TELEMAC-MASCARET Forum: Treatment of boundary for tracers at inlet (1/1)

Welcome, Guest

TOPIC: Treatment of boundary for tracers at inlet

Treatment of boundary for tracers at inlet 5 years 5 months ago #33604

qilong
OFFLINE
Expert Boarder
Posts: 340
Thank you received: 33

Hello everyone,

I have a question about the treatment of lateral boundary for tracers (also for k-epsilon). I found in the test with a tidal current imposed at downstream boundary, the boundary value of tracer uses the value from previous time step when this open boundary has inflow, so the tracer concentration does not change in this period. When the tidal current reverses and this boundary has outflow, the tracer values are changing again (can be seen here).

In the subroutine cvdf3d.f, there is a section for dealing with lateral boundary conditions for tracers:

!     FOR TRACERS (=NOT VELOCITY) : DIRICHLET VALUES ARE NOT RESPECTED IF EXIT
!     THERE IS NO NEED TO TEST KENTU OR KADH FOR TRACERS
!
      IF(NPTFR3.GT.0.AND.NFRLIQ.GT.0.AND..NOT.VELOCITY) THEN
        DO IPTFR=1,NPTFR3
          IF(LIFBOL%I(IPTFR).EQ.KENT) THEN
!           EXITS ARE TREATED AS FREE BOUNDARIES
            IP=NBOR3%I(IPTFR)
            IF(FLUEXTPAR%R(IP).GT.0.D0) LIFBOL%I(IPTFR)=KSORT
          ELSEIF(LIFBOL%I(IPTFR).EQ.KSORT) THEN
            IP=NBOR3%I(IPTFR)
            IF(FLUEXTPAR%R(IP).LT.0.D0) THEN
              LIFBOL%I(IPTFR)=KENT
              FBORL%R(IPTFR)=FN%R(IP)
            ENDIF
          ENDIF
        ENDDO
      ENDIF

I guess this is the code that is responsible for the strange behaviour happening for the tracer at the open boundary.

It seems that this could add some artifacts to the solution if the boundary is not far away from the domain? Is there anyway to improve it?

Regards,
Qilong

The administrator has disabled public write access.

Treatment of boundary for tracers at inlet 5 years 5 months ago #33605

qilong
OFFLINE
Expert Boarder
Posts: 340
Thank you received: 33

The imposed tracer value (not exact) is only taken during the inflow, but it's not taken when this boundary becomes outflow. Is this morel realistic than using the same value for both inflow and outflow?

And also this causes the k-epsilon values at the lateral boundary not continuously evolving.

Here we show k values at 3 nodes, node 38 is the middle node, the other 2 are at two sides with prescribed water level. We thought k-epsilon had the neumann type lateral boundary by default. But it changes the boundary type depending on the inflow or outflow as well.

The administrator has disabled public write access.

Treatment of boundary for tracers at inlet 5 years 5 months ago #33615

pham OFFLINE Administrator Posts: 1559 Thank you received: 602	Hello, What do you mean by: "I found in the test with a tidal current imposed at downstream boundary"? Is it the vasque example of T3D or another one? Have you changed it and if yes, what are the exact changes? Which version of T3D do you use? Can you give your steering file, fortran files and boundary conditions file? Chi-Tuan
	The administrator has disabled public write access.

Treatment of boundary for tracers at inlet 5 years 5 months ago #33622

qilong
OFFLINE
Expert Boarder
Posts: 340
Thank you received: 33

Hello, Thanks for asking.

The test case is not the example in TELEMAC-3D. These are the tests we made for testing different functionalities.

We were using v7p3r1 for these tests. One of the example is with a large-scale domain and the other one is a much smaller model for representing 1DV water column. Both of them indicate the same behaviours of tracers at tidal boundaries.

Here you can find the small test case, which is fast to test, and its results of k-epsilon was shown in the second figure in the first post.

Thanks!
Qilong

Attachments: