Desalination and Water Treatment 217 (2021) 31–41

March
www.deswater.com
doi: 10.5004/dwt.2021.26825

Dynamic modeling and simulation of the multi-effect distillation

desalination process

Jianqing Hua, Lianying Wua,*, Yan Wangb, Weitao Zhanga, Yangdong Hua
a
College of Chemistry and Chemical Engineering, Ocean University of China, Qingdao, Shandong, China, 266100,
Tel. +8613791914873; emails: wulianying@ouc.edu.cn (L.Y. Wu), hujianqing.cool@163.com (J.Q. Hu),
55768616@qq.com (W.T. Zhang), ydhuhd@ouc.edu.cn (Y.D. Hu)
b
College of Chemistry and Chemical Engineering, Qingdao University, Qingdao, Shandong, China, 266071, email: yanwang@qdu.edu.cn

Received 22 May 2020; Accepted 20 November 2020

abstract
Accurate description and analysis of the dynamic behavior of state variables in multi-effect distil-
lation (MED) desalination process is a challenge for its characteristic of numerous variables and
their complex correlation. This paper is aimed at exploring the dynamic response of state variables
with the derived dynamic model in MED desalination process, including evaporate temperature,
salinity, evaporation mass flow rate and brine pool level. Adopting the reported modeling strategy,
a rigorous dynamic model of the MED is established by coupling the dynamic equations of mass,
salt and energy balance of the system, considering the relation between the state variables and the
operating time. The dynamic model is solved in finite time and the influence on state variables is
investigated under specific conditions, that is, by importing disturbance in feed temperature, feed
flow and/or steam flow. From the point view of process control, the indices of maximum deviation
(MD), response time (RT) and transition time (TT) are applied to analyze the transition process of
state variables, serious parameters of which are significant for PID controller and control scheme
deployment, especially in multistage and integrated process. The parameters of a designed nine
effects advection MED plant in China are taken as the initial input data of the model in simulation
and analysis part. Simulation results demonstrate the transition process and the approximate data of
MD/RT/TT for the state variables in the case plant. The evaluation process might promote the prac-
tical application of dynamic simulation result for control scheme implement and the further control
strategy research.
Keywords: Seawater desalination; Process modeling; Multi-effect distillation; Dynamic simulation

1. Introduction osmosis (RO), membrane distillation crystallization, etc.

[1–7]. It is viewed that three processes namely MSF, MED
With the global issue of water shortage to be solved
and RO have been the dominant technologies for the high
urgently, seawater desalination, as an open source throt-
performance ratio (PR) of fresh water [8,9]. Comparing with
tling technology to the worldwide challenge has aroused
membrane separation process, MED/MSF takes the advan-
widespread attention. A variety of desalination technologies
tage of simple material and energy flow. MED and MSF are
have been developed on the basis of thermal distillation
usually driven by low-grade source with lower equipment
or membrane separation technique during the last several
investment as well [10]. Moreover, the power consumption
decades, such as multi-stage flash (MSF), multiple-effect
of an MED plant is significantly lower than that of an MSF
distillation (MED), vapor compression distillation, reverse

* Corresponding author.

1944-3994/1944-3986 © 2021 Desalination Publications. All rights reserved.

32 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41

plant. MED is more efficient than MSF from a thermody- plant operation data obtained from MES evaporator at Abu
namic and heat transfer point of view [11,12]. As reported Dhabi solar desalination plant, while both results exhibit
in the literature [13], some plants have been built to oper- reasonable agreement. Calle et al. [25] reported a dynamic
ate with a top brine temperature in the first effect of about model for simulating a pilot MED plant located at CIEMAT-
70°C, which reduces the energy consumption and the poten- Plataforma Solar de Almería (PSA) using a compound solar
tial of seawater scaling. Recent development in thermally collector, in which the solar energy was employed as heat
activated desalination methods has achieved an energy source. Narmine [26] established a dynamic model of the
efficiency less than 2.5 (kW helec)/m3 [14]. Furthermore, with multi-effect evaporation, who investigated the variation of
the synthesis technology of different desalination processes state variables in the evaporator when the heating steam
and heat transfer enhancement technology developing, flow reduces by 10%. Nevertheless, the fluctuation of other
hybrid desalination process based on MED will play a more input variables was not discussed. Later, another dynamic
important role in the future [15,16]. model based on mass and energy balance equations for
In order to improve the PR and decrease the energy the PF-MED-TVC system was proposed by Narmine’s
consumption of fresh water, current MED is designed group. In this model, a modified Runge-Kutta method
with approximately 10 stages with heat transfer tempera- was applied to solve the differential equations simulta-
ture difference between 2°C and 5°C [17]. As a result of neously. In another case, Mazini et al. [27] developed a
the tiny temperature difference in adjacent stages, oper- dynamic model for MED-TVC plants depending on mass
ating conditions should be controlled within a small mar- and energy balances of brine and water. The brine salinity,
gin of error to guarantee the MED desalination system in brine level and temperature were selected as state variables
stable production state. Otherwise, the MED desalination to describe the dynamics of each effect. However, the vari-
system might run off the designed parameters, leading ation of the TVC motive steam flow rate was not analyzed.
to severe issues such as local overheating, severe scaling Shahzad and Ng [28] presented a transient model for the
and decline of production [18,19]. On the other hand, var- hybrid MED plus adsorption/desorption (AD) system and
ious integrated desalination process has been proposed compared with conventional MED systems. Azimibavil
[20,21], the control system design of which is an import- and Dehkordi [29] reported a method for the predication
ant step for industrial practice. Precise control scheme and of the MED unit dynamics, in which the dynamic behaviors
unit should be designed seriously to deal with the random of both water and vapor streams in the tube bundle of an
external disturbances manually or environmentally. The effect were emphasized. Simulation results are of impor-
large-scale process control of highly complex desalination tance for the control of wall fouling in heat transfer tubes.
plants has been largely occupied with conventional propor- Nevertheless, this work merely focused on analyzing a sin-
tional-integral-derivative (PID) controllers [22]. Although gle effect, while the information on the behavior of an entire
conventional techniques may provide a minimum perfor- MED plant was missing. Ge et al. [30] presented a start-up
mance requirement, they fall short of the increasing con- model of MED in combination with a nuclear heating reac-
trol performance demand of stability as the stage num- tor. In this model, both the evaporators and the preheaters
ber increases. Control systems in desalination plants are in each effect were expressed through mass and energy
responsible for keeping the parameters of the plant within conservations. However, the model formulation is more
the specified allowable design limits. Therefore, the first likely to describe the sequence of stationary conditions of
step for control scheme designers is to observe the behav- a system, rather than the representation of a real dynamic
ior of state variables in the system, once the disturbance is model for the MED process. Roca et al. [31] developed
imposed. Shahzad et al. [23] had carried out an experiment a dynamic model for the solar-assisted MED pilot unit.
on MEDAD hybrid desalination cycle and extracted the The aim of this dynamic model is to implement a control pro-
typical temperature profiles for the start-up of the three- cess for optimizing the distillate production. The proposed
stage MED plant, which provided a great contribution model gives key indications toward improving the thermal
for further application of the hybrid process. However, it performance of MED under transient operating conditions.
might be unrealistic for most of the plant to build the exper- As mentioned above, various dynamic simulation mod-
imental device to get the state variable profiles. Dynamic els have been proposed and validated. What is important
modeling and simulation would be a preferred choice. to the proposed work is the dynamic model for MED with
Obviously, accurate simulation model is vital for the thermal vapor compressor plant from Mazini et al. [27],
purpose of grasp the transition process distinctly. To date, which adopting the modeling strategy that the evaporator
a large variety of modeling works have been performed unit is divided into three main subsystems; evaporators
for the simulation of steady-state operations of the MED (effects), condenser and thermo-compressor, considering
process. It is impossible to design an accurate and efficient the evaporate temperature, salinity, and the brine pool level
process control system for MED desalination process with as the state variables. Thus the inner characteristic could
steady-state model. The dynamic model, which is partic- be observed, especially the transition process once the
ularly important in improving both the stability and the disturbance imposed. As far as the process control engi-
operational efficiency of the plant, can be employed to neer concerned, the process transition identification is of
predict behaviors of the plant under transient conditions. significance for PID controller design and control scheme
El-Nashar and Qamhiyeh [24] developed a dynamic model deployment [32]. For instance, the fluctuant deviation
on MED systems for predicting the transient behavior of and discrete time are employed to evaluate the effective-
multi-effect stack-type (MES) distillation plants. This model ness of the designed control systems [33]. In accordance
was validated by comparing the calculated data with actual with the general process of control system design [34],
 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41 33

maximum deviation (MD), response time (RT) and transi- model is considered to be a key factor. Adopting the
tion time (TT) are adopted to analyze the process identifi- modeling strategy from literature [27], the differential
cation in this paper. dynamic model can be deduced based on mass, heat, salt bal-
This paper is organized as follows: in Section 2, process ance and seawater physical property equations. The deriva-
description of MED is given and mathematical dynamic tion process is presented, as follows:
model of the plant is derived based on the mass, salt and BPE correlation:
heat balance equations as well as numerous physical
property equations. The evaporate temperature, salinity, Tb ,i = Tv ,i + BPEi (1)
and the brine pool level could be observed when solving
the derived model. In sections 3, dynamic simulations of with differentiation from Eq. (1):
single-effect evaporator state variables are carried out by
MATLAB software for analyzing the dynamic characteris- dTb ,i dTv ,i dBPEi
= + (2)
tics of the process. In section 4, the dynamic change behav- dt dt dt
ior of state variables for MED process is investigated by
applying disturbance of input variables including the feed BPE is the function of steam temperature and salinity
water temperature, the seawater flow or the steam flow,
and how the dynamic simulation results could be used for dTb ,i ∂BPEi  dTv ,i dBPEi dXi
control scheme design. Finally, section 5 provides some = 1 +  + (3)
dt  ∂Ti  dt dXi dt
concluding remarks.
The quality and density correlation:
2. Dynamic model
Mb ,i = As × Li × ρi (4)
2.1. Description of the evaporation process
According to the heat and mass transfer process, a sin- with differentiation from Eq. (4):
gle evaporator can be divided into three lumps, that is, the
brine pool, vapor space and tube bundle. Graphical repre- dMb ,i dLi  ∂ρ ∂Tb ,i ∂ρi ∂Xi 
= As × ρi × + As × Li ×  i +  (5)
sentation of a single effect including three lumps is shown dt dt  ∂T dt ∂Xi dt 
 b ,i
in Fig. 1.
As illustrated in Fig. 2, the pre-effective steam is piped The flow rate of evaporation:
through the heating tube bundle while the feed water is
uniformly sprayed from the top of the evaporator on the Wv ,i −1 Li −1 − WfeedC pf (Ti − Tfeed ) + Wb ,i −1C pb (Ti −1 − Ti )
outer wall of the heat transfer tube bundle. The steam Wv ,i = (6)
Li
turns into water entering the water chamber. Part of the
feed water vaporizes into the steam chamber. Meanwhile,
the liquid part is mixed with the salt water from the last Start
effect in the brine pool.
Input [a b], h, x0, y0
2.2. Dynamic model of the MED
Dynamic simulation of the MED process is aimed at
observing the regularity of the state variables in practical x1=x0+h
engineering. Therefore, the consistency of the dynamic k1=f(x0, y0)
k2=f(x0+h/2, y0+(k1*h)/2)
k3=f(x0+h/2, y0+(k2*h)/2)
Feed,i k4=f(x0+h, y0+k3*h)
y1=h*(k1+2*k2+2*k3+k4)/6

Wv,i-1 Water,i x1, y1

Tube bundle
x0=x1
Wv,i y0=y1
Vapor space
N
x1>b?
Wb,i-1 Wb,i
Brine pool Y
End

Fig. 1. Internal structure of the evaporator. Fig. 2. Flow chart of the Runge–Kutta method.
34 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41

The derivation of mass balance equation: The derivation of salt balance equation:
dMb ,i
= WF + Wb ,i −1 − Wb ,i − Wv ,i (7) d ( M b , i × Xb , i )
dt = WF × X F + Wb ,i −1 × Xb ,i −1 − Wb ,i × Xb ,i (15)
dt
Combining Eqs. (3), (5)–(7), the following equation is
obtained for the mass balance equations: Combining Eqs. (3), (5)–(7) and (15), the following
equation is obtained for the salt balance equations:
dLi dT dXi
k1 + k2 i + k3 = k4 (8) dLi dT dX
dt dt dt k9 + k10 i + k11 i = k12 (16)
dt dt dt
where
where
 k1 = As × ρi
  k9 = As × ρi × Xi
 k = A × ∂ρi  1 + ∂BPEi  
 2  

s
∂Tb ,i  ∂Ti   k = A × X × ∂ρi  1 + ∂BPEi 
(9)  10  
 s i
∂Tb ,i  ∂Ti  (17)
 k = A × L ×  ∂ρi ∂BPEi + ∂ρi  

 3  
 ∂T ∂Xi   k = A × L ×  X × ∂ρi ∂BPEi + X × ∂ρi + ρ 
s i
  b ,i ∂Xi  i
 11 s i  i i

 k4 = WF + Wb ,i −1 − Wb ,i − Wv ,i  ∂Tb ,i ∂Xi ∂Xi 
 
 k12 = WF X F + Wb ,i −1Xb ,i −1 − Wb ,i Xi

The derivation of energy balance equation:
Thus, we can write derivatives of three mentioned state
d ( Mb ,i × hb ,i ) variables using Eqs. (8), (14) and (17) as follows:
= WF × hF + Wb ,i −1 × hb ,i −1 −
dt
Wb ,i × hb ,i − Wv ,i × hv ,i + Q E ,i (10) 
 dL C ⋅E − B⋅F
 i =
where  dt A ⋅ E − B ⋅ D
 dTi A ⋅ F − C ⋅ D
QE ,i = Wvi −1 × hvi −1 = Ki × As ,i × LMTD (11)  = (18)
 dt A⋅E − B⋅D
 dLi dTi
 dXi k12 − k9 dt − k10 dt
LMTD =
(Tv ,i −1
+ Tfeed ) − 2Tv ,i
(12)  =
T −T   dt k11
ln  v ,i −1 v ,i 
 T −T 
 v ,i feed 
where
Combining Eqs. (3), (5)–(7) and (10), the following
equation is obtained for the energy balance equations:  A = k1k11 − k3 k9

dLi dT dXi  B = k2 k11 − k3 k10
k5 + k 6 i + k7 = k8 (13) 
dt dt dt C = k k − k k
 4 11 3 12
 (19)
where
 D = k5 k11 − k7 k9

 k5 = As × hb ,i × ρi  E = k6 k11 − k7 k10
 
 ∂hb ,i  ∂BPEi   F = k8 k11 − k7 k12
 k6 = As × Li × ρi × ∂T  1 + ∂T  +
 b ,i  i 
 ∂ρi  ∂BPEi  Eqs. (18) and (19) are the derived dynamic model
 As × Li × hb ,i × 1 + 
 ∂Tb ,i  ∂Ti  of MED. So far, a few points about the model should be
 (14) expounded in details. (i) The model is mainly based on
 k = A × L × ρ ×  ∂hb ,i ∂BP PEi ∂hb ,i 
+ + Mazini’s work, and model validation has been proved
 7 s i i  ∂T ∂Xi 
  b ,i ∂Xi to be consistent with a real plant. (ii) The thermal vapor
 compressor model part is not included, for the proposed
 ∂ρ ∂BPEi ∂ρi 
 As × Li × hb ,i ×  i +  work is aimed to investigate the inner state variables of a
 ∂T ∂Xi 
  b ,i ∂Xi nine-­effect desalination during transition process. (iii) For
 simplification of the computational complexity, the heat
 k8 = WF hF + Wb ,i −1hb ,i −1 − Wb ,i hb ,i − Wv ,i hv ,i + QE ,i
transfer coefficient of steam condensation on each stage is
 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41 35

assumed to be a constant (1,500 W m–2 C–1), which is rec- time is approximately 0.7 h. The brine pool level decreases
ommend in chemical process equipment [35] and validated with a maximum deviation of around 20% m. The settling
with experiment in literature [36]. time is about 1.1 h. These results of single-effect evapo-
Dynamic simulation is the numerical solution of the rator should be considered as the minimum boundary of
model at different time points, which is implemented via MED system. Moreover, in order to perform the validation
MATLAB software. The Runge–Kutta method (Fig. 2) of the adopted model, the simulation result analysis of
is adopted for the numerical solution of the differential the proposed and published work [27] is shown in Table 3.
equations. Interval (h) is set as 1 s.

Table 1
3. Dynamic analysis of the single-effect evaporator Initial conditions of the dynamic simulation
3.1. Initial conditions
Variable Value
For transient simulations, the initialization of all
Feed (kg/h) 16,666
time-differential variables depends on a number of initial
conditions. Thus, fixed initial conditions (based on the data Sea water temperature (°C) 25
available from steady-state simulation results) are listed Seawater salinity (mass fraction) 0.035
in Table 1. The data were extracted from a design case in Steam flow (kg/h) 10,000
desalination engineering [37]. It is important to note that Steam temperature (°C) 110
all initial conditions (Table 1) are only used for the numer- Heat transfer area (m2) 476
ical initialization of the solver, but they cannot describe a Evaporation temperature (°C) 69.1
specific initial status of the plant.
Salt chamber level (m) 0.5
The dynamic simulation of the evaporation process is
based on the determination of the structure and param- Brine salinity (mass fraction) 0.068
eters of the equipment. The plant size of the evapora-
tor is calculated according to the formula in the chemical
process design manual and fundamentals of salt water
desalination [38,39] based on initial conditions. Size of
the plant is shown in Table 2.
Dynamic response of the single-effect evaporator to
step disturbances is presented in the following sections.

3.2. Evaluation indices for transition process

The evaluation indices for transition process are pre-
sented in Fig. 3. MD is the maximum deviation between
the initial and fluctuated value of state variables. RT is the
delayed time of state variables when disturbance imposed.
TT is the transition time of the whole process. The proposed
work is aimed to simulate the multi-effect desalination
(MED) plants in dynamic mode with disturbance of manip-
ulated variables imposing and evaluating the transition Fig. 3. Description of MD/RT/TT.
process reasonably for control scheme research.

3.3. Dynamic characteristic analysis of the evaporator

The dynamic simulation time is 2 h. When step change
MD=20%
of the sea water temperature is given at t = 0.3 h, the
variations of effect variables including evaporation tem-
perature, brine level, brine concentration and evaporation
rate in the evaporator is shown in Fig. 4.
As presented in Fig. 4, the first point that should
be emphasized is that the state variables are in oscil- TT=0.7h
lation all the time as the dynamic model is solved
with numerical methods (zoomed in Fig. 4). When the
temperature of seawater rises by 5% at 0.3 h, the evap-
oration temperature increases from 69.1°C to 71°C. The
maximum deviation is 2°C and the transition time is
approximately 0.7 h. The evaporation rate increases from
9,480 to 10,000 kg/h and the transition time is approximately
0.7 h. The brine concentration increases and the transition Fig. 4. Step response of the evaporator.
36 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41

The temperature of each effect restores stable after the the presence of external disturbances with independent
imposed disturbance (5% increase of feed water tem- operating variables. In this study, effects of (i) seawater tem-
perature). Temperature increase ratio is calculated from perature, (ii) seawater flow rate and (iii) steam flow rate
the initial and stable temperature both in simulation variation were investigated.
result and literature data. It should be noted that the
temperature increase ratio is lower than that in the liter- 4.1. Initial conditions of MED
ature data, especially for the first effect. As the thermal
The initial conditions extracted from a designed
compressor part belonging to the process intensification
case [38] for dynamic simulation are shown in Table 4.
(PI) technique is not involved in the proposed work, it is
The dynamic behavior of state variables (temperature T,
reasonable that the temperature increase ratio is lower.
brine level L and brine concentration X) are observed by
Approximately, the simulation result is consistent with
applying disturbance of feed water temperature, feed
that in the literature and the model is validated.
seawater flow or steam flow.
4. Dynamic simulation and analysis of MED 4.2. Effect of variation in feed water temperature
To evaluate the system behavior under transient condi- The effect of temperature variation in the inlet seawa-
tions, a series of analysis were carried out through simulating ter is of great importance for the design of control systems
and plant operation. As a matter of fact, feed water tem-
Table 2 perature can be easily manipulated by changing the oper-
Size of the plant ation of heat exchangers. Feed water is usually applied
for cooling down the outlet streams of distillate or brine,
Evaporator device size and the feed water is pre-heating at the same time.
Diameter (m) 1 The step change of the feed water temperature is
Tube bundle height (m) 1
increased by 5% at 0.3 h. The dynamic response of the param-
eters (temperature, brine concentration, evaporation rate
Brine pool height (m) 1
and brine level) in the process is shown in Fig. 5.

Table 3
Analysis of the simulation result and literature data

ith Simulation result Literature data Error

effect Initial Stable Temperature Initial Stable Temperature magnitude
temperature (°C) temperature (°C) increase ratio temperature (°C) temperature (°C) increase ratio
1 38.0 40.8 7.37% 46.4 57.2 23.28% 15.91%
2 40.5 42.7 5.43% 53.6 60.8 13.43% 8.00%
3 44.3 46.5 4.97% 57.0 62.1 8.95% 3.98%
4 48.2 50.2 4.15% 59.8 63.2 5.69% 1.54%
5 52.3 53.8 2.87% 62.1 64.8 4.35% 1.48%
6 56.5 57.0 0.88% 64.0 65.5 2.34% 1.46%

Table 4
Initial conditions of MED dynamic simulation

1 1–2 2–3 3–4 4–5 5–6 6–7 7–8 8–9

ΔT (°C) / 4.8 4.4 4.3 4.2 4.1 3.9 3.8 2.5
Tv (°C) 70 65.2 60.8 56.5 52.3 48.2 44.3 40.5 38
Wv (kg/h) 10,742 9,887 9,461 8,889 8,704 8,628 8,693 8,883 8,862
Wb (kg/h) 12,291 28,788 35,994 43,771 51,733 59,771 67,744 75,527 83,331
X (mass fraction) 0.073 0.071 0.069 0.068 0.068 0.068 0.069 0.070 0.070
Heat transfer area (m2) 476 915 956 919 921 935 991 1,039 1,575
L (m) 0.5
TS (°C) 110
FS (kg/h) 13,784
TSW (°C) 25
FSW (kg/h) 166,666
XSW 0.035
 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41 37

As shown in Fig. 5, when the feed water tempera- flow rate and brine pool levels were observed, as reported
ture increases, the evaporation temperature, the evapo- in Fig. 6.
rate generation rate and the brine concentration increase, From Fig. 6, the liquid level of the brine chamber
while the brine pool level drops. The MD from the first increases with increasing the feed water flow, whereas the
effect to the last effect increases gradually and the MD of evaporation and the salinity of brine decrease. The MD
brine level in last stage is 30%. The variables of each effect reduced gradually from the first effect to the end effect.
are responded simultaneously. The TT of each effect is The TT of each evaporator in turn increases along the
different and the TT of the final effect is approximately 2.2 h. brine flow direction, and the transition time of the final
effect reaches the maximum value at 3.1 h.
4.3. Effect of variation in feed water flow rate
4.4. Effect of variation in steam flow rate
A similar analysis was performed to investigate the
response of the system with increasing 5% of the feed sea- Variation in the steam flow rate is key controlling
water flow rate at 0.3 h. The behaviors of evaporation tem- operation in MED plants. Meanwhile, the influence of
perature, salinity (brine concentration), evaporation mass steam flow rate is significant for all variables, since it can

TT=2.2h

MD=30%

Fig. 5. Transient response of the system to a step variation of the feed water temperature: (a) evaporation temperature,
(b) brine concentration, (c) evaporation rate, and (d) brine pool level.
38 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41

TT=3.1h

MD=7.0%

Fig. 6. Transient response of the system to a step variation of the feed water flow rate: (a) evaporation temperature,
(b) brine concentration, (c) evaporation rate, and (d) brine pool level.

4.5. Simulation result analysis

dramatically affect the distillate production rate and the gain
output ratio of the unit. The dynamic simulation has been performed in the
The dynamic response of the system was analyzed for front part. The evaporate temperature, brine concentration,
several operation variables with the steam flow rate increas- evaporate rate and brine level are selected as the observed
ing by 5% at 0.3 h, as shown in Fig. 7. variables, the stability of which are vital in terms of fresh
As shown in Fig. 7, when the steam flow rate increases, water productivity, energy consumption, safety operation
the evaporation temperature, the evaporation and the salin- or evaporator scaling. No doubt that all variables in the
ity increase, while the salt chamber level drops. The MD dynamic model could be observed for different need in con-
from the first effect to the end effect increases gradually. trol scheme design process. In this paper, the mentioned
All effects are responded at the different times. The TT of four variables are involved in the dynamic simulation part
each evaporator increases effect by effect and the transition and the transition process of the variables is summarized in
time of the final effect achieves the maximum value at 3.5 h. Table 5.
 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41 39

Fig. 7. Transient response of the system to a step variation of the feed steam rate: (a) Evaporation temperature, (b) brine concentration,
(c) evaporation rate, and (d) brine pool level.

5. Conclusion
Table 5 has shown the evaluation of the transition pro-
cess when different disturbance imposed. The maximum In this paper, a dynamic mathematical model based
value has been marked out. It should be explained in on mass, salt and energy conservations for the analysis of
details that the maximum value means to the whole MED MED desalination plants was proposed. The Runge–Kutta
system but not to one evaporator/unit, which is usually method was adopted to solve the proposed model via
larger than that of one evaporator/unit. It is important to MATLAB software. The steady-state simulation results
notice that amplification effect is one essential character- were used as initial conditions to carry out a series of
istics of multi process. Energy consumption saving and dynamic simulations. The maximum deviation (MD),
PR are benefitted from the amplification effect. However, RT and transition time (TT) are adopted to evaluate the
the difficulty of keeping the system stable increases at the dynamic behavior of the state variables (temperature T,
same time due to the amplification effect of deviation in liquid level L and salinity X) in the transition process. It is
cascade process. Control scheme designers should not only found that when the temperature of the feed water rises,
consider the evaporators transition process but also the the evaporation capacity, the evaporating temperature and
MED system from top to end. The obtained result could the salinity of salt stream increase, whereas the liquid level
be referred for further research of control scheme. drops. Maximum deviations of the evaporation amount,
40 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41

Table 5
Evaluation of the transition process

Disturbance Nine effect desalination state Evaluation indices

variables MD/% RT/h TT/h
Evaporate temperature – (0.0) –
+5% Feed water Brine concentration – (0.0) (2.2)
temperature Evaporate rate – (0.0) –
Brine level (–30) (0.0) –
Evaporate temperature – (0.0) –
+5% Feed water Brine concentration – (0.0) (3.1)
flow rate Evaporate rate – (0.0) –
Brine level (+7.0) (0.0) –
Evaporate temperature (+20) (0.2) –
+5% Steam flow Brine concentration – (0.2) (3.5)
rate Evaporate rate – (0.2) –
Brine level – (0.2) –

the liquid level, the evaporator temperature and the brine K — Heat transfer coefficient, W/(m K)
salinity in the multi-effect evaporation system increase Δt — Designed temperature difference, °C
with the fluctuation of the steam flow rate and/or the feed QE — Heat transfer rate of evaporator, kW
seawater temperature, while they reduce with the fluctu- Fs — Steam flow rate, kg/h
ation of the feed flow rate. When the feed flow rate and Fsw — Sea water flow rate, kg/h
the feed temperature fluctuate, the state variables of each
effect respond simultaneously. However, the state vari-
ables respond sequentially when the steam flow fluctuates. Acknowledgment
The simulation results are evaluated with the indices of MD/
RT/TT, demonstrating the inner transition process of MED The authors are grateful to the financial support from
system. The dynamic simulation results provide a basis the National Nature Science Fund Program of China
for the design of MED desalination process control system. (No. 21376231 and No. 21776264).

Symbols References

i — i th effect of the MED process [1] M. Shatat, S.B. Riffat, Water desalination technologies
utilizing conventional and renewable energy sources, Int. J.
L — Salt chamber level, m Low-Carbon Technol., 9 (2014) 1–19.
X — Brine salinity, mass fraction [2] F. Salata, M. Coppi, A first approach study on the desalination
T, Tv — Vapor temperature, °C of sea water using heat transformers powered by solar ponds,
Tb — Brine temperature/°C Appl. Energy, 136 (2014) 611–618.
[3] M.W. Shahzad, M. Burhan, N. Ghaffour, K.C. Ng, A multi
hv — Enthalpy of steam, kJ/kg
evaporator desalination system operated with thermocline
ρ — Brine density, kg/m³ energy for future sustainability, Desalination, 435 (2018)
BPE — Boiling point elevation, °C 268–277.
Mb — Brine flow, kg/h [4] M.W. Shahzad, K.C. Ng, An improved multievaporator
As — Heat transfer area, m2 adsorption desalination cycle for gulf cooperation council
countries, Energy Technol., 5 (2017) 1663–1669.
WF — Feed water flow, kg/h
[5] M.W. Shahzad, K.C. Ng, On the road to water sustainability in
XF — Seawater salinity, mass fraction the Gulf, Nat. Middle East, 50 (2016), doi: 10.1038/nmiddleeast
Ts — Steam temperature, °C .2016.50.
Tsw — Sea water temperature, °C [6] Y.-D. Kim, K. Thu, A. Myat, K.C. Ng, Numerical simulation
Xsw — Sea water brine salinity, mass fraction of solar-assisted multi-effect distillation (SMED) desalination
systems, Desal. Water Treat., 51 (2013) 1242–1253.
t — Time, h
[7] H.T. Wang, B.A. Li, B. Liu, Status of seawater desalination
Tfeed — Sea water temperature, °C technology and presents of novel technology, Sal. Chem. Ind.,
Wb — Brine flow, kg/h 6 (2014) 1–5.
Wv — Evaporation flow, kg/h [8] M.W. Shahzad, M. Burhan, K.C. Ng, A standard primary energy
hF — Enthalpy of seawater, kJ/kg approach for comparing desalination processes, NPJ Clean
Water, 1 (2019) 1–7.
hb — Enthalpy of brine, kJ/kg
[9] K.C. Ng, M.W. Shahzad, H.S. Son, O.A. Hamed, An
Cpf — Specific heat capacity of water, kJ/(kg °C) exergy approach to efficiency evaluation of desalination,
Cpb — Specific heat capacity of brine, kJ/(kg °C) Appl. Phys. Lett., 110 (2017) 1841011–1841015, https://doi.
LMTD — Logarithmic mean temperature difference, °C org/10.1063/1.4982628.
 J.Q. Hu et al. / Desalination and Water Treatment 217 (2021) 31–41 41

[10] M.W. Shahzad, K.C. Ng, K. Thu, Future sustainable desali­ [25] A. de la Calle, J. Bonilla, L. Roca, P. Palenzuela, Dynamic
nation using waste heat: kudos to thermodynamic synergy, modeling and simulation of a solar-assisted multi-effect
Environ. Sci. Water Res. Technol., 2 (2016) 206–212. distillation plant, Desalination, 357 (2015) 65–76.
[11] A.D. Khawaji, I.K. Kutubkhanah, J.-M. Wie, Advances in [26] N.H. Aly, M.A. Marwan, Dynamic response of multi-effect
seawater desalination technologies, Desalination, 221 (2008) evaporators, Desalination, 114 (1997) 189–196.
47–69. [27] M.T. Mazini, A. Yazdizadeh, M.H. Ramezani, Dynamic
[12] M.A. Darwish, Desalination Process: A Technical Comparison, modeling of multi-effect desalination with thermal vapor
Proceedings of IDA World Congress on Desalination and compressor plant, Desalination, 353 (2014) 98–108.
Water Sciences, Abu Dhabi, United Arab Emirates, 1995, [28] M.W. Shahzad, K.C. Ng, K. Thu, B.B. Saha, W.G. Chun, Multi-
pp. 149–173. effect desalination and adsorption desalination (MEDAD):
[13] A. Ophir, A. Gendel, G. Kronenberg, The LT-MED process a hybrid desalination method, Appl. Therm. Eng., 72 (2014)
for SW Cogen plants, Desal. Water Reuse., 4 (1994) 28–31. 289–297.
[14] M.W. Shahzad, K. Thu, K.C. Ng, W.G. Chun, Recent [29] S. Azimibavil, A.J. Dehkordi, Dynamic simulation of a multi-
development in thermally activated desalination methods: effect distillation (MED) process, Desalination, 392 (2016)
achieving an energy efficiency less than 2.5 kWhelec/m3, Desal. 91–101.
Water Treat., 57 (2016) 7396–7405. [30] Z.H. Ge, X.Z. Du, L.J. Yang, Y.P. Yang, S.R. Wu, Simulation on
[15] M.W. Shahzad, K.C. Ng, K. Thu, B.B. Saha, W.G. Chun, Multi the start-up of MED seawater desalination system coupled with
effect desalination and adsorption desalination (MEDAD): nuclear heating reactor, Appl. Therm. Eng., 28 (2008) 203–210.
a hybrid desalination method, Appl. Therm. Eng., 2 (2014) [31] L. Roca, L.J. Yebra, M. Berenguel, D.C. Alarcón-Padilla,
289–297. Modeling of a solar seawater desalination plant for automatic
[16] K.C. Ng, K. Thu, S.J. Oh, L. Ang, M.W. Shahzad, operation purposes, Sol. Energy Eng., 130 (2008) 1–8, https://
A.B. Ismail, Recent developments in thermally-driven seawater doi.org/10.1115/1.2969807.
desalination: energy efficiency improvement by hybridization [32] H. Chaudhary, V. Panwar, N. Sukavanum, R. Prasad, Fuzzy
of the MED and AD cycles, Desalination, 356 (2015) 255–270. PD+I based hybrid force/position control of an Industrial Robot
[17] C.J. Gao, G.L. Ruan, Seawater Desalination Technology and Manipulator, IFAC Proc. Volumes, 47 (2014) 429–436.
Engineering, Chemical Industry Publications, Beijing, 2016 [33] M.X. Lu, A. Luo, Chemical Process Control System, Chemical
(in Chinese). Industry Publications, Beijing, 2006 (in Chinese).
[18] F.N. Alasfour, M.A. Darwish, A.O. Bin Amer, Thermal analysis [34] J. Carvajal, G.R. Chen, H. Ogmen, Fuzzy PID Controller: Design,
of ME—TVC+MEE desalination systems, Desalination, Performance Evaluation, and Stability Analysis, Information
174 (2005) 39–61. Sciences, 2000.
[19] C.L. Li, X.Y. He, Study on Model-Free Adaptive Control in [35] J.R. Couper, W.R. Penney, J.R. Fair, Chemical Process
Multiple-Effect Evaporator Method, J. Hunan Univ. Technol., Equipment-Selection and Design, Elsevier, Amsterdam, 2004.
5 (2008) 74–76. [36] M.W. Shahzad, A. Myat, W.G. Chun, K.C. Ng, Bubble-assisted
[20] M.W. Shahzad, M. Burhan, K.C. Ng, Pushing desalination film evaporation correlation for saline water at sub-atmospheric
recovery to the maximum limit: membrane and thermal pressures in horizontal-tube evaporator, Appl. Therm. Eng.,
processes integration, Desalination, 416 (2017) 54–64. 50 (2013) 670–676.
[21] T.M. Missimer, K.C. Ng, K. Thuw, M.W. Shahzad, Geothermal [37] S.C. Wang, Desalination Engineering, Chemical Industry
electricity generation and desalination: an integrated process Publications, Beijing, 2006, pp. 39–46 (in Chinese).
design to conserve latent heat with operational improvements, [38] D.R. Wu, Chemical Process Design Handbook, Chemical
Desal. Water Treat., 57 (2016) 23110–23118. Industry Publications, Beijing, 2009, (in Chinese).
[22] S.W. Sung, J.T. Lee, I.-B. Lee, Process Identification and PID [39] H.T. El-Dessouky, H.M. Ettouney, Fundamentals of Salt Water
Control, Wiley-IEEE Press, 2009. Desalination, Elsevier, Amsterdam, 2002.
[23] M.W. Shahzad, K. Thu, Y.-D. Kim, K.C. Ng, An experimental
investigation on MEDAD hybrid desalination cycle, Appl.
Energy, 148 (2015) 273–281.
[24] A.M. El-Nashar, A. Qamhiyeh, Simulation of the performance
of MES evaporators under unsteady state operating conditions,
Desalination, 79 (1990) 65–83.