From SPHYSICS

Revision as of 09:46, 5 October 2010

04 October 2010 - v2.2.000:

For version 2.2, there are numerous bug features, but most importantly the addition of new complex geometry generator. Details can be found in the new Blender_for_SPHysics_v2.2_GUIDE.pdf Multiple bug fixes including:

• SPHYSICSgen_3D.f bug for h=sqrt(dx*dx + dy*dy + dz*dz)
• SPHYSICSgen_3D.f bug for fill_part for floating objects (ytmin & ytmax)
• Multiple moving objects interacting with object number 1
• File DT being unfinished until end of simulation (now opened and closed consistently)
• Floating_Objects.RESTART file added for restarting simulation
• Intel Visual Fortran replaces Compaq Visual Fortran (CVF)

3 February 2010 - v2.0.001: For version 2.0, see details below of new code.

• Corrected bug movingPaddle_2D/3D.f for single frequency piston paddle
• Corrected bug in SPHYSICSgen_3D.f for ifort for Cases 6 & 7.

28 January 2010 - v2.0.000:

For version 2.0, there are many new features including kernel correction schemes, floating objects, Riemann solvers. Details can be found in the new SPHysics_v2.0.000_GUIDE.pdf

Features:

• kernel gradient correction
• floating objects
• Riemann solvers
• New cases added:
  • Case6 - floating boxes in waves (2-D & 3-D)
  • Case7 - Focused wave group approaching a trapezoidal coastal defence structure (2-D & 3-D)
  • Case8 - Dam-break with 2 falling boxes with X-periodic boundaries (2-D)

Source code bugs:

• Have separate subroutines : poute_2D & poute_Conservative_2D
• INDAT always in columns (better to read files from Matlab)
• Normalization coefficients for Wendland 3D: AWen = 21/(16.pi) and BWen = AWen*-5
• Shepard & MLS filters now appear with ac_Shepard / pre_celij_Shepard / pre_self_Shepard
• Correct TIME for each PART
• ipoute is used instead of ivort_calc
• use vortx/y/z_temp in poute instead of vortx(npar)
• Variables xm1 in Verlet were obsolete and removed
• Some variables xdoto, xdotm1 in Beeman were obsolete and removed
• Vorticity calculation as an option
• Velocity gradients, CSPH, in separate routines
• if(i.ge. nstart_ MLS.and.j.ge. nstart_ MLS)then before drx and call kernel in pre_MLS
• nstart_Shepard = nbp1  !nstart and nstart_MLS = nbp1  !nstart
• if(rr2.lt.fourh2.and.rr2.gt.1.e-18) then instead of if(rr2.le.fourh2.and.rr2.ge.1.e-18) then
• Checkpointing (i.e. using RESTART repetitively for batch submissions on clusters)
• nstart changed
• Bug found by Arno Meyerhofer in LU_decomposition
• message that Verlet and Beeman do not work with conservative formulation
• Error Message for periodic boundaries cases with too few particles
• Hughes and Graham (2010) suggestion to improve Dalrymple BC behaviour
• Number of time steps to apply density filter, verlet or DBC correction as input data
• AC replaced by AC_NONE
• Floating bodies for Dalrymple BC
• Paddle motion can be read from a file, e.g. Case7 for Focused wave group
• Multiple trapezoids can be placed on a beach

COMPILERS:

• Support has now been added for the Silverfrost FTN95 compiler on Windows, and for the GCC gfortran compiler on Cygwin (linux file system emulator)

BATCH files:

• Batch files for unix/linux now report errors in compilation and successful completion of SPHYSICSgen
• New batch files for Silverfrost FTN95 (formerly Salford FTN95)

CaseN.txt:

• Multiple Changes due to new compiling options.
• Option for precision of XYZ variables: single or double

Postprocessing:

• Matlab program to plot properties by colour (e.g. SPHYSICS2D_Plot_Magnitude.m)
• Debugged PART2VTU_2D/3D.f

11 February 2009 - v1.4.000:

Source code:

• Optimisations for gfortran and ifort compilers (-ipo)(SPHysicsgen.f)
• CFL_number is input in CaseN.txt to calculate dt (SPHysicsgen.f; getdata.f; variable_time_step.f)
• dt corrected for Predictor-corrector algorithm (step_predictor_corrector.f)
• Equation of state for an ideal gas (SPHysicsgen.f ; EoS_IdealGas.f)
• Changes in common (obsolete variables removed) (common.2D/3D)
• The Beeman time-stepping algorithm is implemented (SPHysicsgen.f ; step_beeman.f)
• The parameter KIND has been removed from subroutines ac, celij, self, kernel, viscosity.

BATCH files:

• sph.out: output data from SPHYSICS is created in the code and some information appears on the screen, too.
• Now SPHYSICSgen is automatically compiled using SPHYSICSgen_compiler.mak files

CaseN.txt:

• Case3_3D: Case3.txt.highRes or Case3.txt.lowRes
• Changes due to new compiling options.

Postprocessing:

• Paraview programs in Fortran. New file in Postprocessing/fortranRoutines: PART2VTU_2D/3D.f

22 April 2008 - v1.2.000 :

• New version with Faster linked list and routines
  • (changes to divide_2D/3D.f, ini_divide_2D/3D.f, keep_list_2D/3D.f, recover_list_2D/3D.f, celij_BC_#choice#_2D/3D.f, self_BC_#choice#_2D/3D.f)
• Moving Least Squares (MLS) Density Reinitialization added
• RESTART option for stopping and restarting each simulation
• New option for different Equations of State
• Guide updated

12 November 2007 - v1.0.003 :

• Corrected undefined tangents in updateNormals_2D.f
• Corrected bugs and updated plotting routines, e.g. SPHYSICSgen2D/3D_Plot.m
• Corrected bugs in subroutines fill_part & wavemaker for iBC=2 in SPHYSICSgen2D/3D.f
• Changed parameters in some CaseN.txt files to obtain improved results

08 August 2007 - v1.0.002 :

• Corrected permissions on linux .bat files
• Edited README files for compiler choice

06 August 2007 - v1.0.001 :

• Corrected directory SPHYSICS_2D to SPHYSICS2D CaseN_cvf.bat (same for 3D)
• Corrected ..\Release to ..\..\execs in subroutine tocompile_cvf in SPHYSICSgen_2D.f & SPHYSICSgen_3D.f

01 August 2007 - v1.0 :

• CODE RELEASED!