~reducedmodelling/fluidity/ReducedModel : changes

~reducedmodelling/fluidity/ReducedModel » Changes from revision 3621

From Revision 3027 to 3008

Rev	Summary	Authors	Date
3027	Fix periodise.F90 so that its output passes diamond_validati... Fix periodise.F90 so that its output passes diamond_validation	pef	13 years ago
3026	update prescribed fields in the adjoint solve update prescribed fields in the adjoint solve	sf1409	13 years ago
3025	Fix some flmls that are giving diamond_validation errors Fix some flmls that are giving diamond_validation errors	pef	13 years ago
3024	There are currently no short or medium test with an implicit... There are currently no short or medium test with an implicit solid and adaptivity. So add a medium test: flow around a vertical cylinder with 3d adaptivity and without free-surface. I tested locally that the current version of the code gives the same result as a binary from last November. So this should work. Let's hope Buildbot won't complain too much...	avire	13 years ago
3023	If a user derives a P0 mesh but the mesh_continuity is not '... If a user derives a P0 mesh but the mesh_continuity is not 'Discontinuous', they now get a nice FLAbort message, rather than: * glibc detected * ../../bin/fluidity: free(): invalid next size (normal): 0x0000000002ebc900 ***	ctj10	13 years ago
3022	- Merging branch revision r15543 into trunk. Branch now sync... - Merging branch revision r15543 into trunk. Branch now synchronised with trunk. Summary of branch changes: ------------------------------------------------------------------------ r15543 \| ctj10 \| 2011-02-16 18:04:48 +0000 (Wed, 16 Feb 2011) \| 7 lines - Brought the calculate_diagnostic_phase_volume_fraction call out of calculate_momentum_diagnostics as a temporary bug fix. Only the submaterials were being used to compute the diagnostic PhaseVolumeFraction, leading to incorrect results. - Also calling calculate_diagnostic_phase_volume_fraction in Diagnostic_fields_wrapper.F90. - Added a 1D version of the mphase_gravity test case committed in r15499 (also tagged as 'special' for now).	ctj10	13 years ago
3021	- Merging branch revision r15528 into trunk. - Merging branch revision r15528 into trunk. Summary of branch changes: ------------------------------------------------------------------------ r15528 \| ctj10 \| 2011-02-16 02:45:54 +0000 (Wed, 16 Feb 2011) \| 4 lines - Added a new problem_type called 'multiphase' to the schema. - Added a basic options checking subroutine called 'check_multiphase_options' to Populate_State.F90.	ctj10	13 years ago
3020	- Merging branch revision r15499 into trunk. - Merging branch revision r15499 into trunk. Summary of branch changes: ------------------------------------------------------------------------ r15499 \| ctj10 \| 2011-02-12 03:46:32 +0000 (Sat, 12 Feb 2011) \| 5 lines - Adding a new multi-phase test case called mphase_gravity. Tagging this as 'special' for now since nothing works. The test case considers a square 2D domain filled with two incompressible fluids starting with a zero velocity ﬁeld. One liquid is denser than the other (\rho_1 = 1.0, \rho_2 = 1.5) and the volume fraction of each phase is initially set to 0.5 everywhere. The heavier liquid should sink beneath the lighter one as time progresses due to the density difference, causing the volume fraction of the heavier liquid to tend to 1.0 and 0.0 in the lower and upper half of the box respectively.	ctj10	13 years ago
3019	- Merging branch revisions r15410, r15498 and r15527 into tr... - Merging branch revisions r15410, r15498 and r15527 into trunk. Summary of branch changes: ------------------------------------------------------------------------ r15410 \| ctj10 \| 2011-02-03 19:02:15 +0000 (Thu, 03 Feb 2011) \| 3 lines - Multiplying mass element, viscosity element, buoyancy element, and C^T contributions by PhaseVolumeFraction (if multi-phase). ------------------------------------------------------------------------ r15498 \| ctj10 \| 2011-02-11 20:18:32 +0000 (Fri, 11 Feb 2011) \| 16 lines In Momentum_CG.F90: ------------------- - Using a new variable called 'coefficient_detwei' for femtools function (shape_shape, etc) arguments, in case detwei also has to be multiplied by the non-linear volume fraction (nvfrac_gi) among other things. This avoids the need for duplicate function calls with a different detwei argument for multi-phase simulations. - Multiplying advection term through by the non-linear volume fraction. We're only considering the non-conservative form (i.e. beta=0) for now - we will sort the rest out later. - In mass term: removed multiplications by the volume fraction in the moving mesh stuff. - Removed multiplications by the volume fraction in the viscosity term for now. In Multiphase.F90: ------------------ - Setting the intent of 'nvfrac' in get_nonlinear_volume_fraction to intent(inout). ------------------------------------------------------------------------ r15527 \| ctj10 \| 2011-02-16 00:46:40 +0000 (Wed, 16 Feb 2011) \| 7 lines - Subroutine 'get_phase_submaterials' now keeps track of the phase's index within the submaterials array. - Calls to calculate_momentum_diagnostics and calculate_densities now use the submaterials array for multi-phase flow simulations. - Added a PhaseVolumeFraction field to each phase in the mphase_identical_velocities test case and removed the unnecessary MaterialVolumeFraction fields. - Added dependencies on Multiphase.F90 in assemble/Makefile.in. - Set up an array of mass matrices in Momentum_Equation.F90	ctj10	13 years ago
3018	Tiny changes when registering the diagnostics and checking t... Tiny changes when registering the diagnostics and checking the options in Implicit_Solids. This should work with 3d-adaptivity as tested by Frank's short and medium tests. A 3d implicit-solid with a free-surface and a hr-adaptive mesh would currently fail. This is not new. The same set-up run with a stable version of the code from last November also fails when adapting the mesh. The reason is that implicit_solids seems to fail with 2+1 adaptivity, which is required when runnning hr-adaptivity with a free-surface. This will be mantis and fixed in the future.	avire	13 years ago
3017	A first draft of recording whatever the functional might nee... A first draft of recording whatever the functional might need later when you actually have it	pef	13 years ago
3016	Dare to tighten the tolerance a bit... Dare to tighten the tolerance a bit...	cjc1	13 years ago
3015	This test now checks against a Python implementation of the ... This test now checks against a Python implementation of the RK trajectory, validating the Fluidity implementation.	cjc1	13 years ago
3014	This test case now tests the new Runge-Kutta code, in serial... This test case now tests the new Runge-Kutta code, in serial.	cjc1	13 years ago
3013	Disable testcase for now. Disable testcase for now.	cjc1	13 years ago
3012	The long awaited restoration of 15599. For now, I've made The long awaited restoration of 15599. For now, I've made lagrangian_timestepping an optional option underneath detectors. The problem is that it is possible to have a combination of static and Lagrangian detectors, but the option shouldn't sit under lagrangian_detector as then it would be possible to specify different detector timestepping options for different detectors, which is not compatible with how things are coded. Comments welcome on how to do this in the schema. I haven't done a lot of testing yet but tests/lagrangian_detectors/lagrangian_detectors_rk.flml shows how to use Lagrangian detectors with the new implementation. TODO: * Add the RK bits to the serialisation for data transfer. That bit of code is a total mess so that will take some time. * Make a separate serial test case comparing against a pure Python implementation using a linear velocity field. * Add a subcycling option so that some integer number of detector timesteps can be taken per fluids timestep.	cjc1	13 years ago
3011	All short tests should pass now. Adaptivity might still be p... All short tests should pass now. Adaptivity might still be problematic for implicit_solids medium and long tests.	avire	13 years ago
3010	fix the supplied python timestep for the adjoint fix the supplied python timestep for the adjoint	sf1409	13 years ago
3009	... one of the test should actually be a medium one ... one of the test should actually be a medium one	ffm10	13 years ago
3008	adding two short tests for the 1-way coupling instead, seria... adding two short tests for the 1-way coupling instead, serial and parallel	ffm10	13 years ago