Research areas and projects
Multi-component fluid flow
Introduction | lattice Boltzmann method | Binary flow | Result 1 | Result 2 | Result 3 | Applications | References
 |
 |
 |
 |
Introduction
Lattice Boltzmann algorithm for the explicit simulation of dense suspension of deformable particles
Project leaders - Dr Michael Dupin, Dr Ian Halliday, Professor Chris Care
The motivation of this project is to model dense suspensions of drops in any configuration, all of which do not coalesce with any other. That is, to keep the drops separated throughout the whole simulation duration. The behaviour of the flow is mainly driven by the hydrodynamics of the interfaces, it is called 'interface-dominated' (low Re and high density of fluid). The project focuses, then, on the interface properties and aims to improve existing model and, more ambitiously, to build a new model more appropriate for this purpose.
The main applications would take place in
- the food industry (simulation of mixtures in pipes)
- oil industry (simulation of pipelines)
- sludge transport (waste water, etc.)
- sediments transportation problems
- settling, re-suspensions
- blood flow
Complete segregation is ensured by the model and any fluid stays coherent to itself in that it does not diffuses into any other fluid. No interpolation or interface tracking algorithm is needed, the interface between fluids (literally) emerges from the model.
However, at this stage of the project, many further tests are required to verify physical behaviour of the suspensions. We have already improved the model and shown noticeable improvement in the drops' isotropy.
Our current models work with gravity, but in a way which is only quantitatively correct at low Re. But our investigation on density difference aims to extend to moderate Reynolds number flows and to enable the model to handle an important difference of density which would open perspective to real engineering flows.
The major part of this project has been carried in 2D because quicker and less demanding in memory and computational power. However, the model is designed to be easily up-gradable to 3D. Note that a basic 3D programme has been developed showing efficient execution and stability. This simple programme simulating a single drop is undergoing careful measurements in order to validate analytically its hydrodynamics. We stress that the coding had been pushed to minimise memory requirement and speed up the execution.
Qualitative tests have been carried and validated the stability of the model as well as its robustness and ability to handle a large number of droplets. Theoretically, this can be any number and the only limitations are the specifications of the user's machine. With our Origin 300, we were able to model more than 1000 immiscible drops in 2000*2000 nodes.
Viscosity, surface tension between any couple of fluid and wetting properties can be appropriate to simulate flow where different materials for the walls are used which leads to different wetting properties. This also provides a dynamic control of the droplets behaviour with respect to each other: non sticking droplets could be used to model biological cells (assuming that the deformation is small, therefore they keep their surface area) or on the other hand, polymer melt droplets that stick a lot. Via calibration, a coherent set of values for the different parameters could enable real flow related problems. Note that lubrication forces do not have to be embedded into our current model as an explicit layer of fluid between the drop emerges from control applied to the method (via the different wetting properties applied to the fluids). This later property is crucial when dealing with biological cells or other non-sticking materials.
The boundary is implemented by using the well establish mid-link bounce-back method which enables very complicated shapes (because real world is not only made out of straight cubic boxes).
