Академический Документы
Профессиональный Документы
Культура Документы
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Copyright 2011 American Scientic Publishers
All rights reserved
Printed in the United States of America
SENSOR LETTERS
Vol. 9, 947957, 2011
3D-CFD Method Driven with the Dynamic Data
Using Real-Time Online Monitoring for Temperature
Simulation of Greenhouse
Yanzheng Liu
1
, Jing Chen
2
, Guanghui Teng
4
, Tingwu Xu
3
,
Tijiang Xiaokai
4
, Yazhou Lv
1
, and Yunkai Li
4
1
College of Mechanical and Electrical Engineering, Beijing Vocational College of Agriculture, Beijing 102442, China
2
Department of Information Engineering, Yantai Vocational College, Yantai 264670, China
3
International College at Beijing, China Agricultural University, Beijing 100083, China
4
College of Water Conservancy and Civil Engineering, China Agricultural University, Beijing 100083, China
(Received: 28 August 2010. Accepted: 11 November 2010)
The whole information of greenhouse micro-climate environment could be fully explained with Com-
putational Fluid Dynamics (CFD) numerical analysis, whose advantages were shown by previ-
ous studies, while many input parameters of the current CFD simulation model were xed and
unchanged along early experimental results. It will be created a signicant error with invariant input
parameters values to predict the real-time dynamic system, which hinder seriously the simulation
and prediction for the complex dynamic greenhouse environment. In this paper, the principle of
Dynamic Data Driven Application System (DDDAS) was introduced exploringly. The CFD simulation
model driven with the dynamic data using real-time online monitoring the greenhouse thermal envi-
ronment was established. And taken the North China type multi-span greenhouse covered double
polyethylene for a case, it was used to simulate the airow and temperature characteristics of green-
house under many conditions. The simulation results had the high accuracy (error area is 4.09.7%)
but lowered the measured value. The results show that: the airow eld in the planting area was
presented a wave-type distribution. Under shading net of greenhouse opened, the temperatures
were increased gradually form the center line between the wet curtain and fan to the around along
with the direction of airow movement, which presented a parabola type. While under inner shading
net of greenhouse opened with a small slit at the edge, the temperatures under shading net were
uniform, but the local high temperature region was presented at the top of mid-span greenhouse
for the existence of two vortexes in opposite direction at the top of both sides span. And the CFD
simulation results were applied to establish the scheme of optimal sensor placement.
Keywords: Greenhouse, Microclimate Environment, Computational Fluid Dynamics, Dynamic
Data Driven, Real-Time Online.
1. INTRODUCTION
The key elements of micro-climate environment in green-
house presented obvious spatial variability which effected
by the external environment and regulating method, such
as large-scale multi-span greenhouse with maximum tem-
perature difference above 10
C.
1, 2
With the more demand
for the high additional value plant, such as special owers,
more greenhouses need to be achieved the full information
e +
U
2
2
+V
e +
U
2
2
= j q +
o
ox
\
a
oT
ox
+
o
o,
\
a
oT
o,
+
o
oz
\
a
oT
oz
o(u)
ox
o(v)
o,
o(n)
oz
+
o(ut
xx
)
ox
+
o(ut
yx
)
o,
+
o(ut
zx
)
oz
+
o(vt
xx
)
ox
+
o(vt
yx
)
o,
+
o(vt
zx
)
oz
+
o(nt
xx
)
ox
+
o(nt
yx
)
o,
+
o(nt
zx
)
oz
+j] U (3)
Table I. Input parameters of the CFD model.
Name Numerical Units Remarks
Double polyethylene 319.2 K Measurement
temperature
Wall temperature on 297.8 K Measurement
the eastern side
Wall temperature on 297.2 K Measurement
the west side
Wall temperature on 298.2 K Measurement
the south side
Wall temperature on 296.9 K Measurement
the north side
Soil temperature 295.6 K Measurement
Outdoor air 302.0 K Measurement
temperature
Air density 1.225 kg/m
3
Fluent manual
Air viscosity 0.0242 kg/(m s) Fluent manual
Indoor air heat 1006.43 wm
1
k
1
Fluent manual
transfer coefcient
Indoor air speed 1.79E-05 m/s Fluent manual
Indoor water 29.0 kg/mol Fluent manual
vapor content
Acceleration due 9.81 m/s
2
Fluent manual
to gravity
Atmospheric pressure 101324 Pa Fluent manual
Import air temperature 301.0 K Measurement
form the wet curtain
Import air speed 1.07 m/s Measurement
from the wet curtain
Outlet temperature 290.2 K Measurement
from the fan
Table II. Physical parameters of the materials.
Specic Thermal Absorption
Material Density heat conductivity rate of solar
name kg/m
3
J/kg K W/m k radiation
Soil 1975 2120 2.44 0.92
Double polyethylene 100 1380 0.047
Standard k a Model
o
ot
(jk) +
o
ox
i
(jkU)
=
o
ox
i
j+
j
t
u
k
ok
ox
i
+G
k
+G
b
jaY
M
(4)
o
ot
(ja) +
o
ox
i
(jaU)
=
o
ox
i
j+
j
t
u
a
oa
ox
i
+C
1a
a
k
(G
k
+C
j
G
b
)
C
2a
j
a
2
k
R (5)
G
K
= u
t
ou
ox
2
+
ov
o,
2
+
on
oz
ou
o,
+
ov
ox
2
+
ou
oz
+
on
ox
2
+
ov
oz
+
on
o,
(6)
where, U denotes uid velocity: U = u
i +v
] +n
k (m
s
1
), u, v, n denoting the velocities on axes of x, , and z;
x
i
denote the direction of coordinate; j (kg m
3
) and j
(Pa s) denote water density and dynamic viscous coef-
cient; (Pa) denotes uid pressure; ]
x
, ]
y
and ]
z
denote the
mass forces on 3 axes; when gravity is the only mass force,
]
x
= ]
y
= 0, ]
z
= g; j (kg m
3
) and j (Pa s), denote
respectively the uid density and dynamic viscosity coef-
cient; P (Pa) denotes the uid pressure; e (J) denotes the
unit mass of uid which it is to be able to; T (K) denotes
air temperature; \
o
(W m
1
K
1
) denotes the air thermal
conductivity; q (J) denotes per unit volume of uid to the
heat increment; t (Pa) denotes the stress tensor; k(m
2
s
2
)
denotes turbulent kinetic energy; a(m
2
s
2
) denotes tur-
bulent dissipation rate; j
t
(m
2
s) denotes turbulent viscos-
ity; G
k
(kg m
1
s
2
) and G
l
(kg m
1
s
2
) denote average
velocity gradient and turbulent pulse kinetic energy caused
by buoyancy; Y
M
(kg m
1
s
2
) denote contribution of uc-
tuating dilatation in compressible turbulence to the overall
dissipation rate, Y
M
= 0; R(J kgmol
1
K
1
) denote gas-
law constant, 8.3144710
3
; C
1a
=1.44, C
2a
=1.92, C
j
=
0.09, u
k
= 1.0, u
a
= 1.3.
Sensor Letters 9, 947957, 2011 949
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
3D-CFD Method Driven with the Dynamic Data Using Real-Time Online Monitoring Liu et al.
Fig. 3. Input module of temperature eld parameter.
Discrete Ordinates radiation using radiation (DO)
model,
V (I( r, s) s) +(o+u
s
)I( r, s)
= on
2
uT
2
r
+
u
s
4r
4r
0
I( r, s)I( s,
s
) JD
r
I( r,
D) JD
=
N
i=1
o
i
I( r,
D
i
); (7a)
N
i=1
o
i
= 4r (7b)
q =
4r
DIJD =
N
i=1
o
i
D
i
I
i
(7c)
Fig. 4. Numerical simulation process.
where, s denotes the discrete space angle; D
i
(i =
1, 2, . . . , N) denotes the discrete direction; o
i
denotes the
weighted sphere volume surface area; I( r,
D
i
) (W m
2
K
1
) denotes the radiation intensity.
2.2. Solution Process
The whole greenhouse internal space was selected for the
calculation domain. In order to avoiding reex, stability
calculation, and avoiding turbulence caused by boundary
conditions, the wet curtain at the entrance was extended in
the computational domain, as shown in Figure 1.
Within the computational region, the hexahedron grid-
ding unit is used for CFD analysis. The discrete approach
is the method of nite volume. GAMBIT was used for
gridding generation. By the frequentative gridding gen-
eration, it was found that the simulation precision was
950 Sensor Letters 9, 947957, 2011
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Liu et al. 3D-CFD Method Driven with the Dynamic Data Using Real-Time Online Monitoring
Fig. 5. Sampling points.
increased none with decreasing the mesh size, when the
grid size was lower than 200 mm. The length of each grid
was 200 mm, and the total meshes was 0.25 million. The
grid was shown as Figure 2. The grid quality was very
high, which the skew rate was mainly in 0.4 or less, not
more than 0.7.
The numerical calculation method is the non-coupling
implicit algorithm with denite constants. The rst-order
windward pattern is used for the item of pressure, etc.
The SIMPLEC (Semi-Implicit Method for Pressure-Liked
Equations Consistent) algorithm is selected for coupling
pressure and velocity.
21
The accuracy of convergence is
chosen to be 0.0001. FLUENT6.13 software was used for
solving the mathematical model.
2.3. Boundary Conditions and Its Real-Time
Collection
The air ow was the research object of CFD simula-
tion. The cover material, enclosure structure, and soil were
set as the boundary condition, and it were set to no-slip
Fig. 6. The comparison between simulated and measured value.
boundary, as Wall type. The internal shading net was
set as the Wall boundary, which was zero thickness with
transmittance of 35%. The input parameters of the CFD
model was shown in Table I, and a part of the thermal
properties parameters of the material as shown in Table II.
The boundary function method was used for low velocity
region near the boundary.
The CFD simulation software and greenhouse acquisi-
tion system was linked, and the temperature of air form
the wet curtain, air from the fan, outdoor air, wall, soil
andinleted air speed from the wet curtain were real-time
collected through the acquisition module. And it was read
as the parameter setting path of CFD calculation. The
program was designed with LabVIEW 8.0, and the input
module and procedures chart were shown respectively in
Figures 3 and 4.
3. MODEL CALIBRATION AND SELECTION
3.1. Experimental Design
The North China type multi-span greenhouse covered dou-
ble polyethylene was taken as the simulated object, which
was designed by the Key Laboratory of Agricultural Engi-
neering in Structure and Environment, Ministry of Agri-
culture. It is located in Haidian District (North attitude
39.9
.
(2) Under shading net of greenhouse opened, the tem-
peratures of three cross-section of X = 4.0 m, 10.0 m
and 18.0 m were increased gradually form the center
line between the wet curtain and fan to the around along
with the direction of airow movement, which presented
a parabola type. It was created with absorbing the long
wave radiation released from soil and wall when the air
was moving.
(3) When under inner shading net of greenhouse opened
with a small slit at the edge (taken T
m3
for example), the
local high temperature region was presented at the top of
mid-span greenhouse (Fig. 13(c)), which was created with
the existence of two vortexes in opposite direction at the
top of other two sides span (Fig. 13(d)).
(4) When the inner shading net was closed, the tempera-
ture under shading net was uniform, and the error was less
than 1