SLIDE 1 Simu Simula lation tion of
Free ee-Su Surf rfac ace e Flo lows ws Wit ith h ST STAR AR-CC CCM+ M+
Samir Muzaferija and Milovan Perić CD-adapco
SLIDE 2 Contents
Introduction to multiphase flows Theoretical background for VOF-method High-Resolution Interface-Capturing (HRIC) scheme Accounting for surface tension effects Extensions of VOF-method Waves: generation and propagation Free surface flows: application examples Future development
SLIDE 3
Introduction to Multiphase Flows
VOF-approach is suitable, when the grid is fine enough to resolve the interface between two immiscible fluids. Sometimes not all parts of the flow are suited for VOF-treatment… Examples: Atomization nozzle flow and jet break-up (right) and flow around a hydrofoil (below)
SLIDE 4 Interface Conditions
- Conditions at an interface between two immiscibe fluids:
Kinematic condition: No flow through interface.
Dynamic conditions: Balance of normal and tangential stresses (surface tension forces):
SLIDE 5 VOF: Theory, I
- VOF considers a single effective fluid whose properties vary
according to volume fraction of individual fluids:
- The mass conservation equation for fluid i reads:
- It can be rearranged into an equation in integral form:
This equation is used to compute the transport of volume fraction αi.
SLIDE 6 VOF: Theory, II
- The mass conservation equation for the effective fluid is
- btained by summing up all component equations and
using the condition:
- The integral form of mass conservation equation (used to
compute pressure correction) reads:
- The properties of effective fluid are computed according to
volume fractions:
SLIDE 7 Interface-Capturing Method, I
- For sharp interfaces, special discretization for convective
terms in the equation for volume fraction αi is needed (to avoid excessive spreading).
- The method must produce bounded solutions, i.e. each
volume fraction must lie between 0 and 1 and the sum of all volume fractions must be 1 at each control volume.
- Bounded schemes must fall within a certain region of the
normalized variable diagram; the normalized variables are defined as:
SLIDE 8 Interface-Capturing Method, II
- The boundedness requirement:
The normalized variable diagram and the proposed high-resolution interface- capturing (HRIC) scheme (details available in STAR- CCM+ documentation)
SLIDE 9 HRIC-Scheme, IV
Simulation of sloshing in a tank due to sinusoidal sway motion:
- ne-cell sharp interface before wave overturns (left) and smeared
Interface after splashing (right), when the interface is in reality not sharp…
SLIDE 10 Interface Sharpening
- In order to prevent dilution, one can activate “interface sharpening”
by setting “Sharpening factor” to a value >0.
- The sharpening model is based on “anti-diffusion” and acts only in
cells at the interface…
- This is usually required only for violent sloshing and similar
phenomena…
SLIDE 11 Local Grid Refinement, I
- One should, when possible, align grid with free surface where it is
flat…
- One should, when possible, avoid vertical grid coarsening in free-
surface zone where its deformation is small…
- The reason: volume fraction is convected into finer cells and leads
to smeared interface…
Flow around a vertical cylinder – two grids for the same initial free surface position
SLIDE 12
Local Grid Refinement, II
Initial value from this cell feeds into next two, from there into next four – the smeared interface does not get sharper by refining time step (only “Sharpening Factor” helps – but it is better to adapt the grid to free surface that to use artificial anti-diffusion…) Impulsively started flow around a vertical cylinder
SLIDE 13 Surface Tension Effects, I
- The kinematic interface condition is implicitly accounted
for by the transport equation for volume fraction.
- The dynamic interface conditions require additional forces
in the momentum equations in cells containing free surface…
- Surface tension forces are converted to volume forces:
Since the gradient of volume fraction is zero away from interface, these terms are equal to zero everywhere except along interface…
SLIDE 14 Surface Tension Effects, II
- The unit vector normal to interface is obtained from the
gradient of volume fraction:
- The curvature of free surface is obtained from the
divergence of the unit vector normal to interface:
- The volume fraction field needs to be smoothed before
the curvature is computed (sharp interface leads to a non- smooth curvature field).
SLIDE 15 Surface Tension Effects, III
- The so called „parasitic currents“ can develop, if the fluid
moves only slowly or not at all, and the surface tension effects dominate (high curvature or surface tension coefficient)...
- The reason: pressure and surface tension forces must be
in equilibrium when fluid is at rest – but the numerical approximations do not guarantee that (one term is linear and the other is non-linear):
- There are many partial solutions to this problem in
literature, but none works in all situations…
SLIDE 16 Surface Tension Effects, IV
- Recently, a new model called “Interface Momentum
Dissipation” was introduced in STAR-CCM+ to reduce the effects of parasitic currents…
- The momentum dissipation term is added to the
momentum equations only in the vicinity of the interface…
- It acts similarly as an increased fluid viscosity near
interface (more on the gas side): µint grad(v)
- Interface Momentum Dissipation decreases rapidly with
distance from interface…
SLIDE 17 Surface Tension Effects, V
- Where free surface is in contact with wall, contact angle
needs to be prescribed.
SLIDE 18 Surface Tension Effects, VI
- One can distinguish between:
Static contact angle Dynamic advancing contact angle on dry surface Dynamic advancing contact angle on wet surface Dynamic receding contact angle
- The contact angle is enforced as:
nfs = - nw cos θw + tw sin θw
SLIDE 19
Interface Momentum Dissipation: Ink Jet Droplet, I
Without IMD With IMD
Without IMD, parasitic currents are strong (maximum velocity 35.88 m/s); With IMD, parasitic currents are hardly visible (maximum velocity 8.98 m/s)
SLIDE 20
Without IMD, the interface is smeared behind secondary droplet and at nozzle exit; With IMD, the interface is sharp almost everywhere…
Without IMD With IMD
Interface Momentum Dissipation: Ink Jet Droplet, II
SLIDE 21
Without IMD: Strong parasitic currents, maximum velocity 4.97 m/s (10x web speed) With IMD: Very weak parasitic currents, maximum velocity 0.506 m/s (1% above web speed)
Interface Momentum Dissipation: Flow in a Slot Coater, I
SLIDE 22
Interface Momentum Dissipation: Flow in a Slot Coater, II
Without IMD: Front meniscus has irregular shape due to high parasitic velocities With IMD: Smooth front meniscus
SLIDE 23
Interface Momentum Dissipation: Flow in a Slot Coater, II
Without IMD: Flow rate at outlet fluctuates due to high parasitic velocities With IMD: Flow rate at outlet fluctuates less
SLIDE 24
Interface Momentum Dissipation: Flow in and Around a Rising Bubble
Left: Without IMD Strong parasitic currents, maximum velocity 11.68 m/s, interface smeared through high velocity normal to it, the flow inside bubble cannot be recognized… Right: With IMD Hardly visible parasitic currents, maximum velocity 0.39 m/s (30 times lower than before), interface is sharp (resolved by one cell) and one can clearly see the flow inside bubble…
SLIDE 25 Extensions of VOF-Method
- One can add additional models in the equation for volume
fraction (diffusion, sources) in order to model effects like non-sharp interfaces, phase change etc.
- This is the main advantage of this approach compared to
level-set and similar schemes...
- VOF-framework is already used in STAR-CCM+ for the
following models:
Evaporation and condensation Melting and solidification Cavitation Boiling
SLIDE 26
- STAR-CCM+ provides several wave models:
– For initialization of volume fraction, velocity and pressure fields; – For transient inlet boundary conditions.
- Currently available models:
– 1st-order linear wave theory – Non-linear 5th-order Stokes wave theory (Fenton, 1985) – Pierson-Moskowitz and JONSWAP long-crested wave spectra – Superposition of linear waves with varying amplitude, period and direction of propagation (can be set-up via Excel-file)
Wave Models
SLIDE 27
- Accurate wave propagation requires 2nd-order time-integration
method.
- Second-order method (quadratic interpolation in time) requires
that the wave propagates less than half a cell per time step.
- First-order scheme is always stable but less accurate…
Time-Accurate Wave Propagation
Scaled 10 times in vertical direction…
Stokes 5th-order wave after 11 periods (8.977 s), resolved by 80 cells per wave- length (125 m) and 20 cells per wave height (5 m); damping over the last 300 m
SLIDE 28 Internal Wave Generation
- The source term in equation for volume fraction can be used to
simulate injection and suction…
- … which can be used to create waves at free surface…
- By a suitable choice of the position and shape of the “source zone”
and an appropriate source term function, one can generate waves
- f desired shape…
- The advantage of this approach: waves radiated by a solid
structure can pass over the source region without reflection (which happens when waves are created by inlet boundary conditions)
SLIDE 29
- Improvements to the treatment of contact angle (better
recognition of contact line, distinguishing direction of motion etc.)
- Transition to other multiphase models:
– VOF to Lagrangian and vice-versa – Fluid film to VOF and vice versa
- Eulerian or Lagrangian multiphase models within VOF
phases
Future Developments: VoF
SLIDE 30
Simulation of Pouring