iahr 2015 - cfd simulations of a bubble screen, omahoney, deltares, 02072015
TRANSCRIPT
9 juli 2015
CFD simulations of a bubble screen
Tom O’Mahoney
Mike van Meerkerk
Contents of the talk
• What is a bubble screen?
• What do we eventually want to model?
• What is known from literature on bubble modelling in CFD
• Model setup
• Lab scale simulations
• Field scale simulation
• Conclusions
9 juli 2015
What is a bubble screen?
9 juli 2015
What is a bubble screen?
9 juli 2015
Air mass flow
Throughput of salt
Value of 100
corresponds to the
case without bubble
screen
What is a bubble screen?
9 juli 2015
What is a bubble screen?
9 juli 2015
What is a bubble screen?
9 juli 2015
What is known from literature
• Modelling of bubbles is a complex process
• Within literature there is no consensus about which physical
effects to model (depends on application)
• Large scale measurements of bubble screens is sparse
• Available data may not be satisfactory to validate a CFD model
Main problems identified in the literature (on bubble columns)
• The overestimation of turbulent viscosity, limiting the transient
character of the flow
• The missing 3rd dimension resulting in a higher generation of
turbulent kinetic energy
9 juli 2015
Model setup
• The Euler-Euler approach for multiphase flows is used because
volume fraction of air is large but the bubbles small
• Drag is the only interphase force of importance
• Limitations on the CFL number and mesh size
• RANS model used for turbulence
9 juli 2015
Lab scale simulations
• Based on simulations by Wen and Forrest (1987)
• Approximately 1m deep
• Different flow rates
• Velocity profile measured at different air flow rates
9 juli 2015
Lab scale simulations
9 juli 2015
Lab scale simulations
9 juli 2015
Lab scale simulations
9 juli 2015
Field scale simulation
• Field scale tests, 10 times the depth of the lab scale experiments
• Mesh is scaled accordingly but the bubbles are not scaled
• Fewer air mass flow rates measured
• Different limits of the viscosity ratio tested. Finally no limit given
9 juli 2015
Conclusions and future work
• At lab scale 3D simulations are necessary but this could be due to
the side wall effects
• Those effects seem smaller at field scale so 2D may still be
reasonable there
• High turbulent viscosity ratios make the application of RANS
modelling questionable
• Available validation data is only without salt
• The addition of the lock exchange flow is a work in progress
9 juli 2015