Professional Documents
Culture Documents
8
>
> C ðT Tm Þ þ Cps DT T < Tm DT
> ps
>
<
hsl hsl
h¼ Cpm þ ðT Tm Þ þ Cpm DT þ Tm DT < T < Tm þ DT (5)
>
> 2DT 2
>
>
: C ðT T Þ þ C DT þ h T > Tm þ DT
pl m ps sl
where Cp denotes the heat capacity. The subscripts s and l repre- where c ¼ ðDx=DtÞ is the lattice speed and c2s ¼ 1=3c2 . The dis-
sent solid phase and liquid phase. hsl and Tm are the latent heat crete force term in Eq. (8) is computed from
and the phase change temperature. When the liquidus temperature
Tl is equal to solidus temperature Ts (Tl ¼ Ts ¼ Tm ), the 1 ei u ei u
solid–liquid phase change is assumed to occur over a temperature Fi ¼ xi 1 2
þ 4 ei F
2sf cs cs
range (Tm DT, Tm þ DT). For the case that the liquidus tempera- (12)
ture is different from solidus temperature (multiple-component 1 ei u ei u
¼ xi 1 þ e i qgbðT ref T Þ
PCM), the phase change temperature Tm is equal to ðTs þ Tl Þ=2, 2sf c2s c4s
and DT is equal to ðTl Ts Þ=2. Cpm is the heat capacity in the
mushy zone, which is defined as Then, the macroscopic quantities are given by
Cps þ Cpl
Cpm ¼ (6) X
8
2 q¼ fi
0
Furthermore, the thermal conductivity in Eq. (3) is a function (13)
of temperature as well, which can be computed from X
8
Dt
qu ¼ ei f i þ F
8 0
2
>
> ks T Tm DT
>
<
ks þ kl The thermal LB evolution equation is proposed in this paper to
k ¼ ks þ ðT Tm þ DT Þ Tm DT < T < Tm þ DT (7)
>
> 2DT update enthalpy, as follows:
>
:k
l T T þ DT m
1
gi ðx þ ei Dt; t þ DtÞ ¼ gi ðx; tÞ gi ðx; tÞ geq
i ðx; tÞ (14)
sg
2.2 Lattice Boltzmann Model. The momentum and mass are
solved with the density distribution, where the evolution equation where gi is the distribution function for enthalpy in direction i, sg
with BGK collision term is described as follows [37]: is the dimensionless relaxation time for enthalpy, geqi is the equi-
librium distribution function for enthalpy, as follows:
1 " !#
fi ðx þ ei Dt; t þ DtÞ ¼ fi ðx; tÞ fi ðx; tÞ fieq ðx; tÞ þ DtFi (8) ei u ðei uÞ2 u2
sf eq
gi ¼ xi h þ Cpl T þ 2 þ -i ðCpl T hÞ
c2s 2c4s 2cs
where fi is the distribution for density in direction i. sf and Dt are
the dimensionless relaxation time and time-step. fieq is the equilib- (15)
rium distribution function for the density which is defined as
where -i is the additional weighting coefficient
" #
ei u ðei uÞ2 u2 8
eq
fi ¼ xi q 1 þ 2 þ 2 (9) >
> X8
cs 2c4s 2cs < x i¼0
i
-i ¼ i¼1 (16)
>
>
where cs is the lattice sound speed. ei and xi are the discrete : x i 6¼ 0
i
velocity and weighting coefficient in direction i, which are deter-
mined by (D2Q9) With the definitions mentioned above, the relations between LB
8 quantities and macroscopic quantities are
>
> 4
>
> i¼0 X eq
>
> 9 gi ¼ h (17)
<
1 i
xi ¼ i ¼ 1; 2; 3; 4 (10)
>
> 9 X
>
> ei geq
> 1
> i ¼ Cpl Tu (18)
: i ¼ 5; 6; 7; 8 i
36
8 X
> ð0; 0Þ i¼0 ei ei geq 2
i ¼ Cpl Tuu þ Cpl Tcs I (19)
>
>
>
> i
> p
< c cos ði 1Þ ; sin ði 1Þp
i ¼ 1; 2; 3; 4
ei ¼ 2 2 where I is the unit second-order tensor. The enthalpy is computed
>
> from
>
> pffiffiffi p p
>
>
: 2c cos ð2i 1Þ ; sin ð2i 1Þ i ¼ 5; 6; 7; 8 X
4 4 h¼ gi (20)
(11) i
For convenience, the following dimensionless properties are 3.2 Natural Convection Without Phase Change. In this sec-
defined: tion, the problem of natural convection without phase change has
been solved to evaluate the present LB model. The schematic of
T Tm x al t Cpl ðTh Tm Þ natural convection is presented in Fig. 3(a). The square cavity is
T ¼ ; x ¼ ; t ¼ ; Ste ¼ ;
Th Tm L L2 hsl full of air, whose Prandtl number is set as 0.71. The left wall and
DT right wall are kept at the temperature of Th and Tc with Th > Tc .
h¼ (44) The upper and bottom walls are adiabatic. Some additional dimen-
Th Tc
sionless quantities are defined as
where Ste represents the Stefan number. Furthermore, the error
between LB model and analytical method is determined as y gbðT Tref ÞL3
y ¼ ; Ra ¼ ; Pr ¼ (46)
follows: L a a
Fig. 3 Schematics of natural convection (a) without phase change and (b) with phase change
Fig. 7 The temperature contours and solid–liquid interface locations at: (a) t* 5 4.0, (b)
t* 5 10.0, (c) t* 5 20.0, and (d) t* 5 30.0
In order to further verify the enthalpy-transforming based LB fi ¼ distribution function for density in direction i (kgm3)
model, the temperature contours and locations of solid–liquid fl ¼ volume fraction of liquid
interface at different moment have been shown in Figs. 7 and 8 fsuml ¼ total liquid fraction
for cases 1 and 2, respectively. Apparently, the locations of Fi ¼ discrete force in direction i (N)
solid–liquid interface captured by enthalpy-transforming-based Fi ¼ discrete force in direction i (kgm3s1)
LB model are close to that by Mencinger’s model [41]. However, fieq ¼ equilibrium distribution function for density in direction i
as mentioned before, the melting front of present LB model is (kgm3)
behind the result of Mencinger, as shown in Fig. 8(d). g ¼ acceleration due to gravity (ms2)
gi ¼ distribution function for enthalpy in direction i (Jkg1)
4 Conclusions geq
i ¼ equilibrium distribution function for enthalpy in direction
i (Jkg1)
In this paper, the enthalpy-transforming based lattice Boltz- h ¼ total enthalpy (Jkg1)
mann model for solid–liquid phase change has been constructed. hsl ¼ latent heat (Jkg1)
The problems of two-region phase change, natural convection L ¼ characteristic length (m)
without phase change, and natural convection with phase change Ma ¼ Mach number
are solved to evaluate the present LB model. The difference of Nu ¼ Nusselt number
temperature distribution between present LB model and analytical p ¼ pressure (Pa)
solution is almost invisible. The natural convection behaviors of Pr ¼ Prandtl number
flow and heat transfer are able to be revealed by enthalpy- Ra ¼ Rayleigh number
transforming-based LB model, and the results agree well with that Ste ¼ Stefan number
of Hortmann et al. [40] and Mencinger [41]. Furthermore, the t ¼ time (s)
higher computational efficiency can be achieved by using present T ¼ temperature (K)
LB model, since no iterations are necessary at each time-step in Tc ¼ cold temperature (K)
the simulation. Th ¼ hot temperature (K)
t ¼ dimensionless time
Funding Data T ¼ dimensionless temperature
u ¼ velocity (ms1)
This work was supported by the Fundamental Research V ¼ volume (m3)
Funds for the Central Universities (2017XKZD05). x, y ¼ coordinates (m)
x , y ¼ dimensionless coordinates
k ¼ thermal conductivity (Wm1K1)
Nomenclature
c¼ lattice speed (ms1)
cs ¼ lattice sound speed (ms1) Greek Symbols
Cp ¼ specific heat at constant pressure (kJkg1K1) a ¼ thermal diffusivity (m2s1)
ei ¼ discrete velocity in direction i (ms1) b ¼ thermal expansion coefficient (1K1)
F¼ force (N) l ¼ dynamic viscosity (kgm1s1)