I am doing a sloshing problem. I define my boundary conditions as functions of time and don't have fluid draining from my tank(boundary condition is wall).

-The solver must be unsteady (1st order implicit)

-the multiphase model is obviously VOF. Turn on the implicit body force option

-the viscous model is laminar

-set the phases. The primary phase should be the lighter

-on the operating conditions panel, enable gravity and specify the operating density as the one of the lighter phase.

-on the boundary conditions panel, turn on the source terms and use my momentum source

-keep my URFs quite big (0.6 and 0.8 for pressure and momentum respectively, 1 for the other parameters)

-discretization schemes: BODY FORCE WEIGHTED for pressure, PISO for pressure velocity coupling (skewness correction=1, neighbour correction=1, turn on the skewness neighbor coupling), FIRST ORDER UPWIND for momentum.

-once I initialize the flow field to create an adaption register for patching the initial fluid location (adapt>region, then patch the liquid volume fraction solve>initialise>patch, select the volume fraction of the liquid as 1)

-keep my time step quite small(0.001).

This is my UDF

#include "udf.h"

double AlnrX;

DEFINE_ADJUST(Accl, domain) { double t;

t = RP_Get_Real("flow-time");

AlnrX = -0.0363*sin(1.1*t);

printf("time = %f\n",t); printf("accel = %f\n",AlnrX);

}

DEFINE_SOURCE(xmom_source, cell, thread, dS, eqn) { double pos[ND_ND]; double rho; double source;

C_CENTROID(pos,cell,thread);

rho = C_R(cell,thread);

dS[eqn] = 0.0; source = AlnrX *rho;

return source; }

But my calculation is wrong comparing with experiment. Please tall me what is wrong on my setting.

Thank you very much!

Lin

How is it wrong?

Grid and timestep independance?

Higher order spatial and temporal discretisation?

Start with a simple dam breaking validation to check your numerics.