[sthompson]

I am modeling labyrinth spillways in reservoir applications and have almost completed all of my simulations at multiple hexahedral mesh sizes.

I am modeling two different labyrinth geometries and one geometry is crashing at just one mesh size at all discharges. I have tried starting it from time 0 with essentially a wall of water behind the weir, I have also done restart simulations from steady state results of finer meshes, but the mesh blocks have the same dimensions

The error is always time-step < dtmin. From one time step to the next the velocities near the spillway sky rocket from ~4 fps to ~250 fps. Here's a photo: https://imgur.com/bpDQbVL
Any help or ideas of where to look would be much appreciated. I have achieved successful results with the mesh on a similar spillway so I am kind of lost at this point.

Some details on the numerics:
VOF=Split Lagrangian
Momentum advection=Second Order
Fluid flow Solver: Solve Momentum and Continuity Equations

I am using two meshes, a coarse upstream mesh and finer,nested mesh surrounding the weir vicinity, the upstream mesh 1.44 of crest width and the weir mesh is 0.48 of the crest width.

any help is much appreciated!

[MSM1985]

I have faced similar situation while modeling an inverted siphon. The following steps might help you:

1) Check the plot of time step vs stability in the simulation window

2) Keep time step smaller than the time step you saw in simulation graph. This can be done in numerics tab under 'Time Step Controls'

3) As solver is iterating a lot and reducing time step, its better to change the 'Covergence Controls' in 'Pressure Solver Options' in Numerics tab. You can increase maximum number of iterations before time step is reduced. This will also prevent dt<dtmin. error

[thunde47]

You can also try using implicit solver for advection.

[sthompson]

Thanks for the help. I ended up getting the simulations to work by using a monotonicity preserving momentum scheme

[carlostyler]

Originally Posted by sthompson

Thanks for the help. I ended up getting the simulations to work by using a monotonicity preserving momentum scheme google street view

Thanks for the update instruction.

[x6102104]

I have faced this error when simulating Piano key weir model with differerent Q. It occured with a higher discharge rates
Still dont have a solution yet.

[sthompson]

Originally Posted by x6102104

I have faced this error when simulating Piano key weir model with differerent Q. It occured with a higher discharge rates
Still dont have a solution yet.

I ended up changing the momentum advection scheme to a monotonicity preserving method and it became stable. It could be that your mesh is too course for the fluid acceleration at a certain point. You could also try the above mentioned advice.

[x6102104]

Originally Posted by sthompson

I ended up changing the momentum advection scheme to a monotonicity preserving method and it became stable. It could be that your mesh is too course for the fluid acceleration at a certain point. You could also try the above mentioned advice.

I have changed to first order momentum advection. It works. But I am confusing what is it that affects the result comparing to using second order.