ENH: regionFaModels: new filmSeparation models for finite-area framework

Review changes
Download
Patches
Plain diff

Merged ENH: regionFaModels: new filmSeparation models for finite-area framework

feature-film-separation-models into develop

Overview 9
Commits 2
Changes 13

Merged Kutalmış Berçin requested to merge feature-film-separation-models into develop 4 months ago

Overview 9
Commits 2
Changes 13

Summary

This merge request involves refactoring the round-corner curvature separation model within the finite-area framework (i.e. curvatureSeparation) and adding a new model that enables the computation of film-separation properties from sharp edges for both full and partial separation cases.

Resolved bugs

#3278 (closed)

Details of new models

The existing curvatureSeparation model is reformed as OwenRyley model, which is based on Owen & Ryley (1983).
- The governing equations for the model are given in the header file documentation.
A new film-separation model is introduced to compute the film-separation properties from sharp edges for full separation (Friedrich et al. (2008)) and partial separation (Zhang et al. (2018)).
- The governing equations for the model are given in the header file documentation.
- Friedrich et al. (2008) concentrated on the limitations of Owen-Ryley model. They developed a separation criterion based on the force equilibrium within the control volume, encompassing the ligament of film starting to detach from the edge. The authors assumed a maximum film thickness of 1e-4 [m]. Factors such as surface tension, gravity, and the film’s inertia were considered. Despite a tendency to predict premature rupturing, this method is considered to be fairly reliable. The method is also the wall-film separation method implemented in ANSYS Forte 18.2.
- Zhang et al. (2018) improved the Friedrich et al.'s model with a separate mass-fraction criterion. In Owen-Ryley and Friedrich et al.'s models of film separation, the entire film within a computational cell detaches from the wall once the separation criteria are met. However, in the Zhang et al.'s model, only a portion of the liquid film—specifically a mass fraction of the total film in the cell—is separated when these criteria are satisfied. Consequently, the mass of the film predicted to separate in this model, which takes into account the film separation mass fraction, is less than that predicted by models without this sub-model.

Risks

The existing curvatureSeparation model produces different results due to the bug fix of #3278 (closed).
The Owen-Ryley model assumes that:
- when the film is detached from the surface, it instantly disintegrates into a mist of droplets, which are tiny relative to the computational mesh. Therefore, this model may not yield accurate results in situations where the liquid departs from the surface as a cohesive mass. – the film travels around a rounded corner of fixed radius. This may limit the model applications on the sharp corners, where the definition of the curvature could be difficult. – the films are thinner than 1e-4 [m]. – the viscous effects and film-surface disturbances are neglected. – the film velocity and film thickness are uniform.
- the criteria which determines whether the film is separated at the relevant face or not is not explicitly formulated.

Tests

The initial serial/parallel tests for the corner and separation identification are carried out using a small test application: Test-corner-identification.zip.

These test cases include sharp edges of various forms, where separation may or may not occur depending on the edge’s orientation relative to the flow direction. They also span different spatial resolutions and parallelization settings.

The figures below illustrate the setup: blue arrows indicate the direction of the flow and the side of the domain, yellow walls highlight faces where separation can occur, and black faces mark areas where no separation takes place.