open TELEMAC-MASCARET The mathematically superior suite of solvers

hello everyone，
I'm learning the program of C-VSM,then I have some questions:
(1)When I set the pro_max_max as 100 or 200,the initial of the pro_max are 2,but when I set 25,it changed to 8.I wonder what conditions determine the initial assignment of the initial pro_max?
(2) I found that when using C-VSM,the .VSP file showed that some adjacent elevations are the same value but with different fractions,is it reasonable?
(3)Since the "C-VSM MAXIMUM SECTIONS" in the .cas file is not the eventual sections,how the program determine the max section?
Thanks for your reply,
Best regards,
Xue.

Dear Xue,

(1) pro_max_max should not have an influence to pro_max at the initial time step. Depending on your initialising the number of vertical layers (=pro_max) at the initial time (AT=0) should be 2. How did you find out about pro_max? With a print in a fortran file? Then which fortran file? Or from the .VSP files? Beware that the VSP files don't show the initial time step. In case you took the number of vertical layers from the VSP file. At which time was the VSP file? At the first time step? Please let me know so I can see if there is a bug or not.

(2) I am not sure that I understand your question correctly. What is adjacent? The adjacent horizontal nodes? adjacent vertical nodes/layers? The composition of the sediment is changing from horizontal node to horizontal node and also from vertical node to vertical node. Depending on your example the distribution can be more smooth or even highly scattering. If your distribution is reasonable or not is hardly to say... My experiences are that non-stable simulations create highly scattering sediment compositions... so it can be a hint to instabilities. But generally I would not link a non-smooth distribution to non-physical behaviour.

(3) pro_max_max (=-VSM MAXIMUM SECTIONS) is the maximal number of vertical sections / layers for a node. The current number of vertical sections / layers for a node is pro_max. This starts from minimal 2 and is increased in each time step with deposition. In order to avoid pro_max > pro_max_max the number of vertical sections / layers are reduced by a compression algorithm.

Best regards,
Rebekka

Dear Rebekka,
Thank you for your reply.
(1)I know the pro_max won't show in the .VSP file. I found that through debugging...And I noticed that when I set the pro_max_max as 100 or 200,the initial value of the pro_max is 2,but when I set it to 24,the the initial value of the pro_max is 8.I wonder why the initial(AT=0) value of pro_max will be 8 instead of 2 when the pro_max_max is 24.

(2)It was my mistake that I didn't make it clear.I found that the printout of "PRO_D" of adjacent vertical layers have some same values but with different "PRO_F",I cut a picture and attach it.

(3)The program I'm debugging now is from examples under gaia,so please ignore the topic I posted in the telemac2d directory...

Honestly,
Xue.

Dear Xue,

(1) In case of max. number of sections = 25 and 5 sediment classes the code makes a compression for the initial conditions. This is not necessary and leads to strange results. You can correct it in cvsp_compress_dp_gaia.f. Please replace line 219 with

IF(PRO_MAX(J) > PRO_MAX_MAX-4*NSICLA-4.AND.LT.GT.0) THEN

(2) Pro_D is the depth of the section and pro_f is the fraction of the sediment. Because you have 5 sediment classes pro_D is for each sediment class the same but the fractions are varying. PRO_D(K,I) means that pro_d is dependent to K which is the section and I which is the sediment class. Of course it is also dependent to the node J. If you add pro_f for your 5 sediment classes it must be 1.0. The sum of all classes must always be 1.

Best regards,
Rebekka

Hello Rebekka,

I understand the sum of PRO_F for each section is 1.0.I don't know why some adjacent vertical layers(like the section of 24 and 23,their depths are all 9.999600736) have the same PRO_D,but like their PRO_F of class 1 (section 24 is 0.316474876 and section 23 is 0.316385502) are different...The same situation occurs in sections 22 and 21.

Best wishes,
Xue.

Dear Xue,
the vertical layers are described as discrete blocks. Inside the block the sediment distribution is constant. To the next block there is a step to a new sediment distribution. I tried to visualise it with a figure...
Therefore K=23 and 24 have values have the same depth Pro_d but different pro_f, because they belong to different "blocks".
Best regards,
Rebekka

Dear Rebbeka,

I have another question hope you can help me. Since no matter how much I set pro_max_max, the initial value of pro_max starts at 2, and then increases the number of sections as sediment deposition or erosion increases. Why is the calculation time for 1000 layers even 22.23 times that for 100 layers (according to your article published in 2017)? because every time step requires updating all dynamic arrays? Or there are other reasons,if there's any,please let me know.

Sincerely,
Xue.

Dear Xue,

you startet with 1 block = 2 sections because your initial vertical sediment distribution is constant. Then at every time step with sedimentation a new vertical layer is set on top. Only layers which lay within the size of the chosen active layer thickness are touched during calculation. But in case of higher number of maximal layers more vertical layers can be involved in the calculation.
Computing time is needed in cvsm for:
1) creating a new vertical layer in case of sedimentation
2) changing the existing sediment distribution of X vertical layers in case of erosion
3) calculating the average sediment distribution of the active layer from X vertical layers

I guess that there is quite some potential of reducing the computing time for cvsm model.

Beware: As it is written in the 2017 paper the results of the computing costs are only roughly transferable to other projects.

Best regards,
Rebekka

Hello Rebekka，
I analyzed models with different number of sections on the same node and plotted the final moment graph. If all calculations start from section 2, why are there significant differences in calculations divided into different layers. As shown in the figure, there are only the fifth type of sediment at 100 layers, but there are two types of sediment at 1000 layers: fourth and fifth.And I summarized the final number of layers corresponding to different pro_pro_max.Why is the difference so significant?

Sincerely,
Xue.

Dear Xue,

according to pro_max_max the reduction algorithm is called at different times and I would expect much more often for small values than for big ones. This changes the results. The "problem" with multi-grain simulation (and also with nature) is that small differences in the conditions create locally more or less big differences. So if you have a look at one point at one time it is not unusual if you see big differences. But the overall behaviour (like width averaged bottom evolution along the river length) should be similar. May be you can find a useful value for the grain distribution. Maybe the mean grain size of the active layer and the mean grain size of a sub-active layer could be such values could be such values mabye additionally combined with averaging about useful horizontal areas.

As for the hirano model the number of vertical layers and their thickness is quite sensitive, unfortunately ...

Best regards,
Rebekka