Opened 13 years ago

Closed 11 years ago

#195 closed defect (fixed)

Convergence problem

Reported by: ole Owned by: stephen
Priority: high Milestone:
Component: Functionality and features Version:
Severity: major Keywords:
Cc:

Description

A pure sine wave gets attenuated when propagated over a long distance. Will Power identified this problem at the August 2007 workshop at GA (see https://datamining.anu.edu.au/anuga/wiki/FirstAnugaMeeting). A code that will reveal this fact is available in https://datamining.anu.edu.au/anuga/browser/anuga_validation/convergence_study

Attached graphs show that the sine wave gets attenuated unless mesh is very refined.

Initially, Stephen suggested that this may be a timestepping issue. However, refining timesteps or trying high order timestepping (see changeset:4721 and changeset:4722) did not change the attenuation.

The current plan is for Stephen to look at the 2nd order gradient limiters. If they are essentially behaving as first order limiters, this could explain what is now believed to be diffusive effects.

Attachments (1)

time_120.png (20.9 KB) - added by ole 13 years ago.
Graph that shows the attenuation of a pure sine wave

Download all attachments as: .zip

Change History (3)

Changed 13 years ago by ole

Graph that shows the attenuation of a pure sine wave

comment:1 Changed 12 years ago by ole

Following Stephen's new implementation of Edge limiters and associated beta values, this feature needs to be integrated into ANUGA and tested against validation examples like ANUGA and large problems like the Onslow model.

On thought would be to create instances of classes containing options that go together such as tigh_slope_limiters, use_edge_limiters etc etc

See also ticket:262 for one of these options.

comment:2 Changed 11 years ago by nariman

  • Resolution set to fixed
  • Status changed from new to closed
Note: See TracTickets for help on using tickets.