[Nek5000-users] Question on drawing streamlines for Nek5000 data

[nek5000-users at lists.mcs.anl.gov]

Dear All,

I am sorry to launch this question about VisIt here. I submitted it in visit-user.org. But it seems nobody there uses Nek5000. So may I ask any help from you.

I tried to use Visit 2.7.2 to draw streamlines for a flow field through a backward step from Nek5000.

Parameters in Visit:  

step length: 0.1 (size of my mesh between 0.05 and 0.2)

max steps: 10000

source type: line (1 -0.99 1.5) to (1 0.99 1.5) (this line is well inside the flow field down stream of the step (at x=0))

50 sample along line uniformly

integration: forward

scheme: Forward Euler (default Dromand-Prince diverged )



Now the questions is:

1. Only part of the line is positioned with seeds. part of the line is blank with No streamlines at all.

2. According to Visit user manual, step length should agree with average size of mesh, in my case even I reduce the step

to 0.01. The integration is broken .

3. To my knowledge, RK4 is good for its stability. But in this case it cannot even give a good result . However, the 1st -order forward Euler can be used. don't know why yet.

4. Generally before visualizing spectral element results, interpolation is necessary a priori. Sorry I did not do this work here just for simplicity, so the accuracy of streamline may not be quite good. But I don't think this broke the streamline generation. 

5. There is a laminar separation vortex just downstream (about x=2) of the seeding line (x=1), if I want to plot the stream lines across the whole flow field, could you please leave me any hints to do that? In fact, what ever the max steps is set, my streamline seems never extend up to x=1.3

6. Is it possible to introduce adaptive local time step for streamline production? this tech. is quite well known for steady CFD problems.

If you need , I can upload a picture later to show above questions.

I'll appreciate any hints from you.

Zhenyu Zhang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20140425/7ae7d123/attachment.html>