Beruflich Dokumente
Kultur Dokumente
A R T I C L E I N F O A B S T R A C T
Keywords: Large footprint and very low efficiency are main disadvantages of solar chimneys. To resolve this, solar desa-
Solar chimney lination system has been added under the collector of a solar chimney power plant. Generally the collector
Solar desalination ground is completely covered by the desalination pond but here it is shown that more benefit can be achieved by
Hybrid solar systems partial occupation of the collector area. This is performed by implementing the particle swarm optimization
Particle swarm optimization
(PSO) algorithm in conjunction with a one dimensional simulation code. The code is first validated using data of
a laboratory scale solar chimney. Then, optimization results show that for a collector diameter of 250 m and
tower height of 200 m, a solar pond located between radii 85 and 125 m of the collector can maximize the
outcome of the combined system. Generally, dimensions of the desalination system depend on local cost of
building the system and price of electricity and fresh water produced.
⁎
Corresponding author.
E-mail address: akasa@ut.ac.ir (A. Kasaeian).
http://dx.doi.org/10.1016/j.enconman.2017.08.006
Received 31 May 2017; Received in revised form 1 August 2017; Accepted 3 August 2017
0196-8904/ © 2017 Elsevier Ltd. All rights reserved.
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
Nomenclature Subscripts
A area a ambient
C heat capacity bw bracing wheel
c constant c glass cover
d diameter ct cover temperature
eps emissivity ew water evaporation
F force g ground
f friction factor gr ground to roof
Fa aspect factor h heat
g gravity hb beam heat
H section height hd diffusive heat
h convection factor p at constant pressure
I irradiation r roof
k conduction factor ra roof to ambient
L latent heat of evaporation rh roof convection
m mass flow rate rs roof to sky
n number t tower
P pressure wt water temperature
Pr Prandtl number w water
q heat transfer rate g ground
Re Reynolds number
r radius Greek letters
T temperature
t time α absorbtivity
v velocity β transmissivity
w weight θ angle
x particle position vector ε roughness
y local best value ρ density
ŷ global best value σ Stefan-Boltzmann coefficient
z height/depth τ shear stress
μ viscosity
using a condenser installed at the chimney top. The resulting freshwater collector and thus providing warm updraft air for the chimney. So the
having high gravitational potential energy is used for driving power idea is to integrate a solar distillation system under the collector of a
turbines at the bottom of the tower. The performance of this combined solar chimney. With a simplified model, they have shown that the ef-
system was compared with a classical one by Zhou et al. [16]. ficiency of a solar chimney can be improved from 0.74% to 55.35%.
Another concept was presented by Zuo et al. [17] in which the Zuo et al. [18] built a small experimental setup for the combined solar
collector ground is covered with solar distillation ponds. In this system chimney system and found that the efficiency of this system was not
the solar energy is first absorbed by the insulated desalinator ground more than 25% which is notably lower than his prediction.
and thus heating the water layer above it. The resulting vapor is con- In the latter concept, the collector area is fully occupied by the
densed on the desalinator glass roof and collected with predefined desalination system while no obligation exists for this. Here, it is pro-
tubes. The desalinator roof is then cooled by the air moving through the posed that the desalination system may only be installed in some
73
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
regions of the collector instead of completely covering the collector the convection factor suggested by Adams [21]:
ground. The required area and its location under the collector must be
hra = 5.7 + 3.8vw (5)
calculated through an optimization procedure, considering the local
cost of building the system and the price of electricity and fresh water The amount of heat transferred to the air inside the collector is
produced. With this concept better financial benefit can be obtained. calculated using the heat convection factor derived from the
This procedure is started by a 1D code validated by experimental data Gnielinski's relation (Eq. (6)) combined with the Haaland [22] equation
of the solar chimney built by Kasaeian [4]. Then the PSO algorithm is (Eq. (7)) for surface friction:
implemented to optimize the combined system proposed by Zuo [17].
(f /8)(Re−1000) Pr
hrh = ⎛k ⎞
2. Numerical model 1 + 12.7(f /8)0.5 (Pr 2/3−1) ⎝ d ⎠ (6)
3 3.33 −2
In this section, mathematical models used for performance predic- ⎛ ⎛ 7.7 ⎞ ⎛ ∊/d ⎞ ⎞ ⎤
f = 2.7778 ⎡
⎢log10 ⎜ ⎝ Re ⎠ + ⎝ 3.75 ⎠ ⎟ ⎥ , ∊ / d < 10−4
tion of the system are described. For a good estimation, mass, mo- ⎝ ⎠⎦ (7)
⎣
mentum and energy conservation equations and also heat transfer re-
lations are used. Before starting, following assumptions must be made: By taking the same strategy for the ground, we may write:
• No wind condition
∂Tg
(βe αg )b Ihb + (βe αg )d Ihd = qgr −k g + qgh
• No shadow condition ∂z z=0 (8)
• Sun radiation angle effect is neglected.
• Flow conditions are axisymmetric
In Eq. (8), qgr is already calculated. qgh is also calculated as in Eq.
(6). Finally, the energy conservation equation for the flowing air inside
the collector, is
2.1. The collector
∂
qrh + qgh = ρH ⎡v (Cp T )⎤
To write needed equations, an elementary volume of the collector ⎣ ∂r ⎦ (9)
must be considered. Applying the mass conservation law, results in:
1 ∂
(ρvr H) = 0 2.2. The turbine
r ∂r (1)
In this formulation, the slope of the collector roof can be taken into For the turbine, a simplified model is used. Here an ideal wind
account by variation of the height of the passage (H) with radius. The turbine with an efficiency equal to the Betz limit is assumed [7,23,24].
dynamic equations can be obtained by applying the radial momentum Thus:
conservation law, and making use of Fig. 1.
ΔPturbine = 0.5(16/27) ρπrt2 v 2 (10)
∂P Fsupports
−⎛H
⎜ + τr + τg + ⎞ = ρH ⎛v ∂v + ∂v ⎞
⎟
⎝ ∂r r Δθ ⎠ ⎝ ∂r ∂t ⎠ (2)
Kroger [19] has proposed the following equation for the shear stress 2.3. The tower
due to friction for the roof as a smooth surface:
The same strategy as the collector can be applied to the tower for
ρ0.8v1.8μ0.2 the derivation of needed equations. For the momentum, Fig. 3 can be
τr = 0.02
H 0.2 (3) considered.
For the ground as a rough surface, the following is used [19]: ∂pt ⎛ τt πdt + Fbw ⎞ ∂v
− − ⎜ = ρt ⎛g + vt t ⎞
⎟
0.51 ∂z ⎝ At ∂z ⎠
∊ 0.254 ⎡ ⎠ ⎝ (11)
τg = 0.014875ρv 2 ⎛ ⎞ ⎛ μ ⎞ ⎤
⎝ 2H ⎠ ⎢1.75 ⎜ ρv ∊ ⎟ + 1⎥
⎣ ⎝ ⎠ ⎦ (4) The force exerted by tower structure can be estimated according to
the work of Von Backström [25] with Kbw = 0.1:
To finalize, the terms presented in Eq. (2), forces applied from roof
supports must be estimated. To this end, supports are considered as 1
(At Kbw 2 ρt vt2) nbw
cylinders with drag coefficients of 1 [20]. For the collector roof, the Fbw =
Ht (12)
following equations hold for the energy conservation and heat transfer,
as depicted in Fig. 2. By assuming tower walls to be adiabatic, the energy equation for the
In this figure, the amount of heat radiated from the ground to the updraft flow inside the tower may be written as follows:
roof and also from the roof to the sky are denoted by qgr and qrs re-
spectively and obtained by well-known radiation theories. The amount ∂ ∂
ρt vt (Cp Tt ) + (ρ vt gz ) = 0
of heat convection to ambient outside air stream is obtained by using ∂z ∂z t (13)
74
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
Authors believe that the area and location of the desalination system
under the collector must be determined through an optimization pro-
cess, keeping in mind the cost of building the system and the price of its
products. So, covering the whole area may not be efficient (see Fig. 5).
In this study, PSO is implemented. This method was first developed by
Fig. 4. Energy conservation for a section of the combined system. Kennedy et al. [28] in 1995. To optimize the size and location of the
desalination system under the collector, two variables (inner and outer
2.4. The desalination system radius of the desalination annulus) are needed. Thus the position of
each particle in the swarm has two components. Here, 120 particles are
If a desalination system is placed under the collector, the ground at used. These particles are initially spread and can move in a constrained
that section is replaced with an evaporation pond. Here, it is assumed 2D area limited by the radius of the collector. The new position of each
that there is no vapor leakage and that the distillation system is divided particle (x) is obtained by adding a velocity term to its last position.
into independent and separated ponds. Also there is no flow in the x i (t + 1) = x i (t ) + vi (t + 1) (21)
desalination system such that momentum variations are neglected.
Thus the mass conservation for each section implies that The velocity of each particle can be calculated as follows:
vi,j (t + 1) = wvi,j (t ) + c1 ra1,j (t )(yi (t )−x i (t )) + c2 ra2,j (t )(yi (̂ t )−x i (t ))
mew + m w,in + m w,out = 0 (14)
(22)
The energy conservation for the glass cover of the desalination Here, y is the position at which the lowest value of the objective
system reduces to (see Fig. 4): function is encountered by the selected particle during its travel. Also,
75
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
START
Random
Distribution of
Particles in the
Space
Performance
Prediction of The
Combined System
for Each Particle
Determination of
Personal Best
Position
Determination of Calculating the Fig. 7. Solar chimney by Kasaeian et al. [4] [Reprinted with permission from Elsevier].
Global Best Velocity of Each
Position Particle
Check
Termination
NO
YES
END
Fig. 8. Atmospheric conditions in the work of Kasaeian [4] [Reprinted with permission
y ̂ is the position of the particle with the lowest objective function value
from Elsevier].
between all particles. The inertia term (w) was first introduced by
Eberhart [29], representing the radius of the area that each particle can
travel. At the start of the process a large value (near 1) must be assigned parameters (R1 and R2) are constrained between the collector radii.
to the weight such that a particle can move faster and reach the final These costs are calculated as in Ref. [17]. The local price of electricity is
point faster. This value must be gradually decreased during the solution 26 cent per kW h and the price of fresh water is assumed to be 27 cent
to allow a particle to finely move toward the optimum point [30]. ra1 per cubic meter. For better understanding, the optimization flow chart
and ra2 are some random variables, preventing particles from being is shown in Fig. 6.
trapped in local minimums. Coefficients c1 and c2 represent the relative
importance of personal and swarm experiences in determination of the 4. Results and discussion
new path direction. They are generally selected such as their sum,
equals 4. A simulation code has been developed using the discretized form of
The objective function in this work is formulated as follows: equations derived in the last section. The accuracy of this code must be
evaluated before starting the optimization process. So, in Sections 4.1
collector cost + tower cost + desalinator cost
∅= and 4.2, a simple solar chimney and a combined system are simulated
electricity price + freshwater price (23)
and studied.
In this function, the collector and tower cost are constant but the
desalination cost is dependent on its area (R1 and R2). Also, the amount 4.1. Solo solar chimney
of electricity and the produced fresh water are functions of the desali-
nation area (R1 and R2). It should be noted that optimization In this section, the solar chimney discussed in Ref. [4] is studied.
76
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
Fig. 11. Temperature rise of the air stream during a complete day.
77
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
78
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
Fig. 15. The Pareto front diagram. The chimney reported in the work of Zuo et al. [17] is considered as
the base geometry for optimization and further study. In Fig. 14 the
initial distribution of particles and also their final destination are
This solar chimney (Fig. 7) has a collector with the diameter of 10 m
shown. The path of a sample particle is also plotted. This path is just an
and a tower with 12 m height and 25 cm diameter.
illustration of how fast the particles are moving toward the optimal
Several sensors were installed in the collector and the tower to ac-
point. An inefficient or badly configured optimization has its particles
curately measure the temperature and velocity of air in the chimney in
circling around the space without any global aim and thus, needing
different times. Atmospheric conditions were recorded and reported as
much more iterations to attain the final answer.
shown in Fig. 8. The chimney geometry and boundary conditions in-
In the PSO algorithm, the optimum point is where particles finally
cluding weather conditions (temperature and radiation intensity) are
gather in. This point is easily found according to the final position of
imported into the code. Due to the variations of atmospheric conditions
particles. It can be observed that almost all particles are clustered at a
during the day, simulations are performed transiently and the results
point with coordinates 85 and 125. So, the desalination system must be
for the mentioned day are shown in Fig. 9.
located between these radii. In other words the desalination system
It can be observed that the numerical code has predicted the tran-
must be located at the inlet of the collector with a thickness of 40 m.
sient behavior of the chimney with good precision. The maximum de-
The Pareto front diagram as shown in Fig. 15 can help to find an insight
viation is about 0.2 m/s. Another data being reported in Ref. [4] is the
in the relative location of the best answer among all other possible
temperature distribution inside the collector at 12:00. Results are
answers. This diagram shows the locus of performance of all possible
compared in Fig. 10. It can be observed that the maximum deviation is
choices for the location and area of the desalination system. It can be
about 4 K. Generally good agreement between data can be found.
seen that where no fresh water is produced (no desalination system),
Fig. 11 represents the behavior of the solar chimney in a 24 h period
the chimney has the maximum energy yield. Also, when the desalina-
including the night time in which the heat, stored in the ground, reacts
tion system delivers its maximum fresh water by occupying all the
as a source of energy.
collector area (lower right corner of the diagram), the minimum energy
The mentioned deviation can be due to:
yield is obtained. The answer of the optimization problem must be lo-
cated on the upper edge of the diagram, where the maximum electricity
• Error in temperature measurement caused by sensors precision and is obtained with a given amount of fresh water needed. The final answer
location error (non-uniformity of temperature in collector height).
is shown with a star on the diagram.
• Sun radiation angle being neglected in the code The optimization procedure has placed the desalination system at
• Probability of presence of clouds during the test. the inlet section to benefit from cooler air from the environment. So, the
• Neglecting the wind amount of fresh water produced per area of the desalination system is
• 1D Model error maximized. In this configuration the electricity production is about
• Numerical solution error 18.5 kW with 6.5 kg/s of fresh water. It can be seen that the amount of
electricity production is less than a simple solar chimney of the same
4.2. Combined system size with 22 kW of energy production. However, the performance of the
new system is much better than a full hybrid one (11 kW electricity
Since no reliable experimental data of a combined system could be generation). Table 1 is prepared for a better comparison between the
found, results are compared with numerical data in the work of Zuo configurations.
[17]. In his work, a chimney with 200 m tower and 250 m collector In this case, the temperature distribution of air inside the collector is
diameter is simulated. The chimney is simulated in two levels. First the as plotted in Fig. 16. It can be observed that the slope of the graph
simple chimney is studied. Results are shown in Fig. 12. Here a max- experiences sharp changes where the desalination system is terminated.
imum of 2 K deviation in temperature is observed. In this configuration In this zone, the temperature rise accelerates because almost all of the
the system will produce 23 kW of electricity. Zuo has reported an radiation energy is now absorbed by the air.
electricity production rate of 29 kW.
In the second step, Zuo et al. have reported the results for
Table 1
Comparison of different configurations of the combined system.
Simple solar chimney (R1 = R2 = 0) Full combined solar chimney (R1 = 5, R2 = 125) Optimum combined system (R1 = 85, R2 = 125)
79
M. Asayesh et al. Energy Conversion and Management 150 (2017) 72–80
Fig. 16. Air temperature inside the collector of the optimized system.
80