open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I’m not having any luck using COMPUTATION CONTINUED with Sisyphe. It seems to load in bed fractions, but then not use them. It doesn’t seem to use the saved suspended concentrations at all.

Firstly, to clarify, I imagine the purpose of COMPUTATION CONTINUED is to re-use bed sediment fractions and suspended concentrations from the end of a previous run. Correct?

For this I imagine I need to include 1A1,1A2,1A3,CS1,CS2,CS3 in my VARIABLES FOR GRAPHIC PRINTOUTS for the previous run (assuming 3 size-fractions) . Is there anything else that is necessary in the PREVIOUS SEDIMENTOLOGICAL COMPUTATION FILE?

What I’m seeing in the output of the continued run is the bed fractions are correct at time0 (same as the previous run), but by the first timestep have changed completely.

Here is a link to my model: docs.google.com/open?id=0B6vTQb1BDNoAWFh1N2xKOXhxOXM

Really appreciate any suggestions.

Thanks,
Dougal

Upon further investigation..

A run with COMPUTATION CONTINUED and PREVIOUS SEDIMENTOLOGICAL COMPUTATION FILE doesn't use the suspended concentrations from the earlier run - but uses 0 throughout the domain for CONC VOL CL1 and CONC VOL CL2, and then for CONC VOL CL3 it uses the value I have entered in steering file for CONCENTRATION PER CLASS AT BOUNDARIES for class 1, throughout the domain (and for CONC VOL CL4 uses the value I have entered for CONCENTRATION PER CLASS AT BOUNDARIES for class 2, etc.).

Would really appreciate if someone with a better knowledge of how all these various variables are passed around the code could take a look and see what might be going on.

To make sure it's not something I'm doing wrong, could someone confirm that to set up a PREVIOUS SEDIMENTOLOGICAL COMPUTATION FILE, I just make a run with VARIABLES FOR GRAPHIC PRINTOUTS including 1Ax and CSx for the x sediment classes.

And further, does anyone have an example of successfully using COMPUTATION CONTINUED?, ideally with multiple sediment fractions?

Regards,
Dougal

Hello,

We tested this option and it should work.
You basically need to specify all varaibles in your printout file:
*A* : will give all the percent of each class and each layer,
*ES: the thickness of each layers
CS*: the concentration of each class

Hope this works!

Cath

Hi Cath,

Thanks for the clarification.

Are you able to provide an example of a model successfully using COMP.CONT. with multiple sediment fractions, so I might hopefully track down what I'm doing wrong?

With regard to the layers, I haven't specified any number of layers, but the results file when using ..,*A*,*ES,CS* includes two layers. The first layer has the thickness of erodable bed that I define in noerod.f, and changes in time with erosion/deposition. The second layer has thickness 0 (and FRACLAY of 0 for all sediment classes) throughout a run.. Seems odd - is this normal?


Thanks,
Dougal

Anyway,

Simplifying and forgetting about multiple fractions or layers, I still can't get COMPUTATION CONTINUED to use the saved suspended concentrations (CONC VOL CL1) from the previous run.

If I use a value for CONCENTRATION PER CLASS AT BOUNDARIES in steering file, the suspended concentrations within the model are reset to 0 everywhere. If I use EQUILIBRIUM INFLOW CONCENTRATION = YES, I do get suspended concentrations throughout the model domain at first timestep, but they aren't the same as in the previous computation run.

Help?

Hi,

I have the same problem with COMPUTAITION CONTINUED applied on a case with multiple fractions. SISYPHE is reading the start-file and also gives the right fractions in the first output in the result file, but after that it forgets the fractions and run with 100% of Fraction1.
I also had this problem some month ago in another case - there is also a post - but I found no solution.
The computation is working well without start file, but that increases the computation time dramatically.

Many thank,
Gabi

Trying my hand at debugging, I think the previous runs' suspended concentrations are read, but then get overwritten by a call to condim_susp.

I can't understand how condim_susp works - doesn't it just overwrite CS with CS0?

Dougal

To reply to my own question,

When I run with COMP.CONTD., the concentrations are loaded in, then overwritten when init_transport.f calls condim_susp.f

condim_susp.f shouldn't be called if concentrations are present in the restart file. I believe this line of code in init_transport.f is in error..

should the condition instead be
to match the addresses in point_sisyphe.f
This fix works for me, but would appreciate if someone can confirm, and correct as necessary.

Hello,

I think your correction in init_transport.f is right. All variables have been shifted by the newly added orbital velocity, and the shifting should be done for CS in init_transport.f as well.

Thanks and regards,

JMH