simulating flow past a cylinder

LAMMPS LAMMPS Mailing List Mirror
1

dear all

I’m trying to simulate flow past a cylinder using LAMMPS. I’ve been successful in getting it about say 80% right. the only trouble that I’m facing right now is due to the periodic boundary condition along the x axis due to which the cylinder is influencing the part of flow before encountering it. is there any way in which i can delete the atoms exiting from the cell and create fresh atoms to enter the cell, during each step of the run. Is it possible? I’m attaching the code that I’ve written.

regards

in.flowpastcylinder (1.01 KB)

2

LK > dear all
LK >
LK > I'm trying to simulate flow past a cylinder using LAMMPS. I've been
LK > successful in getting it about say 80% right. the only trouble that I'm
LK > facing right now is due to the periodic boundary condition along the x axis
LK > due to which the cylinder is influencing the part of flow before
LK > encountering it. is there any way in which i can delete the atoms exiting
LK > from the cell and create fresh atoms to enter the cell, during each step of
LK > the run. Is it possible? I'm attaching the code that I've written.

you can try using fixed boundaries for the direction of your flow and and then ignore atoms that are lost because they travel beyond the boundary. that would solve one half of the problem.

for the other half, have a look at fix append/atoms. it may need some adjustments, but it seems to serve a similar purpose.

axel.

LK >
LK > regards

3

or use a bigger box with periodic BC.

Steve

4

I have performed similar simulations with granular particles some time ago (albeit in the dissipative hard-sphere limit). There are a number of ways you could do this. I used fix deposit and ignored lost atoms as Axel suggested. This is more like a hack and probably won’t work particularly well if you have dense systems with low mach number.

The periodic system may be easier to set up, but you will have to be careful. Such a system with a rigid cylinder will not technically be statistically stationary (though over what time-scale is certainly a valid question). Energy will cascade from bulk motion to temperature. Some thermostatting and ingenious forcing could help you out.