COMP_DropletphaseSubcycles

For modeling the dynamics of particle-particle and particle-wall interaction very small timesteps might be necessary. These timesteps can be orders of magnitude smaller than the maximum timestep for a participating fluid. If the global timestep is reduced to these small timesteps, then the performance is significantly decreased.

In order to keep a good performance, there is the possibility to resolve the dynamics of the DROPLETPHASE in subcycles.

The functionality is switched on with parameter COMP_DropletphaseSubcycles.

This means that at the beginning of the timestep in DROPLETPHASE it is determined how many substeps are likely needed to fulfill criterions for COEFF_dt_d30 and COEFF_dt_coll in every substep. This number of substeps will be performed. If during subcycling it is realized that the substep size was too big, then this will yield a reduction of the global timestep in the next timestep. The next global timestep ( from DROPLETPHASE perspective ) is only determined by COEFF_dt criterion, as it guarantees sufficient quality of neighborhood information for the particles.

There is also the option to introduce a limit for the maximum number of allowed subcycles: if

is specified, the algorithm will strictly obey a maximum of 10 subcycles, irrespective of possible violations of time step criteria. Other than in the case above, the global timestep will then also be influenced by the specified number and the criterions given by COEFF_dt_d30 and COEFF_dt_coll. If

is smaller than the number of substebs determined by COEFF_dt_d30 and COEFF_dt_coll \( \delta t_{sub}\) is defined by: \begin{align} \Delta t_{sub} = \frac{\Delta t_{glob}}{COMP_DropletphaseSubcycles}\end{align}.

Default: COMP_DropletphaseSubcycles = 0 (subcycling switched off)

The subcycling only gets activated whenever the global time step \( \Delta t_{glob}\) is larger than any of the DROPLETPHASE time steps dictated by

• DELT_dt_AddCond
• COEFF_dt_d30
• COEFF_dt_coll

In this case, the solver will execute multiple subcycles with a reduced time step that satisfies both of these conditions.

Structure of subcycling

At the beginning of a global time step the following is done first:

• Reading of PhysicalProperties
• Computation of layer thickness and curvature (see LiquidLayer)

Then, in each subcycle the following steps are executed:

• Treatment of boundary conditions (in particular wall collisions for %BND_COLLISION%)
• Update body forces defined via gravity, FreeFlight
• Resolve Particle-Particle collisions as defined via ParticleInteraction, see DropletCollisions
• Calculation of the new particle velocities
• Movement the particles (second order displacement)

• For particles near boundary update the distance to boundary virtually by considering the calculated displacement normal to the boundary element.

Currently not included in the subcycling:

• LiquidLayer : modeling of liquid layers as a 2D shallow water phase

Important remarks

Due to the structure of the subcycling procedure the following points should be kept in mind

• Specifying a value not equal 0 here yields that the particle displacement must be done within the DROPLETPHASE-Routine instead of the central displacement-Routine.

• The value supplied via DarcyBasisVelocity will be read before the subcycling and stored in %ind_v0Darcy%. When considering a drag force acting on the droplets (cf. FreeFlight) projecting the LIQUID velocity in every subcycle is often unnecessary. In these cases it is better to store the projected velocity in %ind_v0Darcy% and use this index in the drag equation supplied via gravity.

• The LiquidLayer modeling is not compatible with the subcycling, i.e. using %IDENT_slip% for walls and COMP_DropletphaseSubcycles is incompatible.