Desalination 353 (2014) 98–108

Contents lists available at ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Dynamic modeling of multi-effect desalination with thermal vapor

compressor plant
Mohammad Taghi Mazini ⁎, Alireza Yazdizadeh, Mohammad Hossein Ramezani
Department of Electrical Engineering Shahid Beheshti University, Tehran, Iran

H I G H L I G H T S

• Multi-Effect Desalination (MED) is one of the most common techniques that provides a considerable quantity of potable water.
• The aim is to obtain a dynamic model for MED plants in such a way that the behavior of the system in different working conditions is predicted.
• The model is simulated with the MATLAB/SIMULINK software to simulate the transient behavior of the MED plants under various conditions.
• This model is validated with actual data from an industrial plant that operate in the South of Iran.

a r t i c l e i n f o a b s t r a c t

Article history: This paper aims to develop a mathematical dynamic model for multi-effect desalination (MED) with thermal
Received 14 September 2013 vapor compressor plant. The developed model is based on coupling the dynamic equations of material, salt
Received in revised form 11 September 2014 and energy balance of the system. Towards this, the plant is divided into three main subsystems; evaporators (ef-
Accepted 14 September 2014
fects), condenser and thermo-compressor. Moreover, each effect is considered to be a dynamic system of order
Available online xxxx
three that represents the dynamic behavior of the evaporator. Using material, salt and energy balance physical
Keywords:
relations, three dynamic equations are obtained for each effect where they are modified to get the state equation
Desalination of the effect. This procedure is repeated for the condenser while in the thermo-compressor only static equations
MED are considered due to its fast dynamic. The proposed model is validated by actual data of an operating plant in
Thermal vapor compressor Kish Island (in south of Iran). The transient and steady state behavior of the model is more investigated by sim-
Dynamic modeling ulating some conditions like applying disturbance and changing the operating point which can be happened in
Simulation real conditions. Due to high performance of the proposed model, it can be used in optimal designing of the
Model validation MED plants and control applications as well.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction operating costs, simple operating and maintenance procedures, high

Energy conservation has been intensively studied since years ago.
Climate change, population growth and industrial development have
caused water shortage as a comprehensive crisis in many countries
especially in the Middle East and North Africa (MENA). Therefore,
the Persian Gulf countries such as Iran are confronted with the
water shortage crisis insofar as the International Water Management
Institute predicted that many of these countries would face with
high water shortage by 2025 [1]. This issue has encouraged these
countries to use desalination technology in the recent years.
There are several methods for water desalination. Multi-effect de-
salination (MED) is one of the most common techniques that provide
a considerable quantity of potable water. This type of thermal desali-
nation plants due to its advantages like low capital requirements, low
⁎ Corresponding author at: Iran, Tehran,Daneshju Blvd, 1983963113, Shahid beheshti
university. Tel.: +989112777163; fax:+981132364226.
E-mail addresses: mohsen_mazini@yahoo.com (M.T. Mazini), a_yazdizadeh@sbu.ac.ir
(A. Yazdizadeh), mh_ramezani@sbu.ac.ir (M.H. Ramezani).
thermal efficiency, high heat transfer coefficient, lower energy con-
suming and higher performance ratio than other thermal desalination
methods like MSF has been used more in the recent years [2,4]. There-
fore, MED has been considered by many researchers and several
works have been presented regarding performance evaluation and
optimization of these systems during the past years.
Slesarenko [3] made a comparison between vertical and horizontal
types of MED. He showed that the heat transfer area for the horizontal
is almost twice as the vertical type of MED. In 1997 it was presented
that by increasing top brine temperature (TBT) the heat transfer area
of evaporator reduces (due to increasing the temperature and heat
transfer coefficient of evaporator) [4]. This reference also showed that
the performance of the system can be improved by increasing the
value of TBT, although this increment is limited by some physical con-
straint such as corrosion, scaling and maintenance cost.
In order to improve the performance of the system, it is essential
to obtain a dynamic model to simulate and predict the behavior of
the plant. Whereas some researchers have studied performance

http://dx.doi.org/10.1016/j.desal.2014.09.014
0011-9164/© 2014 Elsevier B.V. All rights reserved.
 M.T. Mazini et al. / Desalination 353 (2014) 98–108
improvement of desalination plants, but there are a few references
that investigate the desalination plant from dynamic modeling
point of view, especially for MED. The steady state modeling of the
MED desalination plant, of course, has been the subject of various
studies in the past.
Al-Juwayhel, El-Dessouky and Ettouney [5] performed a comparison
for four types of single effect evaporator desalination systems. They de-
veloped mathematical models for the proposed systems in steady state
conditions. The analysis was based on comparison of the performance
ration, specific power consumption, specific heat transfer area and spe-
cific cooling water flow rate. In 2009 a MATLAB algorithm was devel-
oped and used to solve a mathematical model optimization problem,
where different numbers of effects were tested to maximize the gain
of output ratio [6]. Ettouney in [7] described a computer package for
simulation of different types of multi-effect desalination systems. This
package intended to serve as an educational tool to study the MED cou-
pling to nuclear reactors or fossil energy sources such as gas turbine
combined cycle. Hanbury [8] presented a steady-state solution to
solve equations of a MED plant. The simulation was based on steady
state behavior in boiling heat transfer coefficient, unequal inter-effect
temperature differences. In [9] a comparison was made between four
different types of single effect desalination systems. For each one, the
steady state mathematical equations were written. Finally, performance
ratio and heat transfer area and cooling seawater flow rate were com-
pared. Results showed that energy consumption of MED with mechan-
ical vapor compressor (MVC) decreased by reducing the boiling
temperature of water. Hatzikioseyian and Vidali [10] developed a math-
ematical model that predicted the performance of MED plant, assuming
horizontal tube film evaporator. The model was based on mass and en-
ergy balance in the steady state conditions. Aly and El-Figi [11] present-
ed a mathematical model to analyze the steady state behavior of the
multi-stage and multi-effect desalination systems. These models show
the role of fouling and its effect on the plant performance ratio.
Regarding the dynamic model of desalination systems there are
some useful references for MSF plants. For example Bodalal et al. [12]
developed a dynamical model and validated it by using data from an ac-
tual plant obtained from an operating MSF unit. But not much attention
has been paid in the subject of dynamic modeling of MED plants. In this
regard, Narmin and Marwan in [13] developed a dynamic model for the
MED process based on mass and energy and salt balance equations,
however without any validation by actual data.
In the present work the same procedure of [13] is used and the
model is extended with more details and less assumptions. Com-
pared to the developed model in [13], the proposed model in this
paper considers distillate level of condenser, time delay between ef-
fects and the model of thermo-compressor. In addition, the pro-
posed model has been validated by actual data of an operating plant.
The aim of this paper is to obtain a dynamic model for multi-effect
desalination (MED) plants in such a way that the behavior of the sys-
tem in different working conditions is predicted. MED process con-
sists of three parts that are evaporators (effects), condenser and
thermo-compressor where the governing physical relations are ana-
lyzed for each one and as a result mass, energy and salt balance dy-
namic equations are written for each effect. These equations are
integrated and three nonlinear differential equations are obtained.
The equations are given based on defining three state variables,
namely, brine level, brine salinity and temperature of the effect
that represents the dynamic behavior of each effect. The same proce-
dure applies to the condenser but due to the absence of salinity in
this part, we have two state variables, namely, distillate level and
feed water temperature. The thermo-compressor has fast dynamic
compared to the other parts, so it is considered in steady state situa-
tion. Finally all equations are combined together as physical system
in which all parts are connected together in order to obtain the com-
plete dynamic model of the MED. To have a more accurate model, the
delays of the system are incorporated to the models. This model
99
expresses relations between inputs and outputs as differential equa-
tions and can evaluate dynamic behavior in different working condi-
tions. The model is simulated with the MATLAB/SIMULINK software
to simulate the transient behavior of the multi-effect desalination
plants under various conditions. Finally the model is validated with
actual data from an industrial plant which operate in the south of
Iran and the results and accuracy of the developed model are
shown practically [24]. It should be considered that this reference
is a confidential document obtained by the authors.
This paper is organized as follows: in the next section a description
about the performance of MED is given and some main variables that af-
fect the MED operation are explained. In Section 3, mathematical dynam-
ic model of the plant is fully derived. In Sections 4 and 5, the proposed
model is validated by the actual data from an operating plant and the re-
sults in different situations of the plant under disturbances are presented.
Finally, Section 6 provides some concluding remarks.
2. Process description
MED process operates in a series of evaporator–condenser vessels
called effects and uses the principle of reducing the ambient pressure
in the various effects. There are some configurations of multi-effect de-
salination process. A process diagram for the horizontal parallel multi-
ple effect desalination process is shown in Fig. 1. This process consists
of six effects, main condenser and steam jet ejector. The steam jet ejec-
tor runs by the motive steam; due to design of ejector it entrains a por-
tion of the output vapor from the other or last effect. Ejector compresses
the entrained vapor by the motive steam to reach the desired tempera-
ture and pressure and also increases its flow rate [14]. The heating
steam leaves the ejector and condensed in the first effect and provides
energy required to evaporate seawater. The steam condensate is
recycled to the power plant to its HRSG boiler feed water.
The seawater enters to the condenser and is preheated to desired
temperature and then is forwarded to two directions; part of the heated
seawater is used as the feed of evaporators and the remaining as cooling
seawater is rejected back to the sea. The feed seawater is equally
sprayed on the last four effects, due to the design of the first and second
effects; the feed water flow rate which is sprayed on these effects is
more than the other effects. In the first effect the feed seawater is
sprayed onto the surface of all tubes of evaporator to raise rapid evapo-
ration. The tubes are heated by externally supplied steam from a
HRSG boiler in the power plant. Each effect has lower pressure and
temperature than the previous effect. During this procedure, pres-
sure drop and high temperature in the effects cause a part of the
sprayed seawater on the tubes in the first effect to evaporate. There-
fore, some water is evaporated and the remaining falls into the brine
pool as brine. In the second effect the same procedure is carried out;
the tubes of this effect are heated by the vapors created in the first ef-
fect. These vapors that run into the tubes are condensed to produce
fresh water. This process of evaporation and condensation continues
all the way to the last effect. On the other hand the collected brine
stored in the brine pool from the first evaporator is flowed into the
second effect through an orifice. This process is repeated for all ef-
fects up to the last one [15,16].
Usually some plants, due to their capacity, have been built to operate
with 4 to 13 effects. The production of the MED plants is related to the
number of effects. The total number of effects is limited by the total
available temperature range and the minimum allowable temperature
difference between two adjacent effects. Some plants have been built
to operate with a top brine temperature (TBT) in the first effect of
about 60–70 °C that reduces the potential for scaling of seawater [17].
Some important variables in the MED process that have role in de-
sign of control structure for the plant are listed as below:
• Top brine temperature (TBT) from the first effect; this variable affects
the distillate production and the performance of plant directly.
100 M.T. Mazini et al. / Desalination 353 (2014) 98–108
Heating steam
TVC
Feed Water
Tv Tv2 Tv3
Condensate to HRSG
Brine
Seawater
Fig. 1. Multi-effect desalination (MED) process.
Increase in TBT, while the last effect brine temperature is fixed, re-
duces the heat transfer area of each effect. On the other hand, the in-
crease of GOR with increasing no. of effects is not a consequence of the
reduction in T difference, but of the nature itself of the multiple effect
distillation process. Moreover, increasing TBT not necessarily requires
an increase in motive steam flow rate, but an increase in its pressure
can be enough [18].
• Brine salinity in the first effect; this can affect the performance of MED
and its value is between 6.5 and 7.5%. If brine salinity is less than nor-
mal value, one may conclude that there is no sufficient steam to heat
the feed water sprayed onto tubes or the flow rate of sprayed sweater
onto tubes is higher than normal.
• Seawater flow rate; this affects the seawater feed temperature
sprayed on the effects. Increasing seawater flow rate at the constant
heat transfer rate of condenser reduces feed water temperature.
• Feed water temperature; this directly affects on brine level and heat
transfer rate of each effect.
• Brine level; this ensures that the heat exchanger tubes are not sub-
merged in brine pool. The brine level must be measured and con-
trolled in order to prevent rising higher than normal value.
• Distillate level in the condenser; reducing this variable causes the con-
denser tubes not to submerge in distillate pool. Because of relatively
high product temperature, storage of distillate in the condenser
Fig. 2. Block diagram of a single effect with interactions between bundles.
Motive steam from HRSG
Cooling Seawater
Tv4 Tv5 Tv6
Seawater
Rejected Distillate
Steam
Condensate
Distillate
increases the heat transfer rate and consequently improves the per-
formance of the plant [16].
3. Dynamic modeling
The multi-effect desalination process is an evaporation and conden-
sation process at low pressure (vacuum), where the pressure and tem-
perature decrease in each effect [19]. Generally this system consists of
three subsystems that are called effects, condenser and steam jet ejector
(thermo compressor). Each effect is an evaporator which distillation
and evaporation processes are performed. From the modeling point of
view, it is easier to describe a single evaporator into three lumps involv-
ing brine lump, vapor space lump and tube bundle. Fig. 2 shows the
graphical representation of the three lumps of a single effect. As
shown there are six fluid streams in each effect; input vapor, feed
water, brine from previous effect as input fluids and formed vapor, dis-
tillate, output brine as output fluids. Each fluid stream has four descrip-
tive variables, flow rate, temperature, pressure, and salt concentration.
Due to the saturated conditions in the process, the pressure and temper-
ature variables are dependent, as a result three variables are considered
in the process. For each lump of effect, mass and energy and salt balance
equations are written. Time derivatives of these equations are applied
and finally the equations are combined to represent a dynamic model.
 M.T. Mazini et al. / Desalination 353 (2014) 98–108 101

In the remainder of this section, the mathematical model for each It is supposed that the temperature of vapor into the tube bundles is
subsystems of MED is derived separately. the arithmetic mean between the inlet and outlet vapor temperatures of
the effect:
3.1. Mathematical model of the effect
T v;i−1 þ T d
TT ¼ : ð8Þ
As mentioned, the effect can be divided into three lumps. The mate- 2
rial, energy and salt balance dynamic equations for each lump of the ef-
fect are written [17]. So the temperature of the distillate exiting from each effect is calcu-
lated as follows:
3.1.1. Brine lump d
 2

 
Material balance for ith effect: hv;i−1 þ hd ¼ W v;i−1 hv;i−1 −hd;i −Q E;i : ð9Þ
dt ρT  V T
d
M ¼ W feed þ W b;i−1 −W b;i −W v;i : ð1Þ Although Eqs. (1) to (9) describe the dynamic behavior of the effect,
dt b;i
however the state variables are not known from these equations. There-
Energy balance for ith effect: fore, in the following the equations are rewritten as the state equations
using partial derivative and variable changes. For this purpose, three
d
 variables called as: brine level (L), brine salinity (X) and temperature
M b;i :hb;i ¼ W feed  hfeed þ W b;i−1  hb;i−1 −W b;i  hb;i −W v;i of the effect (T) are selected as the state variables and their derivatives
dt
 hv;i þ Q E;i ð2  aÞ are obtained.
With differentiation from Eq. (6), the following relation is obtained:
in which
dT b;i dT v;i dBPEi


 ¼ þ : ð10Þ
T v;i−1 þ T feed − 2T v;i dt dt dt
Q E;i ¼ U E;i  AE;i  LMTD ; LMTDi ¼ " # ð2  bÞ
T v;i−1 −T v;i Regarding assumption Tv,i = Ti that means the temperature of the
ln
T v;i −T feed output vapor is equal to temperature of the formed vapor in the corre-
sponding effect, Eq. (10) can be written in detail as:
and salt balance for ith effect:  
dT b;i ∂BPEi dT i dBPEi dX i
¼ 1þ þ : ð11Þ
d
 dt ∂T i dt dX i dt
M b;i  X b;i ¼ W feed  X feed þ W b;i−1  X b;i−1 −W b;i  X b;i : ð3Þ
dt
Mass of brine can be obtained using density relation as:

3.1.2. Vapor lump M b;i

ρ f ;i ¼ → M b;i ¼ As  Li  ρ f ;i : ð12Þ
Material balance for ith effect: As  Li

d Now, by rewriting Eqs. (1) to (3), the differentiation of Mb,i (mass of

M ¼ W e;i −W v;i ð4  aÞ
dt v;i brine) should be computed as a function of state variable.
According to Eqs. (1) and (10) the following relation is achieved:
where !
dMb;i dL ∂ρ f ;i ∂T b;i ∂ρ f ;i ∂X i
W e;i ¼ W vt;i þ W vb;i : ð4  bÞ ¼ As  ρ f ;i  i þ As  Li  þ : ð13Þ
dt dt ∂T b;i dt ∂X i dt

Energy balance for ith effect:

Substituting from Eq. (11) into Eq. (13), we have:
d

∂ρ f ;i
 
M v;i  hv;i ¼ W e;i  hv;i −W v;i  hv;i : ð5Þ dMb;i dL ∂BPEi dT i
dt ¼ As  ρ f ;i  i þ As  Li  1þ þ As  L i
dt dt ∂T ∂T i dt
! b;i
It should be noted that there is a difference between the tempera- ∂ρ f ;i ∂BPEi ∂ρ f ;i dX i
  þ : ð14Þ
ture of the vapor formed into an effect and temperature of brine into ∂T b;i ∂X i ∂X i dt
the brine pool that called thermodynamic losses [15]:
Similarly from Eq. (12), we have the following relations:
T b;i ¼ T v;i þ BPEi ð6Þ
M v;i
ρv;i ¼ → M v;i ¼ As  Lv;i  ρv;i : ð15Þ
where the detail of BPE is given in Appendix A. As  Lv;i

3.1.3. Tube bundle So the following equation can be obtained from Eq. (4-a) and
The tube wall in the effect has very thin layer, therefore, the thermal ca- Eq. (4-b):
pacity of tube is much less than the thermal capacity of the hot vapor [10]
and as a result the heat loss of tube metal can be neglected in calculation. ∂ρv;i dT i dL
As  Lv;i  −As  ρv;i i ¼ W e;i −W v;i : ð16Þ
The steam entering into the tube bundle is condensed and distillate ∂T i dt dt
is produced; due to no changing in the mass of steam, no material bal-
ance is needed. For calculation of output distillate temperature from Substituting Eqs. (14) and (16) into Eq. (1), the following nonlinear
each effect, the energy balance is used as below: equation for the mass balance equations is realized:

d
 dLi dT dX
ðMT  hT Þ ¼ W v;i−1 hv;i−1 −hd;i −Q E;i : ð7Þ k1 þ k2 i þ k3 i ¼ k4 ð17  aÞ
dt dt dt dt
102 M.T. Mazini et al. / Desalination 353 (2014) 98–108

where: rate) are written in terms of state and input variable as:
8

>
> k1 ¼ As  ρ f ;i −As  ρv;i W v;i−1 Li−1 −W feed  C p f T i −T feed þ W b;i−1  C pb ðT i−1 −T i Þ
>
>   W v;i ¼ :
>
> ∂ρ f ;i ∂BPEi ∂ρv;i
>
> k ¼ A  L  1 þ þ As  Lv;i  Li
< 2 s i
∂T b;i ∂T i ∂T i
! : ð17  bÞ ð21Þ
>
>k ¼ A  L  ∂ρ f ;i ∂BPE ∂ρ f ;i
>
>
i
þ
>
> 3 s i
∂T b;i ∂X i ∂X i
>
> Brine flow rate outgoing from each effect can be calculated as:
:
k4 ¼ W feed þ W b;i−1 −W b;i −W v;i
W b;i ¼ ρ f ;i  V b;i  Ab;i ð22Þ
Similarly, combining Eqs. (2), (5), (11) and (13), the following equa-
tion is obtained for the energy balance equations: in which brine velocity (Vb,i) can be obtained using the following
Bernoulli equation [20]:
dLi dT dX
k5 þ k6 i þ k7 i ¼ k8 ð18  aÞ P iþ1 þ ρ f ;iþ1  g  Liþ1 V b;i 2
dt dt dt Pi
þ Li ¼ þ ð23Þ
ρ f ;i  g ρ f ;i  g 2g
where
8 where Ab,i is the cross section of brine pipe.
>
> k ¼ As  hb;i  ρ f ;i −As  hv;i  ρv;i
> 5
>   !
>
> ∂hb;i ∂BPEi ∂hv;i 3.2. Mathematical model of condenser
>
> k6 ¼ As  Li  ρ f ;i  1þ þ ρv;i  As  Lv;i 
>
>
>
> ∂T b;i ∂T i ∂T i
>
>   !
>
> ∂ρ ∂BPE ∂hv;i The output vapor from the last effect is entered into the condenser to
>
> þ As  Li  hb;i 
f ;i
1þ i
þ hv;i  As  Lv;i 
>
> heat the seawater. To model the condenser, it is divided into two lumps;
>
> ∂T b;i ∂T i ∂T i
>
< condenser tube lump and condenser inner lump which are described in
!
∂hb;i ∂BPEi ∂hb;i ð18  bÞ the following, separately.
>
> k7 ¼ As  Li  ρ f ;i  þ þ As  Li  hb;i
>
> ∂T ∂X ∂X
>
> b;i i i
>
> ! 3.2.1. Condenser tube lump
>
>
>
> ∂ρ f ;i ∂BPE ∂ρ f ;i Due to the incompressible nature of the feed flow inside the tubes,
>
>  i
þ
>
> ∂T b;i ∂X i ∂X i
>
> no material balance is required for seawater in the tubes. The energy
>
>
>
> balance for the condenser tubes can be written as:
>
:k ¼ W h
8 feed feed þ W b;i−1 hb;i−1 −W b;i hb;i −W v;i hv;i þ Q E;i :
d
ðMcon  hcon Þ ¼ Q con −W sw  hsw ð24Þ
dt
Finally Eqs. (3) and (14) are used to derive following relation for the
salt balance equation:
dhcon

dLi dT dX V con  ρcon  ¼ Q con −W sw  cp T feed −T f ð25Þ
k9 þ k10 i þ k11 i ¼ k12 ð19  aÞ dt
dt dt dt
where
where
8 T −T f
k ¼ As  ρ f ;i  X i Q con ¼ U con  Acon  LMTD ; LMTD ¼ " feed #: ð26Þ
>
>   T v;n −T f
> 9
>
>
> ρ f ;i ∂BPEi ln
>
> k ¼ A  X  L  1 þ T v;n −T feed
< 10 s i i
∂T b;i ∂T i
!: ð19  bÞ
>
> ρ f ;i ∂BPEi ∂ρ f ;i
>
> k ¼ A  L  X  þ X  þ ρ
>
> 11 s i i
∂T b;i ∂X i i
∂X i f ;i
>
> 3.2.2. Condenser inner lump
:
k12 ¼ W feed X feed þ W b;i−1 X b;i−1 −W b;i X b;i
Because all of the vapors that entered in the condenser distill in the
condenser pool, and since the important variable in this section is distil-
Now we can write derivatives of three mentioned state variables late level, we neglected phase transition and only use these equations
using Eqs. (17-a), (18-a) and (19-a) as follows: for obtaining the distillate level.
Material balance and energy balance for input vapor to the condens-
8
> dLi CE−BF er are as Eqs. (27) and (28), respectively:
>
> ¼
>
> dt AE−BD
>
>
< dT i AF−CD d
¼ W v þ Dn ¼ M þ W out ð27Þ
dt AE−BD ð20  aÞ dt t
>
> dL dT
>
> k12 −k9 i −k10 i
> dX
>
>
: i
¼ dt dt
dt k11
d
W v  hv þ Dn  hn ¼ Q con þ W out  hout þ ðM t  ht Þ: ð28Þ
dt
where
 In Eq. (24), temperature of seawater into the condenser tubes is con-
A ¼ k1 k11 −k3 k9 B ¼ k2 k11 −k3 k10 C ¼ k4 k11 −k3 k12 sidered as the arithmetic mean temperature between inlet and outlet
: ð20  bÞ
D ¼ k5 k11 −k7 k9 E ¼ k6 k11 −k7 k10 F ¼ k8 k11 −k7 k12 temperature, i.e.

On the other hand, using energy and Bernoulli equations [6], output T feed þ T f
T con ¼ : ð29Þ
variables Wv (output vapor flow rate) [9] and Wb (output brine flow 2
 M.T. Mazini et al. / Desalination 353 (2014) 98–108 103

Table 1
Actual parameters of Kish Island MED-TVC plant.

No. of evaporator tubes in the first and second effects 7000 Evaporator tube diameter (mm) 28.575 No. of evaporator tubes in the other effects 3500
Evaporator tube length (m) in the first and second effects 6.1 Condenser tube length (m) 6.5 Condenser tube diameter (mm) 19.05
First and second effects width (m) 3.6 Other effects width (m) 1.8 Length of other effects tubes (m) 3.9
No. of condenser tubes 1890 Condenser length (m) 8 First and second effects length (m) 7
Effects height (m) 4 Condenser height (m) 3.2 Length of other effects (m) 4.8
Condenser width (m) 2.2 CR of ejector 1.2 No. of effects 6
Brine pipe diameter (inch) 14–18 Feed water pipe diameter (inch) 10–12 Distillate pipe diameter (inch) 8

Eqs. (24) to (28) can be used to estimate the transient behavior of “constant-area mixing ejector” and “constant-pressure mixing ejector”.
the feed water temperature and distillate level into the condenser, In this paper, the ejectors have been modeled based on the real plant
respectively. which is “constant-area mixing ejector” [22]. Due to fast response of
Combining Eqs. (25) and (26), the following equation is given for ejectors compared to other parts of MED systems, its dynamic equations
feed water temperature: are not considered in the modeling. The simplified static equations be-
tween inputs and output of ejector are as follows [9]:

    
dT feed Q con −W sw  C p T feed −T f ðP s Þ1:19 P m 0:015 PC F
ω ¼ 0:296 ð32Þ
¼2 ð30Þ ðP ev Þ1:04 P ev TC F
dt ∂h
ρcon  V con  con
∂T con where

−7 2
and the distillate level is obtained from Eq. (27): PC F ¼ 3  10 ðP m Þ −0:0009ðP m Þ þ 1:6101 ð33Þ

dLt W v;n þ D−W out
¼ :
dt Ac  ρt
3.3. Mathematical model of steam jet ejector
Thermal desalination systems operate at pressures lower than the
atmosphere pressure. Therefore, using vacuum devices in these systems
is unavoidable [16]. Ejectors are common thermal devices that provide
vacuum requirement in MED systems in which all processes are per-
formed in the sonic or supersonic situation. Due to the simplicity of
the design and absence of motive parts, ejectors are very reliable and
they require, practically, no maintenance and have a relatively low in-
stallation cost [21]. The ejectors powered by heat are low-cost energy
equipments and it is obviously less expensive to run them than electri-
cal or mechanical-related power. The steam required for the jet ejector
is commonly drawn from boilers. These devices are used in vapor com-
pression desalination systems as a heat pump. The thermo compressor
(as a kind of ejector) is used to compress the vapor from pressure Pv
(which is the vapor pressure leaving one of the effects or condenser
that depends on the system design) to P1 (which is the vapor pressure
entering the first effect) by using an external source of steam (motive
steam) at a pressure greater than the vapor pressure.
Regarding high pressure and temperature of primary input (motive
steam) and low pressure and temperature of secondary input (entrain-
ment vapor) in the ejector, applying jet action on motive steam de-
creases its pressure. This action makes some of the output vapor from
second effect to be sucked. Finally the mix of motive steam and entrain-
ment vapor exits from the ejector and runs to the first effect. The ejector
design can be classified into two categories which are known as
and
ð31Þ
−8 2
TC F ¼ 2  10 ðT ev Þ −0:0006ðT ev Þ þ 1:0047: ð34Þ
The effect of temperature and salinity on the physical properties of
water such as specific heat at constant pressure, density, enthalpy, pres-
sure and latent heat is taken into consideration using mathematical cor-
relations [9,23]. These correlations are given in Appendix A.
4. Model validation
The derived equations are simulated in MATLAB software.
Dormand–Prince method is used for the numerical solution of the dif-
ferential equations. In order to evaluate the proposed model, input
data of a real plant is extracted and applied to the model [24]. The
plant is a 4000 m3/day desalination unit which is operated by Water
and Electricity Company of Kish Island in Iran. Table 1 gives the values
of known parameters of the plant which have been used in the
modeling.
By comparison of real system outputs with the model outputs, the
validity of the model is investigated. For this purpose, values of un-
certain parameters of the model is varied in a range around the nom-
inal values and for each set of parameters, differences between
model output and system output is measured. The assigned variables
that is important for simulation is: TBT, feed water temperature, the
brine level, the brine salinity, condenser level. The initial condition
for modeling is listed in Table 2 and operation condition is listed in
Table 4. Actual data are extracted during 2 PM to 3 PM on 12-Apr-
2012. Figs. 3 and 4 typically show variation of one of the main inputs
and outputs of the plant, namely, motive steam and first effect tem-
perature respectively.

Table 2
Initial conditions of the modeling.

Initial conditions Value Initial conditions Value Initial conditions Value

Feed water temperature 43 Condenser level 0.25 Distillate temperature 47

Temperature of first effect 52 Brine level of first effect 0.2 Brine salinity of first effect 4
Temperature of second effect 48 Brine level of second effect 0.23 Brine salinity of second effect 4.3
Temperature of third effect 45 Brine level of third effect 0.25 Brine salinity of third effect 4.5
Temperature of 4th effect 42 Brine level of 4th effect 0.27 Brine salinity of 4th effect 4.7
Temperature of 5th effect 39 Brine level of 5th effect 0.29 Brine salinity of 5th effect 4.9
Temperature of 6th effect 36 Brine level of 6th effect 0.31 Brine salinity of 6th effect 5.1
104 M.T. Mazini et al. / Desalination 353 (2014) 98–108

Table 3 Table 4
Average absolute and relative error between actual and simulated values of four main out- Operation conditions of actual plant.
puts of the plant.
Operation condition Value Operation conditions Value
Plant output First effect Feed water Condenser Total
Seawater flow rate (t/h) 775–785 Seawater salinity (%) 6.5–7.5
temperature temperature temperature distillate
Plant input Seawater temperature (°C) 24–26 Steam flow rate (t/h) 19–22
product
Steam pressure (bar) 8.7–9.4
Average absolute error 0.0351 (°C) 0.14 (°C) 0.45 (°C) 0.417 (kg/s)
Normalized Average 1.17% 3.5% 9% 20.8%
relative error × 100%
Figs. 6 to 8 show the variation of brine level, salinity and temper-
ature of all effects with respect to the variation of operating point,
Table 3 shows the average absolute and relative error between actu- respectively. At time 6600 s, the feed water flow rate is increased
al and simulated values of three main outputs of the plant. As seen, the to 5% and then at time 7200 s, a 5% increment is applied to the feed
results show good agreement between the model and actual outputs. water temperature.
The proposed model is considerably valid to accurately predict the per- Although by increasing the feed water flow rate and constant heat
formance characteristics of multi-effect desalination plant at both transfer rate of each effect (due to constant input steam specifica-
steady state and transient. As a sample of obtained results, Fig. 5 tion), it is expected that the temperature of the effect decreases
shows the comparison between simulated and actual time response (and brine level rises), however, brine level is dropped here at the
for seawater temperature. first moments (Fig. 6). The reason is that the sudden decrease in
brine levels (starting from the last effects) is likely related to the en-
hanced cooling in the final condenser and last stage, which generates
5. Simulations and results a sudden reduction in stage pressure and a faster exiting flows from
the first effects to the last ones (showing graphs of other variables,
In this section, in order to have more investigation of the obtained such as pressure, exiting flow rates, etc. would probably indicate it).
model, transient and steady state response of model under various As seen in Fig. 7, when the feed water flow rate increases (while
working conditions such as changing the operating point and applying other inputs are constant), the amount of salinity reduces due to
the disturbance is evaluated through simulation. Using these simula- the increase of the level. It is worth noting that due to the presence
tions, which are usually regarded as exaggerated conditions, could of some delays, this reduction in each effect is done slower than
have more evaluated the response of the model and analyzed the results the previous one. At time 7200 s, the feed water temperature is in-
from logical point of view. creased so the brine salinity is also increased because of higher
In the first simulation, the nominal values are selected for input evaporation.
variables and the system is allowed to reach its steady state value. Note that the rates of decrement of the temperature in the last ef-
Then by sudden changes in feed water flow rate and temperature fects are higher than the previous ones in Fig. 8. The reason is that the
in two different times, the variations of the effect variables have temperature reduction of each effect is affected by two factors: the in-
been investigated. crement of feed water flow rate and the temperature reduction of the

Nomenclature

Heat transfer area of effect (m2) AE Condenser area (m2) Ac Heat transfer area of condenser (m2) Acon
Evaporator area (m2) As Boiling point elevation (° C) BPE Specific heat of inlet seawater at constant pressure cp


kJ
kg ÅC
Specific heat of inlet brine at constant pressure cpb Specific heat of feed water at constant cpf Total distillate flow rate entering the D



kJ
pressure kJÅ condenser (kg/s)
kg ÅC kg  C
Enthalpy of brine in the ith effect (kJ/kg) hb,i Enthalpy of seawater in the condenser tubes hcon Enthalpy of distillate exiting the ith effect (kJ/kg) hd,i
(kJ/kg)
Enthalpy of feed water to evaporators (kJ/kg) hfeed Enthalpy of inlet seawater (kJ/kg) hf Enthalpy of distillate exiting the last effect (kJ/kg) hn
, hsw
Enthalpy of total distillate exiting the condenser hout Brine level in the ith effect (m) Li Vapor level in the ith effect (m) Lv,i
(kJ/kg)
Enthalpy of vapor in the tube bundles (kJ/kg) hT Logarithmic mean temperature difference LMTD Enthalpy of distillate pool in the condenser (kJ/kg) ht
Enthalpy of vapor in the vapor space (kJ/kg) hv Mass of brine in the ith effect (kg) Mb,i Seawater mass in the condenser tubes (kg) Mcon
Vapor mass in the tube bundles (kg) MT Mass of distillate pool in the condenser (kg) Mt Mass of vapor in the ith effect (kg) Mv,i
Entrained vapor pressure (kPa) Pev Motive steam pressure (kPa) Pm Pressure of compressed vapor (kPa) Ps
Saturated pressure (kPa) Psat Brine temperature in the ith effect (° C) Tb,i Seawater temperature in the condenser tubes (° C) Tcon
Distillate temperature in the ith effect (kj/kg) Td,i Feed water temperature (° C) Tfeed Inlet seawater temperature (° C) Tf
Entrained vapor temperature (° C) Tev Secondary steam temperature (° C) Ts Temperature of vapor in the vapor space in the ith Ti
effect (° C)
Primary steam temperature (° C) Tp Vapor temperature exiting the ith effect (° C) Tv,i Heat transfer coefficient of evaporator and UE,


condenser kJÅ Ucon
kg  C
Heat transfer rate of evaporator (kW) QE Heat transfer rate of condenser (kW) Qcon Water volume in the condenser tubes (m3) Vcon
Vapor volume in the vapor space (m3) Vv Brine mass flow rate exiting the ith Wb,i Seawater flow rate (kg/s) Wf
effect (kg/s) Wsw
Feed water flow rate to the effects (kg/s) Wfeed Rate of total distillate exiting the condenser Wout Vapor mass flow rate exiting the ith effect (kg/s) Wv,i
(kg/s)
Rate of vaporization from the tube bundle (kg/s) Wvt,i Rate of vaporization from the brine Wvb,i Total rate of vaporization produced in the We,i
pool (kg/s) effect (kg/s)
Salinity of brine in the evaporator (%) Xb Salinity of feed water to the evaporator (%) Xfeed Density of vapor (kg/m3) ρv
Density of distillate pool in the ρt Water density in the tubes of ρcon Density of water (kg/m3) ρf
condenser (kg/m3) condenser (kg/m3)
Density of vapor in the tubes of effect (kg/m3) ρT
 M.T. Mazini et al. / Desalination 353 (2014) 98–108 105

5.95

5.9
Flowrate (kg/sec)
5.85

5.8

5.75

5.7

5.65

5.6

0 500 1000 1500 2000

Time (sec)

Fig. 3. Actual value of motive steam flow rate entering into the thermo-compressor.

68.45

68.4

68.35

68.3
temperature (*C)

68.25

68.2

68.15

68.1

68.05

68
0 500 1000 1500 2000
Time (sec)

Fig. 4. First effect temperature from actual data.

68.5

68.4

68.3

68.2
Temperature (*C)

68.1

68

67.9

Model output
67.8
Actual data

67.7
6600 6800 7000 7200 7400 7600 7800 8000 8200 8400 8600 8800
Time (sec)

Fig. 5. Actual and simulated value of first effect temperature.

106 M.T. Mazini et al. / Desalination 353 (2014) 98–108

First effect
35 Second effect
Third effect
4th effect
5th effect
6th effect
30

25
Level (cm)

20

15

10
6600 6800 7000 7200 7400 7600 7800
Time (sec)

Fig. 6. Brine level variations of each effect.

previous effect which are affected by a delay. So, rate of temperature re-
duction of each effect is higher than the previous.
The variation of temperature, salinity and level of each effect
affected by increasing feed water temperature at time 7200 s is shown
in Figs. 6–8. Since the increment of feed water temperature has reverse
influence versus the increment of feed water flow rate, so the behavior
of outputs is the same as before but in reverse direction.
In the second simulation, the effect of seawater variation on the
system behavior is investigated. Fig. 9 shows the variation of each
effect temperature by increasing in the seawater temperature. It
shows that the seawater temperature is one of the existing distur-
bances in the control of desalination systems. Since the increment
of the seawater temperature at the constant heat transfer of con-
denser increases the feed water temperature, the temperature
increment of the effect diagram shows slower behavior than the
previous case (due to existing delay in the condenser). Two-step in-
crease in temperature of the effects that is evident in some shapes is
due to different delays in two paths.
6. Conclusion
In this paper a dynamic mathematical model for the analysis of MED
desalination plants is proposed. The developed model is based on the
basic equations of mass, salt and energy. The model is simulated by

7
First effect
Second effect
6.8 Third effect
4th effect
5th effect
6.6 6th effect

6.4
Salinity (%)

6.2

5.8

5.6

5.4
6600 6800 7000 7200 7400 7600 7800
Time (sec)

Fig. 7. Brine salinity of each effect at seawater temperature.

 M.T. Mazini et al. / Desalination 353 (2014) 98–108 107

66

64

62

60

58
Temperature (*C)

56

54

52

50
First effect
48
Second effect
Third effect
4th effect
46
5th effect
6th effect
6600 6800 7000 7200 7400 7600 7800
Time (sec)

Fig. 8. Temperature of each effect from proposed model by using constant inputs.

67

66

65

64

63
Temperature (*C)

62

61

60

First effect
59 Second effect
Third effect
58 4th effect
5th effect
6th effect
57
6900 6950 7000 7050 7100 7150 7200 7250 7300 7350 7400
Time (sec)

Fig. 9. Effect of seawater inlet temperature on temperature of effects in the same seawater flow rate.

using MATLAB/SIMULINK software and the results are compared Appendix A. Thermodynamic properties [23,24]
with the logged data of an actual plant in Kish Island south of
Iran. The plant outputs confirm the validity of the proposed There are some variables in the equations that directly depend on
model. Using the validated model, the system behavior under temperature and salinity. Relations of these variables are as below:
various conditions such as changing operating point and applying Saturated temperature:
disturbance is investigated. As a matter of fact, the proposed  
model can be used for optimal design of system and control 3892:7
T sat ¼ 42:6776− −273:15 ðA1Þ
applications. ½ ln ðP=1000Þ−9:48654

where Tsat in °C and P in kPa.

Acknowledgment Saturated pressure:

The authors would like to express their gratitude to the MAPNA P sat ¼ 10:1724607−0:6167302  T þ 1:832849
−2 2 −4 3 −6 4
Company (Tehran—Iran) and Kish Water and Power Company  10 T −1:77376  10 T þ 1:47068  10 T ðA2Þ
(Kish Island—Iran) for their cooperation and also our co-worker
Mr. Yazdani for providing his experiences. where T in °C and Psat in kPa.
108 M.T. Mazini et al. / Desalination 353 (2014) 98–108

Water enthalpy: T in °C, and X is the water salinity in g/kg.

BPE correlation:
−4 2
hb ¼ 0:063635409 þ 4:207557011  T−6:200339  10 T
−6 3
þ 4:459374  10 T : ðA3Þ

2
BPE ¼ X A þ BX þ CX ðA11Þ
Vapor water enthalpy:
8

−2 −4 −6 2
>
> A ¼ 8:325  10 þ 1:883  10 T þ 4:02  10 T
hv ¼ 2501:689845 þ 1:806916015  T þ 5:087717 >
<

−4 2 −5 3 −4 −5 −7 2
 10 T −1:221  10 T : ðA4Þ B ¼ −7:625  10 þ 9:02  10 T−5:2  10 T
>
>

>
: C ¼ 1:522  10−4 −3  10−6 T−3  10−8 T 2

Latent heat of vaporization:

where T is the temperature in °C and X is the salt weight percentage. The
−2 2
L ¼ 2589:583 þ 0:9156  T−4:8343  10 T : ðA5Þ above equation is valid over the following ranges: 1 b X b 16%,
Seawater density: 10 b T b 180 °C.
3
ρ ¼ 10  ðA1 F 1 þ A2 F 2 þ A3 F 3 þ A4 F 4 Þ: ðA6Þ
References
where
[1] F.R. Rijsberman, Water scarcity: fact or fiction, Proceedings of the 4th International
Crop Science Congress, Brisbane, Australia, October 2004.
8
> A1 ¼ 4:032219G1 þ 0:115313G2 þ 3:26  10 G3 8
−4 [2] K. Al-Shayji, Modeling, simulation, and optimization of large-scale commercial desa-
>
> > 2X
< −3 −4 < G1 ¼ 0:5 −150 lination plants, PhD Thesis Chemical Engineering, Virginia Polytechnic Institute and
A2 ¼ −0:108199G1 þ 1:571  10 G2 −4:23  10 G3
−3 −6 ; G2 ¼ B ; B ¼ 1000 State University, April 1998.
>
> A3 ¼ −0:012247G1 þ 1:74  10 G2 −9  10 G3 >
: G ¼ 2B −1
2 150
>
: 3 [3] V. Slesarenko, Seawater desalination in thin film plants, Desalination 96 (1994)
−4 −5 −5
A4 ¼ 6:92  10 G1 −8:7  10 G2 −5:3  10 G3
173–181.
[4] M. Najem Al-Najem, M.A. Darwish, F.A. Youssef, Thermovapor compression de-
salters: energy and availability — analysis of single and multi-effect systems, Desa-
8
> F 1 ¼ 0:5 lination 110 (1997) 223–238.
>
<F ¼A [5] F. Al-Juwayhel, H. El-Dessouky, H. Ettouney, Analysis of single-effect evaporator de-
2 2T−200
2 ; A¼ : salination systems combined with vapor compression heat pumps, Desalination 114
> F 3 ¼ 2A −1
>
:
160 (1997) 253–275.
3
F 4 ¼ 4A −3A [6] A.O. Bin Amer, Development and optimization of ME-TVC desalination system,
Desalination 249 (October 2009) 1315–1331.
X is the seawater salinity in ppm, and T is the seawater temperature [7] H.M. Ettouney, H. El-Dessouky, A simulator for thermal desalination processes,
Desalination 125 (1999) 277–291.
in °C. This correlation is valid over the following ranges:
[8] W.T. Hanbury, An analytical simulation of multiple effect distillation plant, Proceed-
0 b X b 160,000 ppm and 10 b T b 180 °C. ings of the IDA World Congress on Desalination and Water Sciences, Abu Dhabi, vol.
Vapor density: IV, November 1995, pp. 375–382.
[9] M. Shakouri, H. Ghadamian, R. Sheikholeslami, Optimal model for multi effect desa-
−5 2 lination system integrated with gas turbine, Desalination 260 (May 2010) 254–263.
ρv ¼ 0:005059 þ 0:00023748T v þ 1:777  10 T v −4:327 [10] A. Hatzikioseyian, R. Vidali, P. Kousi, Modelling and Thermodynamic Analysis of a
−8 3 −9 4
 10 T v þ 4:342  10 T v : ðA7Þ Multi Effect Distillation Plant, 2006.
[11] Narmine H. Aly, El-Figi, Thermal performance of seawater desalination systems,
Desalination 158 (2003) 143–150.
Heat transfer coefficient of condenser: [12] Awad S. Bodalal, Sayed A. Abdul_Mounem, Hamid S. Salama, Dynamic modeling and
simulation of MSF desalination plants, Jordan J. Mech. Ind. Eng. 4 (3) (June 2010)
−2 −5 2 394–403 (ISSN 1995-6665).
U con ¼ 1:7194 þ 3:2063  10 T vn −1:5971  10 T vn þ 1:9918
[13] Narmin H. Aly, M.A. Marwan, Dynamic response of multi effect evaporators, Desali-
−7 3
 10 T vn : ðA8Þ nation 114 (1997) 189–196.
[14] F.N. Alasfour, M.A. Darwish, A.O. Bin Amer, Thermal analysis of ME-TVC + MEE de-
salination systems, Desalination 174 (2005) 39–61.
Heat transfer coefficient of evaporator: [15] H.T. El-Dessouky, H.M. Ettouney, Multi effect evaporation vapor compression, Fun-
damentals of Salt Water DesalinationMarch 2002.
−2 −5 2
U e ¼ 1:9695 þ 1:2057  10 T b −8:5989  10 Tb þ 2:565 [16] A. Adibfar, Desalination Plant methods, MAPNA Group, 2010.
−7 3 [17] A. Ophir, A. Gendel, G. Kronenberg, The LT-MED process for SW cogen plants,
 10 T b : ðA9Þ Desalin. Water Reuse 4 (1) (1994) 28–31.
[18] A.D. Khawajia, I.K. Kutubkhanaha, Jong-Mihn Wieb, Advances in seawater desalina-
tion technologies, Desalination 221 (2008) 47–69.
Specific heat of water at constant pressure: [19] M. Al-Shammiri, M. Safar, Multi-effect distillation plants: state of the art, Desalina-

 tion 126 (1999) 45–59.
−3 2 3
C p ¼ 10  A þ BT þ CT þ DT ðA10Þ [20] Raymond Mulley, Flow of Industrial Fluids: Theory and Equations, CRC Press, 2004,
pp. 43–44.
[21] R.K. Kamali, S. Mohebinia, Experience of design and optimization of multi-effects
where desalination systems in Iran, Desalination 222 (2008) 639–645.
[22] Y. Li, S. He, R.Z. Wang, Progress of mathematical modeling on ejectors, Renew.
8 −2 2 Sustain. Energy Rev. 13 (2009) 1760–1780.
>
< A ¼ 4206:8−6:6197X þ 1:2288
> −2
 10 X
−4 2 [23] Yunus A. Cengel, Michael A. Bols, Thermodynamics an Engineering Approach, 5 Edi-
B ¼ −1:1262 þ 5:4178  10 X−2:2719  10 X
−6 2 :
tion McGraw-Hill Science/Engineering/Math, January 2010.
−2 −4
>
> C ¼ 1:2026  10 −5:3566  10 X þ 1:8906  10 X [24] Kish Water a Power Company, in: Kish — Iran (Ed.), Data Sheet of Kish Island Desa-
: −7 −6 −9 2
D ¼ 6:8777  10 þ 1:517  10 X−4:4268  10 X lination, 2012.