Colliding Al Jets

Using typical lab parameters, I ran simulations of colliding Al jets. To simplify the problem, I ran in 2-D with the jet width = domain width. This essentially reduces to colliding two slabs.

The slabs are initialized with opposing velocities, and the bottom slab has an initial 10% sinusoidal velocity perturbation on its interface.

I ran two cases: the first is adiabatic, and the second has a special Al cooling table.


Adiabatic

As expected, the slabs collide and create a hot central shock that propagates back towards the boundaries. The central region remains hot and thus at a low density. The interface maintains an awkward shape due to the perturbation, but this shape is fairly constant.

movie​

movie​


Cooling

Note that the legends are much different. This is because I'm trying to highlight the regions of high density and low temperature. With this scaling , you don't see much in the density until about halfway through the simulation. Thus the cooling_rho.gif movie is only of the second half of the simulation.

The simulation starts out differently because cooling occurs immediately at the interfaces of the "jet heads". This quickly morphs into a similar configuration as the adiabatic case in which there is a hot central region with shocks propagating back towards the boundaries.

However, since cooling is now on, this central region eventually cools and forms dense regions. There is a periodicity to this because of the chosen perturbation.

movie​

movie​

There are times at which these high density regions look "clump-like". The key here is in the perturbation. The high velocity regions of the interface collide first and can thus start cooling sooner. Their shock strength is also slightly higher which leads to stronger cooling. This is where high density knots can form.

So unless the jet heads collide at the same time perfectly along the entire interface, you should always get high density knots within the post-shock region.


NTSI

I found that the perturbation that I was using was either too short-lived or not strong enough. I altered it a bit, and got some nice looking instabilities (namely, the Non-linear Thin Shell Instability).

For this simulation, I put a 15% velocity perturbation on both of the colliding flows. The average velocity and the continuously injected velocity from the top and bottom domain boundaries is 60 km/s. As a reminder, here are all the relevant parameters: 

nslab = 1e-5 g/cc
tslab = 10 eV
vslab = 60 km/s (initially +/- perturbation)

namb = 1e-8 g/cc
tamb = 0.1 eV

runtime = 100 ns
domain size = 2 cm x 2 cm
resolution = 64 x 64 cells + 3 AMR levels

The ambient is sufficiently low density and low temperature to play the role of a vacuum.

Note that in the following image/movie, the legend is constantly changing. This is because the cooling keeps increasing the maximum density and I wanted to always have the densest regions in red. The densities are in computation units (so in units of 1e-5 g/cc).

movie​

Once you enter the highly non-linear regime of the NTSI, you get filaments of high density scattered throughout the cooling region. Below is temperature in eV.

movie​

And just for comparison, here is the density image/movie for a simulation without cooling. Due to the shape of the perturbation, you get an interesting pattern, but you can see that it does not have the NTSI. Thus, so there is no high density fragmentation.

movie​

Attachments (15)

Comments

No comments.