[daysley]

Hi

i'm currently simulating a VAWT turbine in ANSYS, using the sliding mesh approach. However, initially the turbine spins the inner domain creating a lot of disturbance in the results and takes roughly 3-4 seconds to reach a quasi-steady solution. i've been trying to use adaptive time stepping to run the first say 4 time steps at a larger step size so i can pass through this zone and reach the periodic time faster. however, when i change the settings and run the simulations, it starts to diverge and is unable to solve the porblem so i have to revert back to fixed time stepping which takes too long. is anyone familiar with how to solve this? all help is greatly appreciated! thanks in advance
sam

p.s. this is the set up for the adaptive time steps i tried

trunc. error - 0.01
end time - 10s
min step size - 0.005s
max step size 0.5
min step change factor 0.005
max change factor - 0.5
fixed time steps - 4

i hoped this would run 4 time steps at 0.5s then decrease to 0.005 when it hopefully reaches a periodic solution?

[daysley]

Anyone have any ideas on adaptive time stepping? anything's a help

[oj.bulmer]

Try with reduced minimum timestep, say 0.001, or smaller if you still see the divergence.

OJ

[daysley]

Thanks again OJ, i'll try that tonight and let you know if it works

[daysley]

OJ,

unfortunately reducing the time step size did not work, it ran for a few time steps and then divergence was detected. and a floating point exception error was produced?? I've attached a screen shot of the error. i also tried increasing the order of the solution methods thinking that would help as the error mentioned momentum divergence but to no avail.
thanks again

[CDollarsign]

Maybe check your viscosity ratios and such in your BCs?

[Far]

which pressure-velocity coupling your are using?

Please show pics of mesh and domain!

[daysley]

Thanks for the input Colin and Far,

After looking more into turbulent viscosity on this forum i altered my ratio, hydraulic diameter to 5m (the width of my domain) and turbulent intensity to 5%. i also totally remeshed my domain instead of using a random tri-mesh i mapped the rectangular area and then used a random tri-mesh for the rotor domain due to its complexity. This seemed to work and i have not been stopped with any floating point or turbulent viscosity ratio errors yet.
Thanks again for both your help it was much appreciated.