search

andreww / slurry

Simple calculations on a slurry in the F-layer
MIT License
0 stars 0 forks source link

Boundary layer and falling velocity #15

Closed andreww closed 5 months ago

andreww commented 3 years ago

I have a first draft of this section (which needs some cleanup). However, there is some code cleanup needed, I have a couple of questions, and we need a figure.

Code cleanup: @andreww will do this. (1) Make the calculation work in one (numpy.vectorized) function and jit the velocity / Ra optimiser. (2) Make a new notebook looking at this and creating the figure.

Figure: Basically what we have now but also the falling velocity. Add hlines and/or vlines for the limits where our scalings are valid.

Questions (for @scs1cd): (1) according to Zhang and Xu the empirical relationship between Re and C_D is valid to Re=3E5. Our current plots go well beyond this (larger radius). Presumably we should limit our plot to R that is around this limit. (2) You have everything in terms of a kinematic viscosity, but we change the density of the liquid. Would it make more sense to work in terms of dynamic viscosity? (3) I've stripped out all the stuff about the thermal boundary layer, but think we should keep this in the code (which will become available). Does this make sense? (4) Zhang and Xu reference a book by Cliff. Should we? Do you have access? (5) Do we need to worry about the Re limits in Inman et al.?

andreww commented 3 years ago

Diffusivity - figure 7 of Pozzo et al. 2013

andreww commented 3 years ago

Use Kinematic viscosity - use eta but make it clear. Avoid mu - checmical potential. Let it be a function but actually use a scalar with density from PREM and dynamic viscosity from Pozzo

andreww commented 3 years ago

I think we need an alternative way to compute pick the boundary layer thicknesses. Will need to discuss properly with @scs1cd as I'm convinced current approach does not work and making choices based on particular numerical parameters is not helpful when we want to be able to change those parameters. Anyway, until we both have time at the same time to look at this no point trying to move this part of the problem forwards.