Back to Start Page Introduction Motivation Contents Current State Outlook New Videos Featflow Home Page Contact



Computational Details:

"Flow over a Backward Facing Step at medium Reynolds number" Back to >>Flow over a Backward Facing Step at medium Reynolds number<<


Description of the flow problem
Description of the spatial discretization
Description of the temporal discretization
Computer requirements
Mathematical details




Description of the flow problem

  • nonstationary 2D channel flow over a "backward facing step"
  • channel width: before/after step/total: 3/10/13
  • channel height: before/after step/total: 0.5/1/1
  • inflow b.c.'s: parabolic with maximum velocity 1
  • outflow b.c.'s: natural "do nothing" b.c.'s
  • other b.c.'s: zero velocity at fixed walls
  • initial condition at t=0: starting from rest
  • viscosity parameter: 1/nu=1,000




Description of the spatial discretization

  • coarse mesh (=level 1): 50 cells, 69 vertices, 286 d.o.f.`s



  • uniform refinements with exact boundary adaption
  • visualization on level 5: 12,800 cells, 13,089 vertices, 64,576 d.o.f.`s
  • computational mesh on level 7: 204,800 cells, 205,953 vertices, 1,026,304 d.o.f.`s
  • nonconforming nonparametric rotated bilinear fem's (meanvalue version), UPW




Description of the temporal discretization

  • equidistant time stepping for computation with k=0.06666667
  • equidistant time stepping for visualization with k=0.20 (= 1 frame)
  • Total time T=100 corresponds to 1,500 time steps
  • fractional step theta scheme




Computer requirements

  • date: 05/28/97
  • simulation by: S.Turek
  • visualization by: J.Wallis
  • SGI PowerIndigo2/R10: 176 MB, 98,026 seconds
  • AVS data: 1,080 MB
  • Software: FEATFLOW1.0 + BOUSS




Mathematical details

  • For more details about numerical and algorithmic aspects see the `Mathematical Background' in the FEATFLOW manual or visit our paper archive for much more details.
  • The problem-specific data for the applied software version including parameter files and input data can be downloaded here!




Please send any comments and suggestions to: featflow@featflow.de