Sie sind auf Seite 1von 15

COMPUTATIONAL MULTICORE ON TWO-

LAYER 1D SHALLOW
WATER EQUATIONS FOR ERODIBLE
DAMBREAK
Cassrio A. Simanjuntak
Bayu Akbar R. H. B
P.H. Gunawan

TELKOM UNIVERSITY
OUTLINE
■ Erodible Dambreak
■ Shallow Water Equation
■ Parallel Programming Using OpenMP
■ Numerical Results and Parallel Performance
■ Conclusion
ERODIBLE DAMBREAK
■ Dambreak problem is widely used as a benchmark in fluid flow problems.
■ Can be categorized in two types: dambreak in hard bed and sediment bed.
■ In sediment bed, the sediment can be moved by the interaction between the water
and sediment itself.

Source: youtube.com
SHALLOW WATER EQUATION (SWE)

■ Hyperbolic equations for simulating fluid flows


■ The wave length greater than the water depth
Source: sciencedaily.com
■ Formed by two equations which are mass and
momentum conservation
■ Two or more layers model more dynamically

Source: sciencedaily.com
TWO-LAYER SHALLOW WATER EQUATION

𝜕ℎ1 𝜕(ℎ1 𝑢1 )
+ =0 (1)
𝜕𝑡 𝜕𝑥
𝑔
𝜕(ℎ1 𝑢1 ) 𝜕(ℎ1 𝑢12 + 2 ℎ12 ) 𝜕(ℎ2 +𝑍)
+ = −𝑔ℎ1 (2)
𝜕𝑡 𝜕𝑥 𝜕𝑥
𝜕ℎ2 𝜕(ℎ 𝑢 )
+ 2 2 =0 (3)
𝜕𝑡 𝜕𝑥
h(x,t) is total depth of fluid, u(x,t) is 𝑔
𝜕(ℎ2 𝑢22 + ℎ22 )
𝜕(ℎ2 𝑢2 ) 𝜌1 𝜕ℎ1 𝜕𝑍
velocity, Z(x) topography, g gravity + 2
= −𝑔ℎ2 + (4)
𝜕𝑡 𝜕𝑥 𝜌2 𝜕𝑥 𝜕𝑥
acceleration, ρ is density, η is water
elevation (η = h+Z), x and t are the
space and time
TWO-LAYER SHALLOW WATER EQUATION
■ The primitive variables in (1-4) in a vector form

■ The finite volume method for approximating (1-4) is given as


(6)

where
and

[8] Audusse E, Bouchut F, Bristeau M O, Klein R and Perthame B t 2004 SIAM Journal on Scientic Computing 25 2050-2065
PARALLEL PROGRAMMING USING OPENMP
NUMERICAL RESULTS AND PARALLEL
PERFORMANCE
■ Simulation of erodible dambreak

Initial Condition
■ Simulation of erodible dambreak

[11] Fraccarollo L and Capart H 2002 Journal of Fluid Mechanics 461 183-228
■ Simulation of erodible dambreak
NUMERICAL RESULTS AND PARALLEL
PERFORMANCE
■ CPU time, Speedup, and Efficiency
NUMERICAL RESULTS AND PARALLEL
PERFORMANCE
■ CPU time, Speedup, and Efficiency

The CPU time and speedup performance


of parallel implementation of computer I.

The CPU time and speedup performance


of parallel implementation of computer II.

The CPU time and speedup performance


of parallel implementation of computer III.
NUMERICAL RESULTS AND PARALLEL
PERFORMANCE
■ CPU time, Speedup, and Efficiency
CONCLUSION
■ The numerical solutions of two-layer shallow water equation for ratio of density
greater than 0.1 are shown in a good agreement with the data experiment in
Louvain case.
■ The parallel performances using platform OpenMP are shown satisfying
measurement in speedup and efficiency, by using parallel computing with 4
processors is shown in a good percentage. The best measurement in speedup and
efficiency is observed for Computer III with processor AMD A6-5200 APU Quad-Core.
Using the maximum grid size Nx = 3200, the speedup is obtained 3.716 times and
the efficiency is calculated 92.9%.
■ In parallel performance during erodible dambreak simulation, Computer I with
processor Intel(R) Core(TM) i5-2500 CPU Quad-Core is shown the best computer in
order to accelerate the computational time.
THANK YOU

Das könnte Ihnen auch gefallen