open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

Hum, this should not occur with monotonous advection schemes, like 3,4,5,13,14, unless it comes from the diffusion (there is no monotonicity of diffusion if you have obtuse angles in some triangle in the mesh). So try with a COEFFICIENT FOR DIFFUSION OF TRACERS = 0. and a TURBULENCE MODEL = 1.
There is also a keyword TYPE OF SOURCES that you can set to 2 to have a Dirac treatment of sources, I am not sure it has an effect but it could be.
Otherwise I'll have a look at your case.

Best regards,

Jean-Michel Hervouet

Dear jmhervouet,

I triad according to your advice,but still didn't work.

How can I show you the cas file and res file?

The "add file" icon below seems not available.

/
/

---

/ COMPUTER INFORMATIONS
/

---

/
GEOMETRY FILE = lvsi-j.geo
BOUNDARY CONDITIONS FILE = lvsi-j.bc
FORTRAN FILE = user.f
RESULTS FILE = lvsi-dec-test.res
LIQUID BOUNDARIES FILE = qsl_dq.txt
/VALIDATION = YES
/REFERENCE FILE = ref
PARALLEL PROCESSORS = 0
/
/

---

/ GENERAL INFORMATIONS - OUTPUTS
/

---

/
TITLE = 'LVSI'
VARIABLES FOR GRAPHIC PRINTOUTS = 'U,V,S,B,H,T1,T2'
GRAPHIC PRINTOUT PERIOD = 60
LISTING PRINTOUT PERIOD = 30
TIME STEP = 60.
/DURATION = 3970800.
NUMBER OF TIME STEPS = 1800
MASS-BALANCE = YES
INFORMATION ABOUT SOLVER = YES
/
/

---

/ INITIAL CONDITIONS
/

---

/
COMPUTATION CONTINUED = NO
INITIAL CONDITIONS = 'CONSTANT ELEVATION'
INITIAL ELEVATION = 3.82
/
/

---

/ BOUNDARY CONDITIONS
/

---

/
PRESCRIBED ELEVATIONS = 0.0
OPTION FOR LIQUID BOUNDARIES =2
/
/

---

/ PHYSICAL PARAMETERS
/

---

/
LAW OF BOTTOM FRICTION = 3
FRICTION COEFFICIENT = 32.
TURBULENCE MODEL = 1
VELOCITY DIFFUSIVITY = 1.E-4
/
/

---

/ NUMERICAL PARAMETERS
/

---

/
TYPE OF ADVECTION = 1;5
SUPG OPTION = 1;2
SOLVER ACCURACY = 1.E-3
DISCRETIZATIONS IN SPACE = 12 ; 11
PROPAGATION = YES
SOLVER = 3
PRECONDITIONING = 2
IMPLICITATION FOR DEPTH = 0.6
IMPLICITATION FOR VELOCITY = 0.6
BOTTOM SMOOTHINGS = 1
MAXIMUM NUMBER OF ITERATIONS FOR SOLVER = 300

/
/

---

/ TRACERS
/

---

/
/DENSITY EFFECTS = Yes
/MEAN TEMPERATURE = 10.0
NUMBER OF TRACERS = 2
NAMES OF TRACERS = 'CODMN MG/L ';
'NITROGEN MG/L '
INITIAL VALUES OF TRACERS = 0.0;0.0
MAXIMUM NUMBER OF ITERATIONS FOR DIFFUSION OF TRACERS = 100
COEFFICIENT FOR DIFFUSION OF TRACERS = 0
ABSCISSAE OF SOURCES =369066.0
ORDINATES OF SOURCES =3556468.0
WATER DISCHARGE OF SOURCES =1.042
VALUES OF THE TRACERS AT THE SOURCES =16.67;10.0
/VALUES OF THE TRACERS AT THE SOURCES =0.0;0.0
TYPE OF SOURCES =2
OPTION FOR THE DIFFUSION OF TRACERS =2
/
&FIN

Hello,

I only see the OPTION FOR THE DIFFUSION OF TRACERS as last possible guilty keyword, so try 1 instead of 2 (it removes divisions by depth in the formula used in the diffusion).

Note also that the keyword TYPE OF ADVECTION should have 3 integers,

TYPE OF ADVECTION : 1;5 is equivalent to:

TYPE OF ADVECTION : 1;5;1

because of the default values, and it means a treatment of the tracers advection with the method of characteristics (which is not mass conservative).

Another thing is that you use the primitive shallow water equations, the wave equation is faster, and there is nothing for tidal flats so you rely on default values.

Try the following parameters :

TYPE OF ADVECTION : 1;5;14
SUPG OPTION : 0;0
COMPATIBILITY OF FREE SURFACE GRADIENT : 0.9 (or down to 0. if unstable)
TREATMENT OF THE LINEAR SYSTEM : 2 SOLVER : 1
TREATMENT OF NEGATIVE DEPTHS : 2
CONTINUITY CORRECTION : YES

it should go faster, with strictly positive values of depth, and you should find monotonicity of the tracers.

With best regards,

Jean-Michel Hervouet

With COEFFICIENT FOR DIFFUSION OF TRACERS = 0,

the area with negative concentration seems larger than before(COEFFICIENT FOR DIFFUSION remains default value).

Dear jmhervouet,

Thank you for your help last time.

But I didn't change my case file totally according to your advice, because your reply reminds me that tidal flats are not taken into account at all.

So I set tidal flats = no and make some changes. The results seemed more reasonable than before.

Recently we enlarged the study area and take tidal flats into account.

New problems occured.

If i set tompson condition=no, then ill-posed problem of free velocity came out.

if i set tompson condition=yes, then interger overflow problem happened.

Do you have any idea about them?

please see my case file in next post.

Best wishes

/
/

---

/ COMPUTER INFORMATIONS
/

---

/
GEOMETRY FILE = lvsi-2.geo
BOUNDARY CONDITIONS FILE = lvsi-2.bc
FORTRAN FILE = user.f
RESULTS FILE = lvsi-large.res
LIQUID BOUNDARIES FILE = bclevel.dat
/VALIDATION = YES
/REFERENCE FILE = ref
PARALLEL PROCESSORS = 0
/
/

---

/ GENERAL INFORMATIONS - OUTPUTS
/

---

/
VARIABLES FOR GRAPHIC PRINTOUTS = 'U,V,S,B,H'
GRAPHIC PRINTOUT PERIOD = 60
LISTING PRINTOUT PERIOD = 30
TIME STEP = 60.
NUMBER OF TIME STEPS = 1800
MASS-BALANCE = YES
INFORMATION ABOUT SOLVER = YES
/
/

---

/ INITIAL CONDITIONS
/

---

/
COMPUTATION CONTINUED = NO
INITIAL CONDITIONS = 'CONSTANT ELEVATION'
INITIAL ELEVATION = 3.82
/
/

---

/ BOUNDARY CONDITIONS
/

---

/
PRESCRIBED ELEVATIONS =0.0
OPTION FOR LIQUID BOUNDARIES =2
/
/

---

/ PHYSICAL PARAMETERS
/

---

/
LAW OF BOTTOM FRICTION = 3
FRICTION COEFFICIENT = 32.
TURBULENCE MODEL = 1
VELOCITY DIFFUSIVITY = 1.E-4
/
/

---

/ NUMERICAL PARAMETERS
/

---

/
TYPE OF ADVECTION = 1;5
/SUPG OPTION = 1;2
SOLVER ACCURACY = 1.E-3
DISCRETIZATIONS IN SPACE = 12;11
PROPAGATION = YES
SOLVER = 3
PRECONDITIONING = 2
IMPLICITATION FOR DEPTH = 0.6
IMPLICITATION FOR VELOCITY = 0.6
BOTTOM SMOOTHINGS = 1
MAXIMUM NUMBER OF ITERATIONS FOR SOLVER = 300
/TYPE OF ADVECTION : 1;5;14
SUPG OPTION : 0;0
FREE SURFACE GRADIENT COMPATIBILITY : 0.9
TREATMENT OF THE LINEAR SYSTEM : 2
/SOLVER : 1
TREATMENT OF NEGATIVE DEPTHS : 2
CONTINUITY CORRECTION : YES
MASS-LUMPING ON H=1
/TIDAL FLATS=NO
/

Hello,

At least you should put now :

DISCRETIZATIONS IN SPACE : 11;11

However I have never seen an integer overflow, this is strange. You will have to send your case if it is still the case.

With best regards,

Jean-Michel Hervouet

done:)

But I don't know why

12;11 is quasi-bubble element for velocities (a way to avoid spurious oscillations due to inf-sup condition). Now the keyword "COMPATIBILITY OF FREE SURFACE GRADIENT = 0.9" has the same role and is much cheaper.

JMH)