Hi Marin, I have been attempting to reproduce the results you show f

Instabilities at high Re about 2d-turbulence-python HOT 1 CLOSED

marinlauber commented on June 3, 2024

Instabilities at high Re

from 2d-turbulence-python.

Comments (1)

marinlauber commented on June 3, 2024

Hi Florian,

Thanks for your message. I suppose you made sure to be up to date with the commits if you pulled some time ago.

For this relatively high Reynolds number you have to use an unpadded spectral method, that is set the flow solver to
flow = Fluid(512, 512, 2000, pad=1.)
such that only 512 modes are used. Padding the spectral method reduced the stability in this case. I was able to run it fine for 1 second on physical time. I have attached the code below and the resulting animation.

McWilliams512x512.mp4

Thanks, Marin

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

import time as t
from src.fluid import Fluid
from src.field import McWilliams

# build fluid and solver
flow = Fluid(512, 512, 2000, pad=1.)
flow.init_solver()
flow.init_field(McWilliams)

print("Starting integration on field.\n")
start_time = t.time()
finish = 1.

# loop to solve
while(flow.time<=finish):
    flow.update()
    if(flow.it % 100 == 0):
        print("Iteration \t %d, time \t %f, time remaining \t %f. TKE: %f, ENS: %f" %(flow.it,
            flow.time, finish-flow.time, flow.tke(), flow.enstrophy()))
        flow.write(file="fluid")

end_time = t.time()
print("\nExecution time for %d iterations is %f seconds." %(flow.it, end_time-start_time))