economical compari- ± Equal specific heat for the brine, distillate, and feed.
sons of the low temperature MEE and the conventional ± The temperature drop between the effects is equal.
MSF process. The study showed that the MSF process needs ± Equal temperature increase across the feed heaters.
about half the heat transfer surface area required for the low ± The amounts of vapor formed by boiling in the effects are
temperature MEE process. The MEE process offers almost equal.
a 50% higher recovery than the MSF process at equal per- On the other hand, Hanbury [11] used the following as-
formance ratios. This results in reducing the size of the brine sumptions in his analysis:
circulating pumps, the piping systems, and the pretreatment ± The thermal loads in the second and subsequent effects
equipment. The electrical energy use for the pumping pro- are identical.
cesses favors the MEE process over the MSF system by a ± The heat transfer coefficient falls linearly from the second
factor of about 30% for the same performance ratio. Morin to the last effect.
[4] showed that the capital and operating costs for the opti- ± Constant specific heat and latent heat.

438 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00438 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

± The heat transfer coefficients and surface areas are equal 2 Process Description
for all feed preheaters.
The assumptions employed in the analysis developed by A process diagram for the forward-feed Multiple Effect
Minnich et al. [3] are: Evaporation (MEE) sea water desalination process is shown
± Here are no feed preheaters. in Fig. 1. The effects are numbered 1 to n from the left to
± The feed flow rate is the same for all effects. right (the direction of the heat flow)1) The feed and vapor
± The contribution of distillate flash is ignored. flow concurrently in the direction of the falling pressure.
± Constant heat transfer coefficients. The system consists of a number of evaporators, n, a series
± Constant heat transfer area for each effect. of feed water preheaters, a train of flashing boxes, last effect
± Vent mass and energy is negligible. or bottom condenser and a venting system. The number of
± Temperature losses are equal in all effects. effects, n, is closely allied to the relationship among the fixed
It is well established that the implementation of the as- charge and the steam operating cost. On the other hand, the
sumptions listed in the previous sections in modeling and upper number of effects is imposed by the difference be-
analysis of the MEE process can cause large discrepancies tween the heating steam temperature Ts and the boiling
between the model predictions and actual data. Accurate point in the last effect T1. The number of the feed preheaters
modeling of the MEE process is a key component in under- is n±2 while the number of flashing boxes is n±1. Each effect
standing the process elements better, improving future de- constitutes of a heat transfer area, vapor space, mist elimina-
sign procedures, and predicting the performance of existing tor, and other accessories.
plants over a wide range of possible operating conditions. The horizontal falling film evaporator is the most widely
The main features of the model developed in this paper are used in the MEE desalination process. The major advantage
summarized in the following points: of the horizontal falling evaporator is its ability to handle
± Adopt the practice case of constant heat transfer areas in sea water scaling, due to high wetting rates and efficient
the evaporators and feed preheaters in all effects. This is water distribution over the heat transfer surfaces by large
necessary to improve the economics and the construction spray nozzles. Thus, dry-patch formation or water maldistri-
procedures of the plant. bution is eliminated. This configuration offers the additional
± Consider the effect of vapor leak through the venting sys- advantages of positive venting and disengagement of vapor
tem. products and/or noncondensable gases, high heat transfer
± Model variations in the thermodynamic losses (boiling coefficients, and monitoring of scaling or fouling materials.
point elevation, non-equilibrium allowance inside the eva- A controlled amount of sea water (Mcw + Mf) is intro-
porators and the flashing boxes, temperature depression duced into the down condenser associated with the last ef-
corresponding to the pressure drop in the demister, vapor fect, where its temperature increases from the sea water
transmission lines, and during the condensation process) temperature Tcw to tf. A part of this water Mcw is rejected
from one effect to another. back to the sea. The function of circulating the cooling water
± Study the effect of boiling temperature, the velocity of Mcw in the last stage condenser is the removal of the excess
brine flowing through the tubes of feed heaters, the tube heat added to the system in the first effect. It is worth men-
material of construction, and the tube bundle geometry tioning that the evaporators do not consume most of the
on the required specific heat transfer area. supplied heat, it simply downgrades it. The remaining part
± Consider the effect of water temperature and salinity on of sea water Mf at tf is chemically treated, deaerated, and
the water physical properties such as density, latent heat pumped through the preheaters. Feed preheating is an im-
of evaporation, viscosity, Prandtl number, and specific portant means of improving the system performance ratio.
heat at constant pressure. The temperature of the feed water increases from tf to t2 as
± Weight the effect of the presence of noncondensable gases it flows inside the tubes of the preheaters. The heat neces-
on the heat transfer coefficients in the evaporators and sary to heat the feed water is supplied by condensing a por-
the feed heaters. tion of vapor formed in each effect at the shell side of the
There are some software packaged that can be used for preheaters. The feed water Mf is sprayed at the top of the
the design and rating of the MEE process. In many situa- first effect, where it falls in the form of thin film down the
tions, however, these packages have limited capacities to de- succeeding rows of tubes arranged horizontally. Within this
termine necessary details, which are specific to local condi- effect, the brine temperature raises to the boiling tempera-
tions. In addition, these packages are not available or easily ture corresponding to the pressure in the vapor space T1 be-
accessible to many engineers in the field, not easy to use, fore a small portion of vapor D1 is evaporated. The heat re-
and difficult to detect the errors. Moreover, the model on quired to preheat the feed and for evaporating D1 is
which these packages were built and the assumptions in- released by condensing a controlled mass of saturated steam
volved are not presented in details (black box syndrome).

±
1) List of symbols at the end of the paper.

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00439 $ 17.50+.50/0 439
 Full Paper

Figure 1. Process flow diagram for the multiple effect evaporator (MEE) desalination process.

Ms inside the tube bundle. The steam is supplied to the sys- BPE and temperature depression corresponding to pressure
tem from an external boiler. The high quality condensate loss in the demister DTm, transmission lines DTt, and during
from the first effect is returned to the boiler. the condensation process DTc reduces the available driving
The temperature of the vapor formed in the first effect Tv1 force for heat transfer in the evaporators and the preheaters.
is less than the boiling temperature T1 by the boiling point Thus, it is necessary to provide excess surface areas to com-
elevation (BPE)1. The vapor generated therein flows pensate for these temperature degradations. In other words,
through a knitted wire mist separator known as a wire mesh the temperature losses present an extra resistance to the
demister to remove the entrained brine droplets. The satura- flow of heat between the condensing steam and the boiling
tion temperature of the vapor departing the demister is less sea water. Nonetheless, the temperature downgrading does
than that of the formed vapor due to the frictional pressure not influence the plant thermal performance ratio PR or
loss in the demister. The vapor flows from the demister have steam economy. The plant performance ratio depends on the
to be transported to the second effect. This transport inevita- heat balance consideration and not on the rate of heat trans-
bly involves a pressure drop and hence a corresponding de- fer.
crease in the saturation temperature. Another pressure fall The vapors formed by flashing inside the effect and the
and consequent depression in the saturation temperature of flashing box (d2 + d2) are condensed on the shell side of the
the vapor is associated with vapor condensation inside the preheaters. The heat given up results in a heating of the
heat transfer tubes in the evaporators or over the heat trans- brine flowing inside the preheater tubes. The distillate con-
fer area in the preheaters. The latent heat of condensation of densed in the preheater (d2 + d2) is carried with the vapor
D1 is exploited for further evaporation in the second effect. formed by boiling D2 into the tubes of the third effect. Al-
The remaining unevaporated brine in the first effect (Mf± though the vapors formed by evaporation and flashing are
D1) goes to the second effect, which operates at a lower drawn separately in the flow diagram to show the process,
pressure. they flow from the evaporator and the flashing box to the
The vapor is formed inside the second effect by two differ- feed heater in the same line.
ent mechanisms. First, by boiling over the heat transfer sur- The processes that take place in the second effect are re-
faces D2. Second, by flashing or free boiling within the liquid peated in each effect all the way down to the last. Fig. 2
bulk d2. The temperature of the vapor formed by flashing shows the different processes that take place inside effect
T¢v2 is less than the effect boiling temperature T2 by the boil- number (i). It is worth mentioning that the amount of steam
ing point elevation (BPE)2 and the nonequilibrium allow- generated by evaporation in each effect is less than the
ance (NEA)2. Another small quantity of vapor d2 is formed amount generated in the previous effect. This is due to the
in the flashing box due to the flashing of distillate condensed increase in the specific latent heat of vaporization with the
in the second effect D1. The flashed off vapor d2 is produced decrease in the effect temperature. Consequently, the
at a temperature T²v2 which is lower than the condensation amount of vapor generated in an evaporator by boiling is
temperature of distillate Tc1 by the nonequilibrium allow- less than the amount of condensing steam used for heating
ance (NEA)2. The flashing boxes offer a way to recover heat in the following evaporator. The unevaporated brine flowing
from condensed fresh water. The boiling point elevation into the last effect reaches its final concentration Xn by eva-

440 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00440 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

mass of fresh product water or the thermal performance ra-

tio PR, the required specific heat transfer surface area sA,
the temperature profile in various effects Ti, and the amount
of cooling water flowing to the condenser of the last effect
Mcw. The analysis of MEE systems is based on developing a
steady state mass and heat balances coupled with the heat
transfer rate equations for each individual effect and con-
joining them with the ratio between the mass of feed to that
of product fresh water. In most cases, the number of equa-
tions which can be developed are equal to the number of un-
known variables.
The dry saturated or very slightly superheated steam ad-
mitted into the first evaporator is used to increase the tem-
perature of the feed brine Mf from the inlet temperature t2
to the boiling temperature T1, supply the latent heat re-
Figure 2. Different processes in effect (i).
quired to evaporate a specified mass of vapor D1, and ac-
porating more vapors. The remaining brine Mb is rejected to count for heat losses to the surroundings q. This is
the sea. The vapor formed by boiling and flashing in the last
effect (Dn + dn) and in the end flashing box dn passes to the (1)
bottom condenser.
The condenser and the brine heaters must be provided Where Cp is the specific heat at constant pressure of the
with good vents, first for purging during start-up and then brine and l is the latent heat of evaporation. The specific
for removing noncondensable gases, which may have been heat depends on water temperature and salinity, while the
introduced with the feed or drawn in through leaks to the latent heat relies only on the boiling temperature. The gen-
system. The presence of the noncondensable gases not only erated vapor has a saturation temperature Tv1 corresponding
impedes the heat transfer process but also reduces the tem- to the pressure in the evaporator vapor space. This tempera-
perature at which steam condenses at the given pressure. ture is less than the boiling temperature T1 by the boiling
This occurs partially because of the reduced partial pressure point elevation (BPE)1; or
of vapor in a film of poorly conducting gas at the interface.
To help conserve steam economy venting is usually cascaded (2)
from the steam chest of one preheater to another. The ef-
fects operate above atmospheric pressure are usually vented The boiling point rise caused by the hydrostatic head DTh
to the atmosphere. The noncondensable gases are always sa- has a negligible effect in the horizontal falling film evapora-
turated with vapor. The vent for the last condenser must be tors. The boiling point elevation BPE at a given pressure is
connected to vacuum-producing equipment to compress the the increase in the boiling temperature due to the salts dis-
noncondensable gases to the atmosphere. This is usually a solved in the water. The boiling point elevation, BPE, is cal-
steam jet ejector if high-pressure steam is available. Steam culated from the following empirical formula.
jet ejectors are relatively inexpensive but also quite ineffi-
cient. Since the vacuum is maintained on the last effect, the (3)
unevaporated brine flows by itself from effect to effect and
only a blow down pump is required in the last effect. with
It is interesting to note that the MSF process, the most fre-
quently encountered in standard desalination practice, is a (3a)
special case of the MEE system. This occurs when the entire
vapor formed in the effects is needed to preheat the feed in
the preheaters and none is available for the evaporator Where T is the temperature in °C and X is the salt concen-
tubes. In this case, the brine heater, distillate collecting trays, tration in ppm. The above equation is valid over the follow-
and the heat rejection section in MSF are replaced by the ing ranges: 20000 < X < 160000 ppm, 20 < T < 180 °C.
first effect, flashing boxes, and bottom condenser in the The dimensions of the evaporator heat transfer surface
MEE process. area Ae is obtained from the amount of transferred heat Qe,
the overall heat transfer coefficient Ue, and the temperature
difference of condensing steam Ts and the boiling brine. This
3 System Analysis is

The main data usually required for the analysis of the (4)
MEE system are the amount of steam consumed per unit

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00441 $ 17.50+.50/0 441
 Full Paper

The effects of steam superheating and condensate sub- Where hl is the heat transfer coefficient, when all the flow-
cooling have negligible roles on the heat transfer process in ing mass is liquid, and is calculated by the well-known Dit-
the evaporators of the MEE system. tus-Bolter equation,
The most critical step in determining the heat transfer
area is the settlement of the overall heat-transfer coefficient (10)
U. The overall heat transfer coefficient based on the outside
surface area Uo is related to the individual thermal resis- The above correlation is valid over the following ranges:
tance by the following expression. 2.8 £ di £ 40 mm, 21 £ TV £ 355 °C, 0 £ w £ 1, 0.158 £ q £
16000 kW/m2, 11 £ G £ 4000 kg/m2 s, 0.7 £ P £ 1 bar, 0.0019 £
Pr £ 0.82, 350 £ Re £ 100000. The average heat transfer coef-
(5) ficient is obtained by linear interpolation between the values
of local heat transfer coefficient hi at the values of w ranging
from 0 to 1.
Where h is the heat transfer coefficient, Rf is the fouling Proper venting of the evaporators reduces significantly
resistance, kw is the thermal conductivity of tube material, the impairing effects of the noncondensable gases on the
and r is the radius. The subscripts i and o refer to the inner condensation heat transfer coefficient. Continuous withdra-
and outer tube surface, respectively. The evaporator heat wal of gases prevents their accumulation and minimizes their
transfer surface area Ae is usually, but not always, taken as effect on the heat transfer coefficient. A decrease of less
that in contact with the boiling liquid, whether on the inside than 5% occurs in the heat transfer coefficient for a gas con-
or the outside of the tubes. centration of 10% in the vent stream. Standiford [14] mod-
Han and Fletcher [12] developed the following experi- eled the effect of noncondensable gases as an additional
mental correlation to calculate the boiling heat transfer coef- fouling resistance, with a value of 6.5 ´ 10±5 (m2 °C/W), which
ficient ho for thin water film flowing over the outside of is then multiplied by the volume percentage of the gases. In
smooth horizontal tubes, water desalination plants, the volumetric concentration of
noncondensable gases is about 4% [15].
Boiling and flashing mechanisms form the vapor in the
(6) second effect. Boiling takes place over the outer surface of
the heating surface. Flashing or free boiling occurs within
the liquid bulk. The mass of vapor generated by boiling is
In the above equation Re and Pr are Reynolds and calculated from
Prandtl numbers, respectively, q² is the heat flux, m is the
viscosity, r is the density, and k is the thermal conductivity (11)
of the fluid. The above correlation is valid over the following
ranges: 770 £ Re £ 7000, 1.3 £ Pr £ 3.6, 30 £ q² £ 80 kW/m2, The amount of vapor flashed off from the brine flowing to
and 49 £ T £ 127 °C. The maximum deviation for the equa- the second effect d2 is estimated from
tion is ± 10%.
There are a wealth of corrections in the literature that can (12)
be used to calculate the heat transfer coefficient of conden-
sation inside a horizontal tube for a particular flow pattern. Where T2© is the temperature to which the brine (Mf±D1)
However, the correlations, which can be used for all flow cools down as it flows to the second effect. This temperature
patterns, are limited. Perhaps the most verified predictive is higher than the boiling temperature by the nonequilibrium
general technique available for all flow regimes in horizontal allowance (NEA¢)2. That is
tubes is the following correlation of Shah [13]:
(13)
(7)
The nonequilibrium allowance is a measure for the effi-
The parameter Z is defined as: ciency of the flashing process. Miyatake, et al. [16] devel-
oped the following equation to correlate data obtained from
flash evaporation experiments in a pool of pure water, which
(8) simulates to the flashing processes inside the evaporators
and the flashing boxes,

Where w is the vapor mass fraction. The local superficial

heat transfer coefficient hf is calculated from (14)

(9) with

442 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00442 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

(15) Where Z is the length of the pipe, di is the pipe inside di-
ameter, and Mv is the vapor mass flow rate. The pressure
and drop during the vapor condensation inside the evaporator
tubes is the sum of the frictional DPr, gravitational DPg, and
(16) acceleration DPa components. That is

On the other hand, the mass of vapor formed by flashing (23)

in the flashing box of the second effect d2 is given by
The two terms on the right hand side of Eq. (23) have op-
posite signs. The first gives a fall in pressure due to wall fric-
(17) tion, while the second represents a rise in pressure because
of the pressure recovery from the flow deceleration and the
with gravitational force.
For the condensation inside horizontal tubes, the gravita-
(18) tional component of the pressure drop DPg is equal to zero.
However, it is usual to design the condenser with a small an-
The condensation temperature of vapor inside the tube gle of inclination such that the condensate tends to run out
bundle of the second effect Tc1 is less than the boiling tem- of the end of the tubes at the opposite end to the steam inlet.
perature in the first effect T1 by the boiling point elevation This makes the flow much more stable than if the tubes are
(BPE)1 and the saturation temperature depressions asso- horizontal. It also improves the efficiency of the venting sys-
ciated with the pressure loss during the vapor flow in the tem [17]. This component of pressure drop is estimated from
demister (DTm)1, vapor transmission lines (DTt)1 and vapor this expression:
condensation inside the horizontal tubes (DTc)1. Thus,
(24)
(19)
Where a is the vapor phase void fraction, Z is the pipe
In general, the pressure loss during the vapor flow through length, and y is the inclination angle. There are many corre-
a wire mesh pad, which is widely used as the mist eliminator lations for the void fraction a. The one suggested most fre-
in water desalination industry, is relatively small because of quently in the literature for condensation in tube is by Zivi
the high void fraction of these pads. Losses in the mist se- [18].
parator are about 10 cm of water for 0.15 m. For the dry
steam (free of any entrained droplets), the pressure loss
through the mesh pad is given by
(25)

(20)

The acceleration pressure drop is calculated from this for-

Where V is the vapor superficial velocity (m/s), a is the mula
specific area per unit volume (approximately = 85±115 m2/
m3), rv is the vapor density, e is the bed void fraction (varied
from 0.97 to 0.99), Lm is the mesh pad thickness, and f is the
friction factor. The value of f depends mainly on the extent (26)
of the modified Reynolds number as follows

(21)
The subscripts (1) and (2) refer to the inlet and outlet con-
The pressure fall due to the vapor flow in the steam line ditions, respectively.
connecting the vapor space of the first effect and the con- The two-phase pressure losses due to friction (dPr/dZ) are
denser tubes of the second one can be calculated from the generally expressed as a function of the corresponding sin-
Unwin formula gle-phase pressure losses, which is multiplied by a correction
factor f21. That is

(22)
(27)

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00443 $ 17.50+.50/0 443
 Full Paper

Friedel [19] developed the following correlation for calcu-

lating f21 (37)

(28) with

(38)
Where
After rearrangement,
(29)

(39)
(30)

The pressure of saturated vapor at Tv1 is predicted from

(31) the steam table or from equation (A±1), which is obtained
by fitting the steam table data.
The vapor formed by flashing inside the evaporator and
(32) the flashing box (d2 + d2) is used to heat the feed water Mf
flowing inside the tubes of the feed heater. The heat balance
around the feed heater yields:
(33)
(40)

Where fg and fl are the friction factors for the total mass Where t2 and t3 are the outlet and inlet temperatures of
flux flowing with vapor and liquids, respectively, G is the the feed preheater respectively, l¢vc2 and l²vc2 are the latent
mass flux, m is the dynamic viscosity, and s is the surface ten- heat of condensation at T¢c2 and T²c2, respectively, and Z2 is
sion. The density of the two phase mixture rTP is defined as the feed heater thermal efficiency. This efficiency accounts
for the heat loss to the surrounding and the vapor escape
with the vented noncondensable gases. The condensation
(34) temperatures T¢c2 and T²c2 are calculated from the following
equations:

Hewitt [20] recommended the use of Friedel equation (41)

when the value of (ml/mv) is less than 1000. In the MEE sys-
tem this ratio ranges from 65.12 at T = 315 K to 19.856 at T and
= 385 K.
The frictional pressure drop term usually calculated in a (42)
stepwise manner. The tube is divided into a number of short
lengths DZ over which the conditions change only moder- Where DTm and DTc are the temperature losses corre-
ately. On the other hand, assuming that the condensation sponding to the pressure drop associated with the vapor flow
process takes place completely inside the tubes for w1 = 0 through the demister pad and vapor condensation outside
and w2 = 1 yields the tubes of the preheater. Eq. (20) predicts the pressure
drop due to the vapor flow through the demister. On the
(35) other hand, the pressure drop due to the vapor flow over the
preheater tubes can be calculated, at best, only roughly be-
and cause changing velocity and flow pattern during condensa-
tion process [21]. This overall pressure drop associated with
(36) the vapor condensation process is the algebraic sum of pres-
sure losses due to the vapor flow in the nozzles and headers,
static pressure head, two phase friction-loss, and momentum
The temperature depression or the vapor saturation tem- change. The static head and baffle losses are insignificant in
perature decrease (Tv1 ±Tc1 = DTt1), because of the pressure the condensers. The momentum change or flow slow down
drop in the demister DPm, vapor transmission line DPt, and during condensation results in a pressure recovery [22]. The
during condensation process DPc, can be calculated using magnitude of this pressure recovery is high in vacuum opera-
the Clausius-Clapeyron equation as follows: tion. The pressure regain can approach or exceed the friction

444 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00444 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

loss [23]. Since most effects in MEE plants operate at va-

cuum, it seems reasonable to assume that the pressure recov- (48)
ery due to flow slow down can compensate for the friction
pressure drop component; therefore, the net pressure fall
and consequently the saturation temperature depression in The correction factors C1 and C2 consider the influence of
the condensation process can be neglected. the condensate dripping down and the presence of noncon-
The heat transfer between the condensing vapor and the densable gases, when they constitute less than 4% by weight,
sea water in the feed preheater of the second effect can be on the condensation heat transfer coefficient, respectively.
written in terms of an overall heat transfer coefficient Up, The size of the coefficients C1 and C2 are given by the fol-
preheater heat transfer area Ap, and the logarithmic mean lowing equations [26]
temperature difference LMTD, thus:
(49)

(43)
(50)

Neglecting the small difference between T¢c2 and T²c2, the Where Xnc is the percentage weight of the noncondensa-
(LMTD)2 is defined as: ble gases and N is the number of tube rows in the vertical di-
rection inside the condenser. The value of N depends on the
total number of tubes Nt, tube arrangement pitch Pt, number
(44) of tube passes, and nozzle diameter. It is customary practice
to arrange the tubes in the feed heaters with a square pitch
pattern to provide adequate mechanical cleaning for the out-
Combining Eqs. (43) and (44) produces er surface of the tubes. In this arrangements each four tubes
occupy an area of (4 Pt2) and the number of tubes in the ver-
tical direction is 2. Thus, the total number of tubes, which
(45) can be installed in a shell of diameter Ds and with a pitch of
Pt, can be approximated by this equation:

Where NTU2 is the number of transfer units. The above (51)

equation can be solved for the outlet temperature of feed
water to give: The following relation determines the number of tubes in
the vertical direction N to the shell diameter and pitch:
(46)
(52)
The term (T¢c2 ± t2) is the preheater terminal temperature
difference, and its value has a strong impact on the preheater For tubes arranged in an equilateral triangular pitch, the
heat transfer area. following equation can be used
Eq. (4) can be used to relate the overall heat transfer coef-
ficient Up to the individual coefficients. The inside tube (53)
heat-transfer coefficient hi is calculated from the empirical
formula developed by Wangnick [24] especially for desalina- The total number of tubes in the feed heater is calculated
tion plants. by

(54)

(47) Where Z is the tube length, do and di are the tube outside
and inside diameters, respectively, and V is the feed water
velocity. The value V is limited at the top end by erosion
damage to the tube materials and excessive pumping costs
Where Xf is the salt concentration in ppm, Tf is the film and at the bottom end by higher fouling rates and the need
temperature, and di and do are the inside and outside tube to maintain high side heat transfer coefficients. It ranges, in
diameters, respectively. thermal desalination units, between 1.3±2.2 m/s.
Henning, et al. [25] developed the following equation to Differences of the second effect and other effects, i.e., ef-
calculate the heat transfer coefficient during vapor conden- fect number i are the brine flow rate to the second effect is
sation outside the tubes. (Mf±D1) while it is

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00445 $ 17.50+.50/0 445
 Full Paper

(63)

for stage number (i) and the mass flow rate of distillate en- The heat transfer area for the feed heater
tering the flashing box of the second effect is D1 while it is

(64)

for effect number i.

Taking these differences into consideration and following The outlet temperature of feed water flowing inside the
the same procedure presented in the previous sections, the feed preheater tubes is
following equations can be developed for effect number i.
The mass of vapor formed by evaporation Di is (65)

(55) There are two basic distinctions between the feed prehea-
ters of any effect and the last effect condenser. The first one
The amount of vapor flashed off from the unevaporated is that the mass of vapor condensed in the shell side is not
brine di is only the vapor formed by flashing inside the last evaporator
and flashing box (dn + n), but also the vapor formed by boil-
ing Di. The other difference is that the mass of water flowing
(56) inside the tubes of the last condenser is the feed water plus
the cooling water (Mcw+Mf). Accordingly, the following re-
lationships can be developed for the last effect condenser.
with

(57) (66)

and
Where Tcw is the sea water temperature and Mcw is the
(58) mass flow rate of cooling water. Arranging the above two
equations and substituting the value of (LMTD)cw from Eq.
The mass of vapor formed by flashing in the flashing box di is (44) produces

(59) (67)

and
with
(68)
(60)
On the other hand, the overall mass balances around the
and plant assuming that the product fresh water is free of salt
(Xd = 0) gives
(61)
(69)
The concentration of salt in the effect is
and

(62) (70)

The thermal performance PR of the MEE plant is defined

as the mass of distillate water produced per unit mass of
The heat transfer surface area is heating steam used. That is:

446 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00446 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

(71)

Where Md is the total mass of distillate formed by boiling

and flashing in all effects and flashing box. Thus:

(72)

The mass of heating steam Ms is calculated from Eq. (1).

The exercise of the basic design requirement of constant sur-
face area for the preheater and the evaporator in all effects
produces

(73)

and

(74)

4 Computational Algorithm

The model developed in the preceding section is com-

posed of a set of highly nonlinear equations. Thus, the use of
any numerical procedure to solve them such as the well-
| |
s s
known Newton-Raphson, TDM, and like methods is tedious,
because it requires the linearization and/or differentiation a
| |
high number of nonlinear relationships. The modified fixed
point iteration procedure developed by El-Dessouky and
Bingulac [27] is a simple, but yet powerful and convergent
algorithm. According to this technique, the whole algorithm
consists of 10 calculation blocks denoted by B through B10
and 6 logical blocks expressed by L1 through L6 as shown in
Fig. 3. The algorithm is implemented using the L-A-S Com-
puter Aided Language [28].
To facilitate the understanding of the algorithm, calcula-
tions of some variables in the following flowchart are repre-
sented symbolically by corresponding functions (subrou-
tines). These variables are l, m, rl, rv, and Pr; LMTD, Eq.
(44); ho and hi for the preheater sections, Eqs. (47 and 48),
these functions are called hop and hip, respectively; ho and
hi for the evaporator sections, Eqs. (6 and 7), these functions
are called hoe and hie, respectively; and Dtl the total thermo-
dynamic losses.
The first block B1 defines all necessary geometrical con-
stants of the plant and design values as well as the initial
guesses for subsequent iterative calculations. The most outer
loop consisting of blocks B2 through B7 and L6 performs se-
quential effect-by-effect solution of equations describing the
MEE system. The calculation process starts at the lower
temperature side, i.e., effect n.
The block B2 sets the initial guesses for the preheater in
effect i. The loop consisting of the blocks B3 and L1 solve
iteratively the equations describing the preheater of effect i.
This involves calculations of outlet temperature from the Figure 3. Computational Algorithm.

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00447 $ 17.50+.50/0 447
 Full Paper

preheater of effect i, i.e., t2 = ti, given the temperature t3 =

ti+1 at the inlet of the preheater of effect i.
The loop consisting of blocks B2 ± B5 and L3 iteratively
adjusts the temperature T2 = T, at the outlet of the ith effect
of the evaporator, so that the temperature difference Ti±1 ±
Ti is equal to the difference Z(ti ± ti+1), as required by the
heat balance. The blocks B7, B8, and L5 constitute the loop
where iterative solution of equations describing the ith effect
of the evaporator. In other words, the quantity D satisfying
Eqs. (11, 73, and 74) is calculated.
Block B9 stores the results of the ith effect of the evapora-
tor as well as sets the initial guesses for the next effect of
both the preheater and evaporator. Finally, in the block B10
overall plant effectiveness parameters are calculated and
stored, which include the plant thermal performance PR, the
specific heat transfer area sA, and the specific cooling water
flow rate sMcw.

Figure 4. The overall heat transfer coefficients in the evaporator and the con-
5 Results denser.

The model developed and the suggested algorithm are N s/m2. Moreover, the salt concentration in the forward feed
used to obtain the data required from modeling the MEE MEE system increases with the boiling temperature de-
desalination process. The parameters needed in process opti- crease. The sea water viscosity is directly related to the salt
mization and product water cost calculations are the plant concentration. The rate of heat transfer is impeded as the
thermal performance PR, the specific heat transfer surface viscosity increases.
area sA, and the specific flow rate of cooling water sMcw. The variation of the plant thermal performance ratio with
These parameters are computed as a function of the most both the top brine temperature and the number of effects is
important design variables, namely, the top brine tempera- illustrated in Fig. 5. As can be seen, the performance ratio is
ture T1 and the number of effects. The main assumptions nearly independent of the top brine temperature and is
upon which these relationships were generated are strongly related to the number of effects. The performance
± The sea water temperature Tcw and salinity Xcw are 25 °C ratio is measured by the kg of distillate water produced per
and 42000 ppm. kg of heating steam condensed in the first effect. In n-effect
± The temperature of rejected cooling water Tf is 35 °C. evaporator system, there is approximately n kg of water pro-
± The boiling temperature in the last effect Tn is 40 °C. duced per kg of steam supplied. In other words, the number
± The sum of the heat transfer resistances due to the tube
material, fouling inside the tube, and fouling outside the
tube is 731 ´ 10±6 m2 °C/W.
± The thermal efficiency of the preheater Zi is 90%.
± The tubes outside diameter do is 31.75 mm and the inside
diameter di is 19.75 mm.
± The brine velocity inside the preheater tubes V is 1.55 m/s.
Fig. 4 shows the effect of the boiling temperature of the
brine on the overall heat transfer coefficients in the evapora-
tor and the feed preheater. The heat transfer coefficient is
augmented with the boiling temperature; however, the heat
transfer coefficient in the evaporator is always higher than
that of the feed preheater. The heat transfer in the evapora-
tor is associated with change of phases on the two sides of
the tubes. In the preheater, steam condenses in the shell side
and the feed water is heated by single phase forced convec-
tion inside the tubes. The increase of the heat transfer coeffi-
cient with the boiling temperature can be attributed to the
effect of boiling temperature on the water physical proper-
ties, specially the dynamic viscosity. The viscosity of the Figure 5. Effect of the top brine temperature and number of effects on the
water at 40 °C and 100 °C are 665 ´ 10±6 N s/m2 and 279 ´ 10±6 plant thermal performance.

448 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00448 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

of effects is directly allied to the number of reusing the heat-

ing steam in generating nearly the same amount of this
steam.
Fig. 6 depicts the impact of the top brine temperature and
the number of effects on the specific heat transfer area. The
specific heat transfer area or the heat transfer area required

Figure 7. Effect of top brine temperature and number of effects on the speci-
fic flow rate of the cooling water.

minishes with respect to the total plant capacity as the num-

ber of effects increases. The result of this is the reduction of
the thermal load of the condenser and consequently the re-
quired cooling water flow rate per unit mass of fresh water
Figure 6. Effect of top brine temperature and number of effects on the speci- produced.
fic heat transfer area.
It is necessary to emphasis that the increase of the top
brine temperature from 60 °C to 100 °C decreases the re-
per unit mass of product water increases by reducing the top
quired specific heat area tremendously. This behavior is
brine temperature and raising the number of effects. The
more pronounced with a high number of effects. For exam-
overall heat transfer coefficient and the available driving
ple, the required specific heat transfer areas at 100 °C are
force for heat transfer affect the specific heat transfer area
30.3% and 26.1% of that required at 60 °C when the number
for heat transfer in each effect. The heat transfer coefficients
of effects are 6 and 12, respectively. The top brine tempera-
in the evaporator and the feed preheater are improved with
ture has minor effects on the plant performance ratio and
the boiling temperature as shown in Fig. 4. Furthermore, the
the specific flow rate of cooling water.
total effective temperature difference between the heating
steam and the bottom condenser is directly related to the
top brine temperature. Note that in all calculations, the boil-
6 Design Correlations
ing temperature at the last effect is kept constant at 40 °C.
On the other hand, the increase in number of effects is con-
The model predictions are used to develop the following
nected with a reduction in the temperature difference for
correlations. These correlations are useful in the design and
heat transfer across the heat transfer area by a factor ap-
optimization of the MEE desalination plants.
proximately equal to the number of effects. That is, the in-
crease of required specific heat transfer area is the compen-
(75)
sating disadvantage to the increase in thermal performance
resulting from increasing the number of effects.
Fig. 7 displays the effects of the top brine temperature
(76)
and the number of effects on the specific flow rate of cooling
water. The figure shows that the specific flow rate of cooling
is nearly constant at different values of top brine tempera-
ture and decreases at a high rate as the number of effects
(77)
goes up. The cooling water is required mainly to condense
the vapor formed by boiling in the last effect. The total capa-
city of an MEE desalination system remains basically un-
changed if the number of effects is varied and if the total
temperature difference is kept constant. This means that the
(78)
fraction of the vapor formed by boiling in the last effect di-

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00449 $ 17.50+.50/0 449
 Full Paper

M [kg/s] Mass flow rate

(79) n [±] Total number of effects
N [±] Number of rows of horizontal tubes
NT [±] Total number of tubes
The standard deviations for the above equations are 1.76 , NTU [±] Number of transfer units
2.03 , 2.4, 5.67, and 1.6%, respectively. NEA [°C] Nonequilibrium allowance
P [kPa] Pressure
Pt [m] Tube pitch
7 Conclusions PR [±] Thermal performance ratio, kg of
distillate/kg of heating steam
The present study introduces an efficient and accurate DP [Pa] Pressure drop due to flow acceleration
mathematical model describing the Multiple Effect Evapora- q [W] Heat loss to the surroundings
tion desalination system. The following conclusions are made q² [W/m2] Heat flux
in the light of results and analysis performed in this study: Q [W] Heat transfer rate
± The heat transfer coefficients in the evaporators and the R [kJ/kg °C] Universal gas constant
preheaters augmented with boiling temperature. r [m] Tube radius
± The heat transfer coefficients in the evaporator are always Rf [m2 °C/W] Thermal resistance of the fouling
high when they are compared to that in the preheater at sA [m2/(kg/s)] Specific heat transfer area
the same boiling temperature. sMcw [m2/(kg/s)] Specific heat transfer area
± The plant thermal performance ratio is nearly independ- t [°C] Temperature of sea water flowing
ent of the top brine temperature and strongly related to inside the feed heaters
the number of effects. T [°C] Temperature of brine boiling in the
± The specific heat transfer area increases by reducing the effects
top brine temperature and raising the number of effects. T¢ [°C] Temperature of vapor formed by
± The effect of top brine temperature on the specific heat flashing inside the effects
transfer area is more pronounced at high number of effects. T² [°C] Temperature of vapor formed by
± The specific flow rate of cooling water is nearly constant flashing inside the flashing boxes
at different values of different top temperature and de- DT [°C] Temperature decrease corresponding
creases at high rate as the number of effects is increased. to pressure drop
Received: July 8, 1997 [CET 929] U [W/m2 °C] Overall heat transfer coefficient in the
evaporator
V [m/s] Velocity
Symbols Used X [ppm] Salt concentration
Z [m] Pipeline length
a [m2/m3] Specific surface area per unit volume
of the demister
Greek Symbols
A [m2] Heat transfer surface area
BPE [°C] Boiling point elevation
a [±] Vapor phase void fraction
Cp [kJ/kg °C] Specific heat at constant pressure of
w [±] Vapor phase mass fraction
sea water
d [m] Tube diameter
Cp [kJ/kg °C] Specific heat at constant pressure of
e [±] Wire mesh void fraction
distillate water
Z [±] Thermal efficiency
d [kg/s] Mass of vapor formed by flashing
m [kg/sm] Dynamic viscosity
inside the effects
n [m2/s] Kinematic viscosity
d [kg/s] Mass of vapor formed by flashing in
f [rad.] Angle of inclination
the flashing boxes
r [kg/m3] Mass density
D [kg/s] Mass of vapor formed by boiling
rTP [kg/m3] Density of the two phase mixture
Ds [m] Shell inside diameter
s [N/m] Surface tension
f [±] Friction factor
g [m/s2] gravitational acceleration
G [kg/m2s] Mass flux Dimensionless Numbers
h [W/m2 °C] Heat transfer coefficient
k [W/m °C] Thermal conductivity. Nu [±] Nusselt number, hD/k
L [m] Length Pr [±] Prandtl number, Cp m/k
LMTD[°C] Logarithmic mean temperature Re [±] Reynolds number, rVD/m
difference We [±] Weber number, rV2L/s

450 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00450 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1998) 5
 Full Paper

Subscripts

1,2,...,n Effect number [21] Sinnott, R.K., Coluson & Richardson's Chemical Engineering, Vol. 6,
2nd ed., Butterworth Heinemann, Oxford, 1996.
a Acceleration component [22] Marto, P.J., Heat Transfer in Condensers, in: Boilers, Evaporators and
b Reject brine Condensers, (S. Kakac, Ed.), John Wiley, NY 1991.
c Condensate or condenser [23] Muller, A.C., Condensers, in: Hemisphere Handbook of Heat Exchan-
ger Design, (G. F. Hewitt, Ed.) Hemisphere, NY 1991.
cw Cooling sea water [24] Wangnick, K., How Incorrectly Determined Physical and Constructio-
d Product fresh water nal Properties in the Sea Water and Brine Regimes Influence the Design
and Size of an MSF desalination plant ± stimulus for further thoughts,
e Evaporator Proceedings of the IDA World Congress on Desalination and Water Sci-
f Feed sea water ence, Abu Dhabi, November, 1995, Vol. II, pp. 201-218.
g Gravitational component [25] Hennig, S; Wangnick, K., Comparison of Different Equations for the
Calculation of Heat Transfer Coefficients in MSF Multi-Stage Flash
i Inside the tubes Evaporators, Proceedings of the IDA World Congress on Desalination
l Liquid phase and Water Sciences, Abu Dhabi, November, 1995, Vol. III, pp. 515-524.
m Demister [26] Oak Ridge National Laboratory (ORNL), United States Department of
the Interior, Research and Development Progress Report No. 315, De-
o Outside the tubes cember 1967.
p Preheater [27] El-Dessouky, H.T.; Bingulac, S., Computer Methods in Applied Mecha-
nics and Engineering 141 (1997) pp. 95-115.
r Frictional component [28] Bingulac, S.; Van Landingham, H.F., Alogrithms for Computer Aided
s Heating steam Design of Multivariable Control Systems, Marcel Dekker, NY 1993.
t Transmission line component
v vapor phase
w Tube wall APPENDIX: Physical Properties of Water and
Water Vapor
1 Vapor Pressure of Saturated Water
References
[1] IDA, Int. Desalination Association, Gnarrenburg, Worldwide Desalting (A.1)
Plants Inventory, Wangnick, K., Report No. 13, December, 1995.
[2] Darwish M.A.; El-Dessouky, H.T., Applied Thermal Engineering 18
(1996) pp. 523-537.
[3] Minnich, K.; Tonner, J.; Neu, D., A Comparison of Heat Transfer Requi-
rement and Evaporator Cost for MED-TC and MSF, Proceedings of the
Where P is kPa and T is °C.
IDA World Congress on Desalination and Water Sciences, Abu Dhabi,
November, 1995, Vol. III, pp. 233-257. 2 Saturation Temperature
[4] Morin, O.J., Desalination 93 (1993) pp. 69±109.
[5] Hess, G.; Morin, O.J., Desalination 87 (1992) pp. 55±68.
[6] Rautenbach, R., Desalination 93 (1993) pp. 1±13. (A.2)
[7] Rautenbach, R.; Widua, J.; Schafer, S., Reflections on Desalination Pro-
cesses for the 21st Century, Proceedings of the IDA World Congress on
Desalination and Water Sciences, Abu Dhabi, November, 1995, Vol. I, Where P is in kPa and T is in °C.
pp. 117±136.
[8] Michels, T., Desalination 93 (1993) pp. 111±118.
[9] El-Dessouky, H.T.; Assassa, G.R., Desalination 55 (1985) pp. 145±168.
3 Specific Volume of Water Vapor
[10] Darwish, M.A.; El-Hadik, A.A., Desalination 60 (1986) pp. 251±265.
[11] Hanbury, W.T., An Analytical Simulation of Multiple Effect Distillation
Plant, Proceedings of the IDA World Congress on Desalination and Wa-
ter Sciences, Abu Dhabi, November, 1995, Vol. IV, pp. 375±382. (A.3)
[12] Han, J.; Fletcher, L., Ind. Eng. Chem. Process Des. Dev. 24 (1985)
pp. 570±597.
[13] Shah, M.M., Heat Transfer, Pressure Drop, Visual Observations, Test
Data for Ammonia Evaporating Inside Tubes, ASHRAE Trans., Vol. Where V is in m3/kg and T is in °C.
84, Part 2, 1978.
[14] Standiford, F.C., Evaporators and Evaporation, Encyclopedia of Chemi-
cal Processing and Design (Meketta, Ed.) Vol. 20, Marcel Dekker, Inc.,
4 Specific Volume of Water
NY 1984.
[15] Office of Saline Water (OSW)±United States Department of the Inte- (A.4)
rior, Distillation Digest, Research and Development Progress Report No.
538, March 1970.
[16] Miyatake, O.; Murakami, K.; Kawata, Y.; Fujii, Heat Transfer Jpn. Res. 2 Where V is in m3/kg and T is in °C.
(1973) pp. 89±100.
[17] Whalley, P.B., Boiling-Condensation and Gas-Liquid Flow, Oxford Sci-
ence Publications 1987. 5 Latent Heat
[18] Zivi, S.M., J. of Heat Transfer 86 (1964) pp. 247±252.
[19] Friedel, L., New Friction Pressure Drop Correlations for Upward, Hori- (A.5)
zontal and Downward Two-Phase Pipe Flow, Proceedings of the HTFS
Symposium, Oxford 1979.
[20] Hewitt, G.F., Gas-Liquid, in: Handbook of Multiphase Systems, Hemis- In the above equation, T is the saturation temperature in
phere Publishing, NY 1982. °C and l is the latent heat in kJ/kg.

Chem. Eng. Technol. 21 (1998) 5 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/98/0101-00451 $ 17.50+.50/0 451