open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I have modified the code with Jean-Michel Hervouet's suggestions, and it worked well. Thanks a lot!

I now have a last question. If I want to include a loop for comparing the structure's diameter and the water depth HN%R(I), should I do it for I=1,BIEF_NBPTS(11,MESH)? I guess I should use HN and not H (not computed yet, right?).

Any comments regarding my previous post about orifice laws used in tubes?

Thanks in advance,
PL

Hello,

Yes, that's it, HN%R(I), but BIEF_NPTS(11,MESH) is a bit heavy, you can use NPOIN if locally available, or HN%DIM1.

Regards,

JMH

Hi,
I have a related question to pilou1253 for when applying multiple drag zones in DRAGFO.
First I considered a single drag zone. The drag is defined as a bed drag parameterisation of the form:

F= rho * CD * U**2 * A (Eq 1)

Where CD is the drag coefficient, rho is water density and A is the drag zone area (=AIRE). In DRAGFO I express FUDRAG and FVDRAG as shears by dividing Eq 1 through by A (and rho & depth as standard), as shown below:

DO I=1,BIEF_NBPTS(11,MESH)
IF(INPOLY(X(I),Y(I),XSOM,YSOM,NSOM)) THEN

depth=max(H%R((I)), 0.01)
UNORM = MAX((SQRT(UN%R(I)**2+VN%R(I)**2)),1.D-4)
FUDRAG%R(I) = -(CD*UNORM)/depth
FVDRAG%R(I) = FUDRAG%R(I)

ENDIF
ENDDO

In doing this I remove the need to calculate AIRE and SOM at all in DRAGFO. I ran this code and typical velocities within the drag zone reduced to 1.9 m/s. However, if I leave the code that calculates AIRE and SOM in my DRAGFO, the values of velocity inside the drag zone reduce to around 1.5 m/s. I do not understand why this is given that FUDRAG and FVDRAG do not rely on AIRE or SOM?

On inspection it appears that the code that calculates AIRE for Quasi Bubble discretization (shown below) is what is affecting this (I use quasi bubble discretization), but I don't understand why. So my second question is, what does the quasi bubble code do to FUDRAG and FVDRAG so that velocities within the drag zone reduce more compared to results where calculations for AIRE and SOM are removed completely?

IF(FU%ELM.EQ.12) THEN
!
DISCLIN=11
CALL CHGDIS(FUDRAG,DISCLIN,12,MESH)
CALL CHGDIS(FVDRAG,DISCLIN,12,MESH)
!
DO IELEM = 1 , NELEM
I4=IKLE%I(IELEM+3*NELMAX)
X4=(X(IKLE%I(IELEM ))+ &
X(IKLE%I(IELEM+ NELMAX))+ &
X(IKLE%I(IELEM+2*NELMAX)))/3.D0
Y4=(Y(IKLE%I(IELEM ))+ &
Y(IKLE%I(IELEM+ NELMAX))+ &
Y(IKLE%I(IELEM+2*NELMAX)))/3.D0
IF(INPOLY(X4,Y4,XSOM,YSOM,NSOM)) AIRE = AIRE + T1%R(I4)

ENDDO
!

ENDIF

Another observation is that if FUDRAG and FVDRAG are calculated after the Quasi Bubble calculation, the velocities remain higher at 1.9m/s. However in the validation dragforce case the quasi bubble code comes after the FUDRAG calculation, so which one is correct, given that in JMH’s earlier reply he suggests calculating AIRE before FUDRAG?
I have attached the 2 DRAGO files, the first does not include calculation for AIRE and gives typical velocities within the drag zone of 1.9 m/s. The second includes calculation for AIRE and gives typical velocities of 1.5 m/s. I am not running in parallel.

Any help would be greatly appreciated.
Cheers
Danny

Hi Danny,
my understanding but I may be wrong of this dragforce subroutine for vertical structures such as the resistive effect of a number of vertical piles or perhaps tree trunks in a section of floodplain area is to calculate the drag force for such structures whereby in the example given N is the number of these structures and D the typical diameter or width of an individual pile and applies this drag or resistive force over a defined area (using a user input polynomial) in the form of a stress similar to and in addition to the bed roughness. Your proposed formulation uses the entire polynomial planar area divided by water depth as basically the vertical width of the structures.

also your formulation leaves out the multiplier of 0.5 from the drag force equation.

regards
Tony

Hi,

Thanks for your reply, however my problem is not related to the form of the drag equation that I define in DRAGFO. I am asking why it matters what order you have the calculation for drag and AIRE in DRAGFO when using Quasi-bubble(I encounter the same problem if I use the standarg drag equation that you mention).

To be clearer, I have run 2 cases with the same geometry and boundary conditions. In the first case, my dragfo file (Dragfo_no_aire.f90) includes only the code for calculating drag. In teh second (Dragfo_with_aire.f90) I also include the code for calculating aire. My question is, why am I getting different results out when my equation for drag is independant of aire?

Cheers
Danny

The files do not seem to attach in .f90 format, here they are as text files.

Hi Danny
I ran both of your files for my own test case and found no difference in results between the two fortran files ??

regards
Tony

Hi Tony,

Thanks for looking at this, can you send me your run files so I can have a look? Also did you use quasi-bubble discrtitization?

Cheers

Danny

Yes i used the 12, 12 and 12, 11 (both same result) and the test case was from the examples telemac2d for drag force
I am away for the remainder of the day but you should be able to generate the same test case

Hello,

I do not see any problem in your implementation except the lines:

open (unit=101,file="U.txt")
write(101,*) UNORM


I do not know what can be the result of this since you open the file in a loop, so it is always opened and never closed. It should be opened before the loop and closed after.

The quasi bubble implementation is correct because for the areas around points we take the quasi-bubble areas, so AIRE would be wrong if we only take nodes at the apex of triangles and not the nodes in the middle of triangles.

However the result may be different between a linear velocity and a quasi-bubble velocity, even the area AIRE could be slightly different.

With best regards,

Jean-Michel Hervouet