myprojects-hg-reborn

Fork

I added the code Python-codes/yd_green.py. It works out Yannis's new
expression for the Green function and the velocity distribution at the
wall. It is also able to read and plot some numerical results for a
better comparison.

I added the code postprocess2.py, which is able to work out the average
axial velocity along the whole tube as a function of r (radial
coordinate) on a single snapshot of the velocity field.

I modified the code Python-codes/postprocess.py by adding some extra
functionalities to read & postprocess the output of the DNS simulations.

I added the code Python-codes/postprocess.py which reads some data
arrays and plots out some quantities of interest.

The problems with abs() were due to the bad choice of an initial
parameter. Using a more realistic value, this does not happen any longer
(after setting A1 to 1e8 the problem disappears and I do not need to
use abs(mu1) and abs(myvar1))

I modified the code Python-codes/read-and-plot.py simply by turning into
a comment a part I did not need any longer (the definition of peval)

I updated the code in Python-codes/read-and-plot.py. Now I do not have
to raise the mean mu and the standard deviation myvar1 to the second
power to avoid problems. I just expressed use abs(mu1) and abs(myvar1)
in the fitting function. I now get the same results as before and as
when using R, which means that a log of a negative number is most likely
the origin of these problems.

I added the Python code Python-codes/read-and-plot.py. This code reads a
.csv file and fits the data to a log-normal distribution. The results
are fundamentally identical to those I get using R.
NB: the code does not work if I write down a log-normal distribution in
the usual way (still investigating). I need to write the mean and std of
the distribution as something squared and then the code works OK.

I created a folder for Python scripts and a readme file to keep better
track of them

I updated the R-codes/readme file with the description the
R-codes/nls-LM-read-and-plot.R file.

I slightly modified the file R-codes/nls-LM-read-and-plot.R. Now the
labelling of the axis in the main plots (mu and sigma) is clearer and
the means and standard deviations of the distributions are saved in
separate .csv files.

I added the file R-codes/nls-LM-read-and-plot.R. This file simply fits a
single-mode log-normal distribution using the LM alghoritm (as other
codes used to do), but it also reads and plots the results for multiple
distributions (so it can handle several distributions simultaneously)

I added the code R-codes/green.R which works out the Milne extrapolation
length both for a pulse source and a steady-source. I updated the readme
file by adding a brief note for the code R-codes/green.R.

This time some trivial modifications to R-codes/plot-boundary.R. I
simply estimates the deposition velocity using some expressions in
literature.

Another major modification to the code R-codes/plot-boundary.R which is
now able to read sequentially temperature and concentration profiles to work out the
deposition rate or efficiency.

This time a major update of the code R-codes/plot-boundary.R. Now the
code is able to read a velocity profile and a concentration profile and
work out the axial flux considering separately the contributions of the
layer and of the core. Careful about the input data files: temperature
and velocities may be defined on different domains and have a different
number of non-zero elements.

Again a small correction to R-codes/plot-boundary.R

I modified slightly the code R-codes/plot-boundary.R by correcting what
was probably a small bug.

I modified the project VELA/CVS-and-anaconda.mph. My modifications
amount to de-activating the artificial diffusion terms in the k-epsilon
model inside the anaconda tube and to fix by hand two singular points in
the subsequent thermal analysis.
The singular point were located at the interface between the fluid and
the diaphragm and at the final restriction of the CVS. My previous tests
ran on the CVS alons show that these singularities are unable to modify
globally the solution, but one should be aware that they exist and that
they have been fixed by hand.
I am not solving the problem of thermal transport in the boundary layer
close to the wall neither in the CVS nor in the anaconda.

Again I modified the project VELA/anaconda.mph. This is to prove that in
the end of the day the choice of whether seeing the problem as highly
nonlinear or not does not matter (nor probably does the exact width of
the boundary layer). What rather matters is whether I activate the extra
diffusion terms or not (it definitely seems that the thermal analysis
give better results without any artificial diffusion active).

I solved again the problem for the CVS but with a thickness of the
boundary layer which is now 2.5 mm. Actually the problem is not the
tickness of the layer itself but either the artificial diffusion which
is activated or the fact that I am solving the equations by ticking the
highly nonlinear problem option.

I modified the project VELA/anaconda.mph in order to make some
comparison with the velocity inside the anaconda in the anaconda-and-CVS
project. I slightly changed the width of the boundary layer (from 2.5 to
3 mm) and I also activated some terms in the artificial diffusion
(which were de-activated previously).

I renamed the workspace VELA/test-grid-square2-bis+anaconda.mph which is
now called VELA/CVS-and-anaconda.mph

I added the missing pdf files insice vela2-document

I messed up slightly the content of vela2-document, but I am going to
fix it by re-adding the pdf files

I updated the file VELA/CVS.mph by fixing a problem with the temperature
close to the layer.

I corrected a small bug in the code R-codes/read_two_sets.R (before
lendat was appearing twice unnecessarily).

I added the file VELA/CVS.mph which tries modelling the CVS by using a
mesh with bricks and a boundary layer.

I added the file VELA/anaconda.mph which contains the modelling with
Comsol of the anaconda (mesh with bricks and the use of wall functions
for the velocity)
)

I added the file VELA/test-grid-square2-bis+anaconda.mph, which contains
the results of the latest vela simulations. It has both the anaconda and
the CVS. It contains a boundary layer both along the anaconda and along
the CVS. I maninly used bricks mesh the geometry, but I also used a mesh
with triangular elements for some domains.