A pressure inlet was set on the top of tank cross
section. The circular cross section model was
meshed with elements of quadrangle, and pave was
selected as the meshing type.
2.2 Sloshing condition
Liquid fill percentage was used to describe the
volume or the mass of liquid bulk in the tank. It was
defined as the ratio of the height of liquid free
surface to the diameter of the tank. Lateral liquid
sloshing under the condition of different liquid fill
levels was considered, and the liquid fill level was
set to be 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9,
respectively.
In practical situation, lateral acceleration of tank
vehicles is hardly to exceed 1.0 g. therefore, lateral
acceleration acts on tank was set to be 0.1 g, 0.2 g,
0.3 g, 0.4 g, 0.5 g, 0.6 g, 0.7 g, 0.8 g, 0.9 g and 1.0 g,
respectively.
The purpose of this study was to investigate
inherent dynamics of liquid sloshing and to establish
equivalent mechanical model. There hence, free
oscillation of liquid bulk should be simulated to
acquire sloshing parameters. To simulate free
oscillation of liquid bulk, liquid bulk’s CM should
be away from equilibrium position. Two different
methods to obtain the initial condition of free
oscillation were proposed. For the first one, forced
oscillation under the action of lateral acceleration
was carried out first. While the center of mass of
liquid bulk arrives at the highest point, the
simulation of forced oscillation was stopped. Then,
the ended moment in this forced oscillation case was
taken as the initial condition of free liquid sloshing.
For the second one, liquid free surface was set to be
tilted at the very beginning of free oscillation
simulation. Gradient of tilted liquid free surface was
obtained by
()
0
=atan
y
ag
θ
(1)
where a
y
is lateral acceleration acting on tanks.
While the second method to simulate liquid free
oscillation was much easier than the first one, it was
used to simulate liquid sloshing in this study.
2.3 Simulation settings
A 2D planer, transient simulation was used to
simulate liquid sloshing in flat cylinder tank.
Pressure-based was selected as the solver. Air and
water were multiphase flows in tank model, and air
was set to be the primary phase. The volume of
water, and intersection line between air and water
were defined by user defined function. Laminar was
used as the viscous model. Pressure inlet was set on
the top of circular cross section, and the pressure
inlet was also set to be reference pressure location.
Gravity was acted on the liquid bulk, who directs to
the negative y-axis. The scheme of PISO was used
as pressure-velocity coupling method. For other
parameters, they were accepted as software default
setting.
Sloshing force and moment were monitored
during simulation. Liquid bulk’s CM was calculated
and recorded by user defined function.
3 DATA FILTERING BY
WAVELET
Liquid bulk’s CM was used to investigate sloshing
dynamics in this study. Before dynamic analysis,
data filtering was carried out. Many filtering
methods were tried, it turned out that wavelet
filtering has the best performance. Therefore, liquid
bulk’s CM, including x-coordinate and y-coordinate,
was filtered by wavelet. Sym8 was chosen as the
wavelet function, and 8 layer decomposition was
done. Comparison of liquid bulk’s CM before and
after wavelet filtering was presented in Figure 1.
Also, oscillation angle of liquid bulk’s CM, which
was a function of x-coordinate and y-coordinate, was
also presented to illustrate data filtering result.
In Figure 1, oscillation angle was obtained by
()
=atan
y
θ
−
(2)
where x is the x-coordinate of liquid bulk’s CM, y is
the y-coordinate of liquid bulk’s CM.
Noise in raw data was with small amplitude and
did have regular frequencies. Reason that bring
irregular noise should be calculation accuracy error.
For this situation, wavelet filtering was a pleasure
select.
Compared with x-coordinate, y-coordinate has
noise with rather large amplitude. That is to say that
y-coordinate of liquid bulk’s CM has much violent
changes in sloshing, and the change trend along y-
axis has much lower possibility than that along x-
axis.
Oscillation angle obtained by filtered liquid
bulk’s CM is much smoother that that obtained by
raw data, which reveals the correctness of data
filtering.