TELEMAC-MASCARET Forum: GAIA Soulsby & van Rijn Suspension formula [trunk] (2/2)

Welcome, Guest

Page:
1
2

TOPIC: GAIA Soulsby & van Rijn Suspension formula [trunk]

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 4 months ago #34027

cyamin
OFFLINE
openTELEMAC Guru
Posts: 997
Thank you received: 234

Hello Agnès,

Enabling SETTLING LAG with options:

BED LOAD FOR ALL SANDS                   = YES;YES;YES;YES;YES;YES
BED-LOAD TRANSPORT FORMULA FOR ALL SANDS = 5;5;5;5;5;5 /SOULSBY-VAN RIJN (bed load)
/
SUSPENSION FOR ALL SANDS                   = YES;YES;YES;YES;YES;YES
SUSPENSION TRANSPORT FORMULA FOR ALL SANDS = 4;4;4;4;4;4 /SOULSBY-VAN RIJN (suspension)

did not solve the issue.
I am also confused with the definition of D90. In SUSPENSION_SANDFLOW_GAIA it is hardcoded to 0.002. According to the manual for Soulsby-Van Rijn formula, D50 and D90 are both needed but D90 keyword is deprecated since GAIA.
Best Regards,
Costas

The administrator has disabled public write access.

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 4 months ago #34028

aleroy	Ok, I'm having a look into this. I'll let you know when I have fixed the issue. Cheers, Agnès
	The administrator has disabled public write access.

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 3 months ago #34370

aleroy	Hi Costas, I have just pushed some changes to GAIA in the trunk, so the D90 value can be set by the user when there is only 1 class of sand (otherwise it is calculated), and CSRATIO is correctly used for the suspension. Please let me know if you still have problems even with these changes. Cheers, Agnès
	The administrator has disabled public write access.

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 3 months ago #34371

cyamin
OFFLINE
openTELEMAC Guru
Posts: 997
Thank you received: 234

Hello Agnes,
Thank you for letting me know. I have been testing the squalius branch in the meantime and on the technical side, the computation works. However, even with the latest trunk, I cannot obtain results because I keep getting NaNs in the DIFFUSION OF TRACERS STEP. These occur after some time, not straight away, so it is puzzling.
I admit it is the first time I enable suspension (or tracers in general). However, previous attempts with SISYPHE worked but eventually halted by precision errors in layer.f.
So at this stage, I am investigating the choice of SOLVER, SOLVER ACCURACY etc but nothing seems to tackle the problem. The issue probably lies in my specific project which, is quite complex.
Here are some steering files if you care to have a look:

File Attachment:

File Name: GAICAS.txt
File Size: 5 KB

File Attachment:

File Name: T2DCAS.txt
File Size: 7 KB

Regards,
Costas

The administrator has disabled public write access.

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 3 months ago #34372

aleroy	Ok, just to make sure: during the simulation, there is nowhere where you start getting tidal flats? Have you tried to use a disztributive scheme instead of the characteristics for the advection of velocities and tracers? Cheers, Agnès
	The administrator has disabled public write access.

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 3 months ago #34373

cyamin
OFFLINE
openTELEMAC Guru
Posts: 997
Thank you received: 234

Usually, I don't expect tidal flats due to deposition that soon and water elevation tends to rise near the coastline, but of course I have to check. I will do a scalar run and check on the temporary results.
My desire would be to run distributive schemes with tidal flats, but it proves that characteristic scheme works so much better with Thompson boundary conditions for velocity.
I take your suggestion and will try the 'tidal flat setup' with distributive schemes and see if it gets better.
Costas

The administrator has disabled public write access.

GAIA Soulsby & van Rijn Suspension formula [trunk] 5 years 3 months ago #34383

cyamin OFFLINE openTELEMAC Guru Posts: 997 Thank you received: 234	And you were right! There was a spot where the velocity pattern was such as to cause negative water levels. Enabling tidal flats did solve the issue. Many thanks for your help! Costas
	The administrator has disabled public write access.

GAIA SEDIMENT SLIDE [trunk] 5 years 2 months ago #34450

cyamin
OFFLINE
openTELEMAC Guru
Posts: 997
Thank you received: 234

Hello Agnès,
From a technical point, I keep having issues with SEDIMENT SLIDE, even with suspension de-activated. I get no particular error, but the computation freezes at some point. Cancelling the computation completes the log and the final steps are:
BEDLOAD ONLY:

BEDLOAD_EVOL_GAIA :            6 /           6
 BEDLOAD_SOLVS_FE_GAIA
 END_BEDLOAD_SOLVS_FE
 END_BEDLOAD_EVOL
 END_BEDLOAD_MAIN
 QS_RESULT
 END_QS_RESULT
 APPEL DE MAXSLOPE_GAIA
 RETOUR DE MAXSLOPE_GAIA
 BED1_UPDATE
mpiexec aborting job...

WITH SUSPENSION:

SUSPENSION_SANDFLOW_GAIA
 END SUSPENSION_SANDFLOW_GAIA
 END_BED1_SUSPENSION_ERODE
 END_SUSPENSION_COMPUTATION
 SUSPENSION_BETAFACTOR
 END SUSPENSION_BETAFACTOR
 GAIA PART 1, COMPUTE_SUSP TRUE
 BED1_SUSPENSION_DEPOSIT
 END BED1_SUSPENSION_DEPOSIT
 QS_RESULT
 END_QS_RESULT
 APPEL DE MAXSLOPE_GAIA
 RETOUR DE MAXSLOPE_GAIA
 BED1_UPDATE
mpiexec aborting job...

Best Regards,
Costas

The administrator has disabled public write access.

GAIA SEDIMENT SLIDE [trunk] 4 years 10 months ago #35236

pavans OFFLINE Moderator Posts: 46 Thank you received: 17	Hello Costas, there was a bug in sediment sliding for parallel computations. The bug is now fixed in the trunk. Sorry for this late reply. Best regards Sara
	The administrator has disabled public write access. The following user(s) said Thank You: cyamin