Learning Forum

[Hui Wang]

Recently i want to model the well known experiments by Liu and Agarwal (1974) and the software adopted in my simulation is Fluent 2022 R1. 

The experiment aimed at obtaining the particle behavior under the effect of turbulence dispersion in a fully-developed pipe flow. And the length and diameter of the pipe are 1.02m and 0.0127m respectively. The aerosol particle is olive-oil with the density of 920 kg/m3. The minimum diameter of the oil droplets is 1.4um.

In my mesh, the y+ of the first layer mesh is about 1, which meets the requirements of the Enhanced Wall Function, and the turbulence model is k-w sst. According to many researchers' work, the default DRW model would overestimate the aerosol deposition for tau_p+ 0.2 (particle diameter is 1.4um on the condition of Re 9894), but it is very strange that there is no deposition in my case.

The attached files show my model settings, can anyone help me to check the settings or gieve me some advice on DRW modelling. Thanks in advance.

[Rob]

You shouldn't need that many integration steps! I think the default is around 50k for a steady model. Otherwise the solver settings look sensible (from a quick look). 

How high is the near wall cell? How does that compare to the droplet diameter? Given the droplet size what mechanism are you expecting to help the droplet penetrate the viscous sub layer?

[Hui Wang]

Hi Rob, thank you for your reply! Very sorry for the delayed response! 

The height of the first cell is 2E-05m (center of the cell), and the diameter of droplet is 1.4um-21um. 

In my case, the turbulence dispersion should drive the particles deposits on the pipe wall. Actually, i model this case just following a research paper with the name of  "Towards quantitative prediction of aerosol deposition from turbulent flows", the conclusion is that "The use of isotropic turbulence models resulted
in over-prediction of V+ by more than 3 orders of magnitude for tao_plus 0.2, whilst the anisotropic RSM gave results in good agreement with experiment. " But in my case, even i use the same model settings as the research paper, there is little deposition for the case of 1.4um (tao_plus 0.2). In fact , a lot of research papers show the same results. So i am really confused. It seems that the particles are getting stuck in the boundary layer and never colliding with the wall.

The attached link https://www.researchgate.net/publication/222936725_Towards_quantitative_prediction_of_aerosol_deposition_from_turbulent_flows shows the reference paper. 

BTW, how can i show you my case in this forum? 

Thanks in advance!

[Hui Wang]

@Rob

Hi Rob, thank you for your reply! Very sorry for the delayed response! 

The height of the first cell is 2E-05m (center of the cell), and the diameter of droplet is 1.4um-21um. 

In my case, the turbulence dispersion should drive the particles deposits on the pipe wall. Actually, i model this case just following a research paper with the name of  "Towards quantitative prediction of aerosol deposition from turbulent flows", the conclusion is that "The use of isotropic turbulence models resulted
in over-prediction of V+ by more than 3 orders of magnitude for tao_plus 0.2, whilst the anisotropic RSM gave results in good agreement with experiment. " But in my case, even i use the same model settings as the research paper, there is little deposition for the case of 1.4um (tao_plus 0.2). In fact , a lot of research papers show the same results. So i am really confused. It seems that the particles are getting stuck in the boundary layer and never colliding with the wall.

The attached link https://www.researchgate.net/publication/222936725_Towards_quantitative_prediction_of_aerosol_deposition_from_turbulent_flows shows the reference paper. 

BTW, how can i show you my case in this forum? 

Thanks in advance!

[Rob]

I can only look at images - I'm not permitted to download anything or follow most links. 

At that scale I'd expect the droplet inertia to be very low. So it's likely the turbulent effects will be needed to push the particles through the viscous sub layer. Given the paper was published in 2008 they'll have been using Fluent 6.x you may find the near wall models have been improved a bit!