In this week’s blog post we will look at the news from the latest version regarding Eulerian multiphase simulations. My colleague Christoffer posted [New features in Simcenter STAR-CCM+ version 2206 overview – VOLUPE Software] a short summary of the Simcenter STAR-CCM+ version 2206 release, a few weeks back. In that post an overview of the news is presented, but in this post, we will dig a little bit deeper and look at the consequences of some of the news and new model types.
Eulerian multiphase modelling
In the Eulerian framework we consider a cell and look at what is entering and leaving that cell in terms of momentum, mass, and energy. The Eulerian models we have in Simcenter STAR-CCM+ are EMP, MMP, VOF, DMP and fluid film.
S-Gamma for EMP-LSI and Multiple Regime Models
S-gamma is a population balance model that is designed to work for the continuous-dispersed workflow. Typically looking at bubbles in water where droplets are not explicitly resolved, but instead represented by a population. New for Simcenter STAR-CCM+ is that the S-gamma model now is also available for multiples flow regime modelling. Before this version, you did set the interaction of the dispersed phases “manually” with a constant value (for interaction length scale and area density) representing bubble and droplet sizes in the dispersed phases on either side of the interface regime (mixed regime). You could of course also use a field function to describes variations in sizes of droplets and bubblers in the dispersed phase. But now, you can give a more complete description of the dispersed phase using the S-gamma model. The S-gamma model accounts for coalescence and break-up (the same as for the continuous dispersed workflow). The biggest addition, is in fact that the model now also predicts bubble entrainment at the free surface.
Looking at the below picture, a description is made of the system now possible to solve. For the mixed regime (where no one phase is dominating over another), the free surface regime and for each dispersed “side” of either the mixed regime or the large scale interface.
The below picture shows an example of the new methodology. A bubble column is simulated, and the large bubbles are solved together with the free surface resolution using LSI. While the smaller sub-grid bubble sizes are determined by S-gamma. There are two new models in Simcenter STAR-CCM+ to predict bubble entrainment rate and diameter at the free surface. The model are Yu (scale separation) and Ma (bubble surface energy). Details on the models can be found under the theory section of S-gamma in the 2206 documentation.
MMP-LSI to LMP resolved transition
A few versions back VOF to Lagrangian transition was introduced. This allows for a sub-grid solution for the smallest droplet in VOF-simulation and that typically avoids long simulation time or dissipated VOF-droplets. In Simcenter STAR-CCM+ version 2206 the same functionality has been provided for MMP-LSI simulations. This now allow for modelling free surfaces, ballistic droplets, and mixtures in the same simulation. The addition compared to VOF-LMP transition is the unresolved mixture which typically is not captured in VOF. The transition parameters are essentially the same as for VOF-LMP transition.
Non-Equilibrium Droplet condensation Model for DMP
In Simcenter STAR-CCM+ version 2021.3, population balance for DMP was introduced, in the form of the S-gamma model. Up until that point the DMP model had limited applicability, being limited to a dispersed phase without any interaction internally and more or less constant size particles. In the version after (2022.1), DMP was made possible together with user defined AMR, another step forward for the applicability of the dispersed multiphase model. In this version,(2206) the inclusion of a non-equilibrium condensation model, moves the DMP model another step towards further applicability. Without getting into too much technical detail in this post for news, the non-equilibrium condensation model calculates the mass transfer rate between dispersed liquid droplets and a single-component gas phase. We now have the ability to model wet steam in low pressure turbine stages and this model can be combined with the further mentioned DMP-news, and with Fluid film together with LMP stripping.
The below picture shows an example of a Moses and Stein Nozzle, where droplet condensation occurs downstream of the throat in the supersonic diverging section. The upper of the two contour plots show the nucleation rate at the location where droplets are forming. The lower contour plot shows the resulting steam wetness. The graph shows a comparison between experimental data compared to the simulation, with relatively good results for the pressure ratio through the nozzle. The methodology contains two major steps, nucleation (Classical nucleation theory) and droplet growth. Details are available in the documentation.
Filtering for adaptive timestep providers
When running your simulation with an adaptive timestep provider, you are generally limiting your timestep to the inherently smallest scale that your provider can find. Since the provider aim to predict the timescale (timestep) required to resolve the physical models simulated, like a free surface in a VOF simulation for instance, where the time step provider maintains the CFL number at a certain level specified to maintain the free sharp surface, the liquid interface between water and air typically. What this functionality does is that it “filters” out the percentage cut-off you allow in the specification. Often you have areas of poor resolution or disjointed droplets of a size smaller than what is needed to answer the engineering question at hand. Then, if you allow for some cut-off, in the example below they use 3%, you essentially look at the frequency of where the timestep provider want to step in and reduce the time step, and you say: Allow the timestep to be slightly higher than my provider suggests. Meaning that you remove the 3% most conservative occurrences of the time step-provider. The example below shows going from 0 percent cut-off to 3% cut-off and that gives for this sloshing tank a 20 times faster simulation.
This is another step towards a more pragmatic simulation, a simulation that is better suited to answer you engineering problem. Earlier release of Simcenter STAR-CCM+ has shown this for VOF particularly in the introduction of the explicit multi step VOF and the MHRIC scheme for the liquid interface. The compatible time-step providers that this work with are:
- Free surface CFL
- Free surface implicit multi-step
- Smoothed convective CFL
Linked VOF Waves – you can now link waves as pointer, this reduces the need to specify a new wave in the tree on several locations when you wish to increase the fidelity of your simulation.
Free surface Implicit Multistep provider for VOF – increase ease of use and manual input.
Minimum diameter for fluid film wave stripping – Prevents unphysically small droplets stripped.
I hope this has been an interesting read. As usual, do not hesitate to reach out if you have any questions to firstname.lastname@example.org.