Professional Documents
Culture Documents
INTRODUCTION 9
SOFTWARE CAPABILITIES 9
SOFTWARE ORGANIZATION 11
USER MANUAL PRESENTATION 13
WHAT'S NEW 14
VERSION 2009.1 14
VERSION 2009.0 17
VERSION 2008.0 24
VERSION 2007.0 30
VERSION 2006.1 34
VERSION 2006.0 38
VERSION 2005.0 45
VERSION 2004.1 50
VERSION 2004.0 55
IDENTIFIED BUGS, PROBLEMS AND LIMITATIONS 59
SOFTWARE MANAGER 62
FILE MANAGER 62
MODULE CALLS 66
ADVANCED MODULE CALLS 74
RUN LIST 77
SOFTWARE CONFIGURATION 80
CUSTOMIZED INSTALLATION 84
FLEXLM 85
GETTING STARTED 89
SOFTWARE LAUNCH 89
PROBLEM SET-UP 91
CALCULATION 99
RESULTS DISPLAY 101
PRE-PROCESSING 104
INTRODUCTION 104
GEOMETRY IMPORT 108
THERMAL 110
GEOMETRY ASSIGNMENTS 111
MATERIALS ASSIGNMENT 111
INTERFACES ASSIGNMENT 116
BOUNDARY CONDITIONS ASSIGNMENT 122
PROCESS CONDITIONS ASSIGNMENT 127
INITIAL CONDITIONS ASSIGNMENT 128
RUN PARAMETERS ASSIGNMENT 129
FLUID FLOW & FILLING 130
RADIATION 134
STRESS 149
DATABASES 158
MATERIAL DATABASE 158
MATERIAL PROPERTIES 162
THERMODYNAMIC DATABASES 169
INTERFACE DATABASE 198
BOUNDARY CONDITIONS DATABASE 201
PROCESS DATABASE 218
STRESS DATABASE 224
STRESS MODELS AND PROPERTIES 227
EXTRACT AND MAPPING 269
RESULTS EXTRACTION 270
RESULTS MAPPING 273
DOMAIN REMOVAL 275
DOMAIN ADDITION 277
EXTRACT EXAMPLES 281
RUN PARAMETERS 302
GENERAL RUN PARAMETERS 304
THERMAL RUN PARAMETERS 311
CYCLING RUN PARAMETERS 317
RADIATION RUN PARAMETERS 320
FLOW RUN PARAMETERS 324
TURBULENCE RUN PARAMETERS 337
STRESS RUN PARAMETERS 339
MICRO RUN PARAMETERS 344
PRE-DEFINED RUN PARAMETERS 348
RUN PARAMETERS RECOMMENDATIONS 350
POROSITY MODELS 353
POROS=1 354
POROS=4 358
POROS=8 358
DENSITY DEFINITION 359
ACTIVE FEEDING 360
SOLVER 446
TROUBLESHOOTING 449
INTRODUCTION 453
FIELD SELECTION 459
DISPLAY TYPES 463
DISPLAY PARAMETERS 473
TAPE PLAYER 491
CURVES 492
GEOMETRY MANIPULATION 498
INTRODUCTION 530
GEOMETRY 531
RADIATION FACES 532
TEMPERATURE 533
HEAT FLUX 536
DISPLACEMENTS 537
STRESS 538
ABAQUS IMPORT 542
SYSWELD IMPORT 545
INTRODUCTION 547
HOW DOES PROCAST PARALLEL WORKS ? 550
USE OF THE PARALLEL SOLVER 555
REPEATABILITY 561
LIMITATIONS 562
MACHINE CONFIGURATION 563
HARDWARE AND OS 565
INTRODUCTION 568
ADVANCED POROSITY PRE-PROCESSING 571
MATERIAL PROPERTIES 572
GAS AND BUBBLE PROPERTIES 577
PROCESS INFORMATION 581
CALCULATION SETTINGS 589
EXAMPLES OF PREFIX_PORO.D INPUT FILE 596
APM EXAMPLES 600
ADVANCED POROSITY SOLVER 611
ADVANCED POROSITY POST-PROCESSING 614
APM APPENDIX 619
APM FLOW CHART 619
GROWTH MODEL OF A PORE 620
GAS THERMODYNAMIC DATABASE 623
LIST OF THE APM PARAMETERS 625
APM REFERENCES 628
MICROSTRUCTURES 629
INTRODUCTION 630
CASE SET-UP AND RESULTS 633
EXAMPLES 639
IRON AND STEEL 646
INTRODUCTION TO IRON AND STEEL 646
IRON AND STEEL MODELS 653
CASE STUDIES 659
"2-D" 698
INTRODUCTION 699
MESHING 701
CASE SETTING 708
CAFE-3D 711
INTRODUCTION 718
MODEL SET-UP 720
INVERSE RUN 727
FILE FORMATS 731
INVERSE APPENDIX 735
PRINCIPLES 746
PRE-PROCESSING 752
STEADY STATE CASES 752
NON STEADY STATE CASES (MILE) 768
INTRODUCTION 794
THE RDG HOT TEARING CRITERION 795
STRUCTURE OF THE INPUT FILES 799
RUNNING A CASE 803
VIEWING OF THE RESULTS 806
REFERENCES 811
CENTRIFUGAL-1 862
RESULTS COMPARISONS (2008.0) 865
GRAVITY-1 866
GRAVITY-2 868
GRAVITY-3 870
HPDC-1 872
LPDC-1 874
INVESTMENT-1 877
STRESS-1 879
MICROSTRUCTURE-1 883
THIXO-1 886
CENTRIFUGAL-1 888
RESULTS COMPARISONS (2007.0) 891
GRAVITY-1 892
GRAVITY-2 894
GRAVITY-3 896
HPDC-1 898
LPDC-1 901
INVESTMENT-1 904
STRESS-1 906
MICROSTRUCTURE-1 910
THIXO-1 913
CENTRIFUGAL-1 915
TUTORIALS 917
INTRODUCTION
SOFTWARE CAPABILITIES
ProCAST is a software using the Finite Elements Method (FEM). It allows the
modeling of Thermal heat transfer (Heat flow), including Radiation with view
factors, Fluid flow, including mold filling, Stresses fully coupled with the thermal
solution (Thermomechanics). Beside that, it includes also microstructure
modeling and porosity modeling.
Special models are included in order to account for thixo casting and lost foam.
Specific features are included to account for processes such as high pressure die
casting, centrifugal, tilt. Finally, customized models for foundry processes, such
filters, sleeves are included.
Thermal calculation
Thermomechanical calculation
SOFTWARE ORGANIZATION
The software is organized around a Manager, which calls the different modules :
The following figure is presenting the structure of the software. First, the
geometry, in the form of a CAD model is loaded into MeshCAST, to generate a
FEM mesh. Then, the calculation is configured in PreCAST, the Pre-processor.
PreCAST is linked to Thermodynamic Databases for the automatic
determination of the material properties from thermodynamic databases. Before
the solver ProCAST is launched, a "data conditioner" named DataCAST is run.
Finally, the results can be viewed or exported (for further processing) in the
Post-processor ViewCAST.
The ProCAST solvers are divided in "Physical modules" with the following
capabilities :
Thermal module
• Heat conduction (Fourier equation)
• Latent heat release during solidification
• Cycling in die casting
• Sleeves (insulating and exothermic)
• Non-coincident meshes
• Solidification time
• Secondary Dendrite Arm spacing
• Porosity indicator
Radiation module
• Net radiation method
• Full view factors capabilities
• Mirror and rotational symmetries
• Relative motion of materials
• Solid or surface enclosures
Stress module
• Elastic, Elastic-plastic, Elasto-visco-plastic
• Rigid or vacant materials
• Automatic calculation of the air gap heat transfer
• Contact algorithm between the different materials
• Contact pressure
• Die Fatigue prediction
• Hot Tearing indicator
The features linked to specific processes (e.g. cycling, tilt, ...) are embedded in the
corresponding physical modules.
After a "Getting started" chapter, which presents briefly the set-up of a simple
thermal case, the different modules of the software, starting by the "Software
Manager" are presented.
After the Run of the calculation chapter, three chapters are dedicated to the
Results viewing, the Results analysis and the Results exports.
Finally, Tips & Traps and Tutorials will illustrate the use of the software.
WHAT'S NEW
VERSION 2009.1
This section is describing the main news of ProCAST v2009.1.
This version is an update of version 2009.0 which corresponds to the first release
of the new STL Mesh Generator. It also includes bugs and problems correction.
The links are referring to the corresponding section of the manual for more details.
MeshCAST.
• STL Meshing : a new meshing algorithm has been introduced in MeshCAST.
It allows to mesh fully automatically multiple cavities (assemblies) defined by
one or several STL files (see the MeshCAST Manual for more details).
PreCAST
• A new mapping tool is now available. It allows to extract Temperature, FVOL
(filling information) and Porosity results from one calculation to an other, in
the case where the two calculations were made with different meshes (see the
"Extract and Mapping/Results Mapping" section for more details).
• New commands were added in the PreCAST Batch script (see the "Batch
Pre-processing/Script language" section for more details).
Thermodynamic Databases
• The Ni database was improved by the addition of Mn (up to 3%) and Cu. In
addition, the quality of the calculations of the liquidus and of the gamma prime
solvus were improved.
• The Fe database was improved in the case of alloys containing Nb (in
conjunction with C).
• A new section was added to the manual to illustrate the effect of the alloy
composition on the computed liquidus temperature, for Ni alloys (see the
"Influence of alloying elements/Ni alloys" section for more details).
• The model for the calculation of the Yield stress of Al alloys was improved.
Comparisons with experimental data are giving good results. Accordingly, the
Yield stress of the "A356-Stress" and "A356-Visco-Stress" entries of the
PreCAST database (stress.db) were modified to take into account this
improvement. As the Yield stress is depending strongly on the microstructure,
which is a result of different cooling rates, the A356-Stress single entry was
replaced by four entries, which are valid for 4 typical cooling rates of 0.1, 1,
10 and 100 K/s.
• The calculation of the Young's modulus of Cast iron was improved in order to
take into account the influence of the graphite shape (i.e. for gray, compact or
ductile iron). For this Young's modulus calculation, the effect of the graphite
shape is triggered by the Mg content, according to the following amounts :
Gray iron for Mg > 0.01%; Compact for 0.01% < Mg < 0.03%; Ductile iron
for Mg > 0.03%.
VisualCAST (please refer to the Visual 5.5 Release notes for full details)
• The "Scale dialog box" (e.g. to change the temperature scale for instance) can
be opened automatically by a double-click on the scale.
• The liquidus and solidus are now displayed on the temperature scale. This
display can be deactivated in the Scale dialog box (under Preferences - Datum
Line)
• The plot of free surface information (like temperature, velocity, ...) was
improved.
• Improvements in the Exports are now available.
Solver
• The limitation of 10000 steps in Radiation calculation has been removed.
• A bug in the reinitialisation of the displacement in the case of stress and
cycling was corrected.
• In the case of a FLOW only calculation, the viscosity (if defined as a function
of temperature) is taken at the temperature corresponding to the Tinitial value
defined by the user.
Manual
• A new section was added to the manual to illustrate the effect of the alloy
composition on the computed liquidus temperature, for Ni alloys (see the
"Influence of alloying elements/Ni alloys" section for more details).
VERSION 2009.0
This section is describing the main news of ProCAST v2009.0.
The links are referring to the corresponding section of the manual for more details.
General / Manager
• When the APM, the CAFE or the HCS solvers are called from the Manager, it
is possible to run the ProCAST solver first (for the thermal calculation for
instance). The script used by the Manager has been complemented in order to
allow the usage of "User functions" in the ProCAST run.
• The printout of the Wall clock time and Solid fraction has been corrected in
the "Status" in the case the "statpro" directory was not defined.
• VisualCAST is now the default post-processor. It is still possible to activate
ViewCAST in the "Installation Settings" of the Manager. If activated,
ViewCAST will be accessible with a right click on the "VisualCAST" menu.
• "Pyramid elements" have been introduced (i.e. elements with one quadrilateral
face and four triangular faces). At this stage, no ESI software is able to
generate such elements (but this will come in the future, especially for
boundary layer meshing), but PreCAST, DataCAST, ProCAST, VisualCAST
and ViewCAST are now compatible with this new element type.
PreCAST
• The writing of the d.dat file is performed after the completion of the
optimization (previously, if an old d.dat file was loaded and the optimization
was failing, the d.dat file was erased).
• New export capabilities of the stress results have been added. It is now
possible to perform a direct export in the Abaqus, ANSYS and Sysweld
format, as well as in a Neutral format (see the "Results Exports/Stress" section
for more details).
• The "Extract" capability allows now to extract also the Porosity results (in
addition to the free surface (FVOL) and the stress results) (see the
"Extract/Results Extraction" section for more details).
• The Die Combo capability was extended in order to allow distinct opening,
closing and spraying times for different parts of the molds. It will be possible
to define TOPEN, TCLOSE, TBSPRAY and TESPRAY with different values,
in each Die Combo Entry. However the possibility to have centralized values
in the Run Parameters will be still available if desired (see the "Interface
Database" section for more details).
• The model describing the nucleation and growth of the pearlite was improved
and a bug was corrected (to predict more accurate ferrite and pearlite
fractions). Two new Run Parameters (PERNUCL and PERGROW) were
introduced which allow to well calibrate calculated and measured
pearlite/ferrite fractions (see the "Micro Run Parameters" section for more
details).
Thermodynamic Databases
• The Ni database was improved by adding Re and Hf. In particular the density
calculation was improved.
• The Fe database was modified. The element "B" was added. In addition, the
database quality was further developed and improved by the addition of 13
new binary and 10 new ternary systems.
VisualCAST
• A Context based help is available in VisualCAST. The user can now press F1
key to get context based help, corresponding to the active panel.
• Information about the Free surface can be obtained as x-y plots. Time
evolution of the following quantities during the filling can be calculated (min,
max and average) : Temperature / Fraction Solid / Fluid Velocity-Magnitude /
Fluid Velocity-U / Fluid Velocity-V / Fluid Velocity-W / Pressure / Front
Tracking (i.e. Junction) / Area.
• The display of the SDAS can now be seen in microns.
• A bug which was creating a crash in the RGL calculation was corrected.
• The Contour scale is removed when the selected Contour is "None".
• The step frequency of the "Calculate New Particles" can now be set to values
lower then 10.
• The "Entity selector" and the Display control can now work together. This
means that while picking nodes for plotting temperatures, the user can switch
on/off parts in the same time.
• The scale of the porosity is now in % (instead of fraction).
• By default, the velocity magnitude results are shown for vectors when the
Results are "none".
• The calculation of the Alpha case (for Ti alloys) was improved and better
documented (see the "Criterion functions" section for more details). It is now
possible to use the Alpha case without a mold (this was not possible before). In
previous version it was needed that the casting material name was starting by
"TI". This limitation was removed and any material name can be used now.
• The pick node was improved in sections or cut-off contours. The nearest node
to section plane or cut-off iso-surface is now selected.
• Scan can now be exported as AVI animations.
• In the "Save Page as", it is now possible to save the Current window, the
current page, selected pages or all the pages.
• Each window has the "standard" icons on the top right to "iconize",
"maximize" or "delete" the window.
• In the "Entity selector list", it is possible to select a range with the Shift and
Control keys. For instance, it is possible to select several nodes (or a range of
the list) if required.
• The "Measure" panel was improved.
• In the "Save Page As" panel, it is now possible to export Animated GIF
movies.
ViewCAST
• The calculation of the Alpha case (for Ti alloys) was improved and better
documented (see the "Criterion functions" section for more details). It is now
possible to use the Alpha case without a mold (this was not possible before).
In previous version it was needed that the casting material name was starting
by "TI". This limitation was removed and any material name can be used now.
• A bug which was creating a crash in the RGL calculation was corrected.
Solver
• DataCAST was modified in order to allow spaces and special characters in
material names.
• When the APM, the CAFE or the HCS solvers are called from the Manager, it
is possible to run the ProCAST solver first (for the thermal calculation for
instance). The script used by the Manager has been complemented in order to
allow the usage of "User functions" in the ProCAST run.
• A problem of Restart in the case of FLOW = 9 has been fixed.
• The mass balance correction algorithm at the free surface in the case of
multiple free surface (e.g. multiple inlets) was significantly improved. This
allows to obtain now well balanced filling in the case of multiple inlets. This
new algorithm is now activated with FGROUP=2 (which is the default value
for both the Scalar and the DMP solvers) (see the "Run Parameters/Flow"
section for more details on FGROUP).
• A Restart problem (when a value of INILEV different or higher than the last
available step was specified) has been corrected. This was affecting restarts
when time dependant functions were used.
• In some cases, the flow solution was different when JUNCTION = 1 or
JUNCTION = 2 were used. This abnormal behavior was corrected and now
the flow solution is in dependant from the fact that JUNCTION is calculated
or not.
• The stress solver was modified in order to reset stress and strain data when
remelting was occurring (i.e. fs < CRITFS). This is correcting odd behavior in
the stress results which were occurring in the case of remelting.
• The free surface algorithm was further improve in order to prevent some
undesired flow patterns observed in some rare cases.
• In the case of a stress calculation, the stresses are now also stored at step 0.
This allows to visualize a stress calculation which have not started yet as well
as to have the right stresses and deformations at step 0.
• The version 2.3.3 of the PETSC library is now used, as well as the Intel
compiler 9.1 for both Linux and Windows.
• In the case of Stress calculations, when deformations become quite important,
some elements may become too distorted, leading to Negative Jacobian's. A
new procedure of "Mesh smoothing" has been introduced in order to cure this
kind of problem. This algorithm will only move "interior nodes", keeping the
same mesh topology (no surface nodes are moved, no new nodes or elements
are created). The Mesh smoothing is activated in the following situations : a)
globally at the beginning of each cycle, b) globally upon "restart" and c)
locally, when a negative Jacobian does appear. The Mesh smoothing is totally
transparent for the user as the post-processor is automatically taking these
movements into account (i.e. the displacements shown in the post-processor
are only the real "physical" displacements).
• When stress calculation during cycling are performed, often the deformations
can be rather large which can lead to highly distorted elements (and thus to
Negative Jacobian's). In order to prevent such problems, the mesh can be reset
to the original mesh at the beginning of each cyle, but the deformation is still
accumulated. Thus, the result is still correct, but without having the problem of
too important accumulated mesh distortions. If the calculation is able to go
through the first cycle, the calculation of the other cycles should be then
possible without problems. However, in the post-processing, we will still be
able to see a mesh which is distorting more and more along the cycles. To
activate this algorithm, the Run Parameter "CYCLE_ALGM" should be set to
2 (the default value is 1, which correspond to no special action and thus no
re-alignment) (see the "Stress Run Parameters" section for more details).
• The possibility to compute the final stress properties of a part, as a function of
its final microstructure and defects has been added (see the "Final Stress
properties as function of microstructure" section for more details).
• A bug was corrected in the writing of the p.out file. The translation
information of an enclosure is now well written.
• A bug about the flow restart when TFREQ > VFREQ was corrected.
• A true 64 bits DataCAST is now available on Windows.
• In the case of stress calculations, the heat transfer coefficient is automatically
affected by the gap width. In previous version, the gap was considered
automatically as air or as vacuum. In the case the gap is filled with something
else (e.g. flux or water), it is possible to define the gap properties in the
"interhtransfer.c" user function
2D
• A bug has been fixed for 2D quadrilateral meshes in the DMP flow solver.
• The possibility to perform an "Extract" in 2D axisymetric cases has been
activated.
• The problems of colors in ViewCAST while displaying 2D models were fixed.
Manual
• The description of the usage of the "Digitized hardening" was improved.
• A new description for FGROUP was added (see the "Run Parameters/Flow"
section for more details).
• APM : the multigas model is now fully described in the On-line manual
(previously it was only available in PDF) (see the updated version of the
"APM Multigas manual").
• APM : Information on the requirements about the possibilities for the Thermal
calculation to use "Restart" or "Extract" in conjunction with the APM were
VERSION 2008.0
This section is describing the main news of ProCAST v2008.0.
The links are referring to the corresponding section of the manual for more details.
General / Manager
• When the ProCAST solver is called before a CAFE or an APM calculation on a
64 bits Windows machine, the 64 bits ProCAST solver is now used.
• For large screens, it was necessary in the past to change the font sizes. The file
"procastFonts.tcl" (in the bin directory of the software installation) was
modified in order to allow an easy adaptation. Fonts can be adapted to large
screens by setting the variable "adaptFontsLargeScreen" to 1 (line 11 of the
procastFonts.tcl file).
MeshCAST.
• It is now possible to import large Patran meshes into MeshCAST (meshes with
more than 10 million elements).
• The "show feature" display is now remaining active during graphical
manipulations.
• It is now possible to restart an Assembly with different parameters (i.e.
different tolerances, ...).
• The Assembly algorithm was further improved. The feature lines are better
taken into account.
PreCAST
• A "Velocity calculator" has been introduced in both the "velocity BC" and the
"inlet BC". The calculator is now also applicable if time functions are used
(see the "Pre-processing/Databases/Boundary Conditions Database" section
for more details).
• It is now possible to automatically chain a cycling and a filling calculation. To
do so, a new run parameter CYCLEF has been introduced to activate the
filling at the last cycle only. At the beginning of each cycle, except the last
one, casting volumes will be set to full (see the "Cycling Run Parameters"
section for more details).
• It is now possible to automatically restart a calculation by setting INILEV to
an arbitrary large value. The last complete step corresponding to TFREQ,
VFREQ and SFREQ will be selected and the calculation will be restarted with
this last time step. It is assumed that MFREQ=TFREQ (see the "General Run
Parameters" section for more details).
• A new method to Extract data from one model to an other one has been
introduced in order to add flexibility in the use of ProCAST (see the new
"Extract" chapter for more details). In particular, it is possible now to
Delete/Add materials or domains while preserving the model set-up (with
Thermodynamic Databases
• The Yield stress can now be automatically calculated (for Al alloys only),
together with the Young's modulus, the Poisson ration and the Thermal
expansion coefficient. This option is only available with the "Back diffusion
model" (see the "Calculation of Stress Properties" section for more details).
• The Ni Database was improved with respect to the previous version. In
particular, the liquidus temperature are now much predicted. The difference
between the database of this version and the one of version 2007.0 can be
summarized briefly in the following way. Version 2007.0 did include many
VisualCAST (see the VisualCAST 4.0 on-line Release Notes for more
details)
• The algorithms have been improved in order to reduce the memory
consumption. Depending upon the cases, the memory needed to display a
given model has been reduced by 20-30%.
• The zoom is now maintained when the "Export Page movie" is used.
• It is now possible to maintain the "View settings" in Page Export.
• When an export is done, it is now possible to define a new folder location.
• The Void Pressure contour is not anymore active when a calculation was run
without the GAS model.
• An option to "Save the image to the Clipboard" has been introduced.
• When AVI movies are generated, it is now possible to select the compression
algorithm.
• To select nodes for XY plots, it is now possible to either pick the node
interactively or to use the coordinates (the selection mode is the same for both
evolutions and profiles).
• In the Scale definition panel, it is now possible to set negative delta values.
• All the contours can now be plotted with XY graphs.
• The calculated particles can now be stored/deleted/modified.
• In the case of a cycling calculation, the cycle information are now displayed in
the Contour and Plot windows. Tcycle, Topen and Tclose are indicated on
evolution graphs.
• The Plot interval definition is now only applied to visible parts.
• A new option to set the number of colors in the color scale has been
introduced.
• A new option to couple windows in order to apply the same graphical
transformations simultaneously in all windows has been introduced.
• The initial view when gravity was along a positive direction was corrected.
• A bug in the gap width export was corrected.
• A bug was corrected in the case of the display of a non-time dependant "usf"
file (i.e. with an output frequency of 0).
• A bug in the display of Tilt casting cases with stress has been fixed.
• A bug was corrected for the display of cases having solid rotations.
ViewCAST
• In the "Pick node", a 3rd digit was added in the display of the values.
• A bug was corrected for the display of cases having solid rotations.
• The -T2, -T2 and -T3 options (to change the display from seconds to
milliseconds or hours, minutes, seconds) have been re-introduced in the
manager.tcl file.
Solver
• Interpenetrating meshes can now be used together with cycling (e.g. cycling
with shot piston). This is available both with the Scalar and the DMP solver.
• In continuous casting simulations (steady state algorithm), the solid transport
term can now be applied to more than one domain.
• For the modeling of stress in large heavy castings, the solver was improved to
account for the casting weight on the mold, with the new Run Parameter
GLOAD = 1 (see the "Stress/Run parameters" section for more details).
• In the case of a stress calculation where the casting is not fully constraint by
the mold, a new Run Parameter GLOAD = 2 was introduced. This allows to
guarantee that the casting will stay in contact with the mold in the direction of
the gravity (see the "Stress/Run parameters" section for more details).
• A new Run parameter, named CLAYERS, has been introduced to adjust the
number of constraint layers used by GLOAD = 1. This allows to reduce the
effect of the constraints linked to the liquid nodes (see the "Stress/Run
parameters" section for more details).
• The stress solver was improved in order to handle the situation where one
solid is separating into two distinct volumes. This can happen during piping,
where a few solid elements can be separated from the main casting due to
piping.
• The possibility to define "visco-elastic" materials has been introduced (see the
"Visco-elasticity" section for more details). Such properties are useful for the
modeling of waxes used in investment casting.
• The new functionnality for the calculation and viewing of the "metal front
tracking" (also called the "JUNCTION" model), which has been introduced in
the previous version, has been significantly improved. It is now possible to
view the "accumulated free surface volume" (free surface multiplied by the
travelled distance), the "accumulated free surface time" (free surface
multiplied by the time), the "Free surface age" and the "Flow length" (see the
"Results analysis/Fluid Front Tracking" section and the "Run
Parameters/Flow" section for more details about this new functionnality and
about the Run Parameter JUNCTION).
larger models). Please note that the input data (defined in the prefix_poro.d file)
are different from the standard version and have been simplified. In particular
the gas properties are now defined in a database and do not need to be specified
in the input data.
• The Advanced Porosity solver (multi-gas version) is now available on 64 bits
machines (Windows and Linux). This allows to release the limitation of a
maximum of cells of 1-2 milions.
• The algorithm (multi-gas version) has been improved in order to reduce the
CPU time quite significantly.
Inverse
• The Inverse solver is now launched via a script file. This allows to customize it
in order to use the ProCAST DMP solver, if desired, (instead of the scalar one)
for the direct calculations of the inverse solve.
2D
• The 2D solver of ProCAST was reactivated (beside the pseudo-2D solver). In
MeshCAST 2D, it is possible now to create real 2-D meshes (and not only 3D
extruded meshes). PreCAST, ViewCAST, VisualCAST and the solvers were
adapted.
• In order to use the 2D version, nothing special has to be done, except to load in
PreCAST a 2D mesh instead of a 3D one (if a "planar" surface mesh *.sm is to
be used, just rename it as *.mesh before loading it into PreCAST).
• As this is the first release of the 2D, it is possible that some functionnalities are
not yet available or may exhibit problems (e.g. Radiation in DMP, stress
axisymmetric, ...).
Manual
• The on-line help manual is now working in the MeshCAST 64bits Windows
version.
• A new section about all the new possibilities of the Extract has been added
(see the "Extract" section for more details) including examples illustrating how
it can be used (see the "Extract examples" section for more details).
• A new section was added in the Tips & Traps chapter about LPDC
recommendations (see the "LPDC recommendations" section for more
details).
• A new section about Viscoplastic properties recommendations has been
added (see the "Viscoplastic Properties Recommendations" section for more
details).
• A new section on Visco-elastic models has been added (see the
"Visco-elasticity" section for more details).
• The section about how to launch DMP calculations with HP-MPI on clusters
was updated (see the "Use of the Parallel solver" section for more details).
VERSION 2007.0
This section is describing the main news of ProCAST v2007.0.
The links are referring to the corresponding section of the manual for more details.
General / Manager
• The possibility to select true 64bits executables for Windows platforms has
been added in the Manager (see the "Software Manager/Software
configuration" section for more details). 64 bits executables are available for
MeshCAST, ProCAST scalar and DMP solvers on Windows (please note that
the 64 bits executables are available on Linux, as before).
• The manager has been changed in order to allow the selection of pre-defined
nodes for DMP calculations (see the "Use of the Parallel solver" section for
more details.
PreCAST
• Automatic alignment of nodes along symmetry planes for rotational and mirror
symmetries (see the "Pre-processing/Radiation/Check of the symmetry planes"
section for more details.
• Introduction of FADING, a new microstructure Run parameter for the
modeling of Graphite fading in SGI (see the "Microstructure Run Parameters"
section for more details).
Thermodynamic Databases
• Improved description of Fe-S binary systems.
• Introduction of Sn element for Fe-Si-Sn systems (only for small addition of
Sn).
VisualCAST (see the VisualCAST 3.5 on-line Release Notes for more
details)
• Customizable menus and toolbars
• Possibility to define visibility and color attributes in Part Table menu
• Possibility to save and restore page templates
• Improvement of foreground/background handling in the slice menu
• Option to display the complete study path information in graphics window
• Definition of symmetries
• Model explode and move options
• Possibility to visualize relative displacements (w.r.t a plane)
• Possibility to visualize relative velocities in centrifugal casting
• Possibility to visualize a different contour on a Cut Off display
• New Cut Off info option to show average results based on selection
• Arithmetic and logical operators to create user defined contours
• Possibility to select different abscissa for curves (e.g. Plot of the Density as
function of the temperature when using Cast iron micro module)
• Quick results template
• Handling of time units
• Mesh export in various new formats including PAM-CRASH
• A bug in the calculation of the SDAS has been corrected.
• A bug in the "contact pressure" display has been fixed.
ViewCAST
• A problem in the visualisation of TILT + STRESS calculations has been fixed.
• The FACE-TO-GROUP information can be displayed also on the enclosure.
• A new functionnality for the calculation and viewing of the "metal front
tracking" is now available. This allows to view where the oxydes and
impurities trapped at the free surface are most likely to end-up as well as to
study the flow junctions (see the "Results analysis/Fluid Front Tracking"
section for more details).
• User defined contours in a new "Additional" contour panel are now available
(see the "Results viewing/Field Selection" section for more details).
Solver
• True 64bits executables for ProCAST solver are now available on Windows
for both scalar and DMP versions (please note that the 64 bits executables are
available on Linux, as before)
• In the scalar version (Windows and Linux) the PETSC solver was introduced
for stress calculations. This leads to better convergence and quality results. In
some cases, significant speed-up (30% to 100%) were observed.
• Improved core blowing option (NNEWTON=7) (see the new "Core Blowing"
chapter for more details).
• A new functionnality for the calculation and viewing of the "metal front
tracking" is now available. This allows to view where the oxydes and
impurities trapped at the free surface are most likely to end-up as well as to
study the flow junctions (see the "Results analysis/Fluid Front Tracking"
section and the "Run Parameters/Flow" section for more details about this new
functionnality and about the Run Parameter JUNCTION).
• Introduction of a new FADING parameter affecting the time during which the
Mg treatment has an effect (see the "Microstructures/Iron and Steel/Iron and
Steel models" section for more details).
• A new Run parameter (FREESFBAL) was introduced to stabilize large free
surfaces in gravity filling (See the "Flow Run parameters" section for more
details).
• For stress calculation, the handling of the penalty (for the contact algorithm)
has been improved by the introduction of the new Run parameter PENMOD
(see the "Stress Run Parameters" section for more details).
• The HCS solver (Hot cracking sensitivity calculation for continuous casting
which was available in calcosoft-3D) is now integrated in ProCAST (see the
"Hot Cracking (for continuous casting)" chapter for more details).
• A bug in the usage of FREESFOPT = 1 together with GAS = 1 was corrected.
• A problem in the TOPFILL algorithm has been fixed.
Innovations
This section was added in order to describe new innovative developments which
are included in the version although still in the industrialization/validation stage.
If you are interested to have access to such developments, please contact your
local support. Manuals about these subjects (in PDF format) are available in the
software installation (dat/manuals/Pdf directory). See the "Innovations" section
for more details.
• A new approach for Multi gas systems has been developed for the Advanced
Porosity Module (APM).
• The possibility to calculate heat treatment, based upon TTT diagrams has been
developed.
• User functions for material properties.
• Mold removal and Gate cutting in stress.
Manual
• A new section about Core Blowing (NNEWTON = 7) was introduced (see the
"Core Blowing" chapter for more details).
• A new section about Fluid Front Tracking has been added (see the "Results
analysis/Fluid Front Tracking" section for more details).
• The 2007 Release notes of CompuTherm are available in PDF format in the
software installation (in
../dat/manuals/PDF/CompuTherm_Release_notes_2007.pdf
• A new chapter about the HCS solver (Hot cracking sensitivity calculation for
continuous casting which was available in calcosoft-3D) was added (see the
"Hot Cracking (for continuous casting)" chapter for more details).
• An Innovation section was added (See the "Innovations" section for more
details).
• A section presenting comparisons of results between version 2006.1 and 2007.0,
as well as Scalar and Parallel calculations (2007.0) has been added (see the
"Results comparisons between versions" section for more details).
VERSION 2006.1
This section is describing the main news of ProCAST v2006.1.
The links are referring to the corresponding section of the manual for more details.
General / Manager
• The possibility to launch the VisualCAST post-processor from the manager
(see the "Software Manager/Module calls" section for more details).
MeshCAST
• A new shell mesher algorithm has been designed in order to improve greatly
the quality of shells (see the MeshCAST manual for more details).
• The handling of symmetries with the shell mesher has been totally revisited
(see the MeshCAST manual for more details).
• Boolean operations algorithms have been improved and made more robust. In
particular, it now accommodates models with enclosures (enclosures should
belong to the master part).
• The mesh smoothening operation following mesh assembly and Boolean
operations has been improved so as to better preserve geometry features.
• The logic of the "Check Mesh button" has been changed : A left mouse click
action will display cracks and multiple edges while a right click will display
bad triangles only.
• In the Assembly, the Master and Slave domains can be viewed with different
colors (with a right click).
• An option has been added to calculate the geometrical modulus from a volume
mesh. The geometrical modulus is printed for clip selections (see the
MeshCAST manual for more details).
• The Export of a volume mesh in SYSWELD format is now available.
• In Geomesh, it is now possible to save a model in "tos" format.
PreCAST
• A bug in the translation defintion x(t) was corrected.
• A problem of units when using a translation defined by v(x) has been
corrected.
• It is now possible to define a velocity BC with only one component (leaving
the other components empty, thus free).
• A bug in the use of Die Combo interface (with a temperature dependant table,
multiplied by a constant different than 1) was corrected. The constant was not
multiplying the table.
• It is now allowed to set an emissivity value of 1.0 (before this was creating
problems).
• The GRAPHITE and MGTREAT Run parameters have been added to the
"Micro" tab (see the "Micro Run Parameters" section for more details).
• The MOLDRIG Run parameter has been added to the "Thermal" tab (see the
"Thermal Run Parameters" section for more details).
Thermodynamic Databases
• The Cu thermodynamic database has been introduced (see the
"Thermodynamic Databases" section for more details). Please note that the Cu
database (provided by CompuTherm) corresponds to a first version and some
compositions (especially with Pb) may encounter problems (see the
"Databases limitations" section for more details). A document from
CompuTherm, entitled "CompuTherm_Copper_Database.pdf", describing the
Cu database is available in the software installation in the dat/manuals/PDF
directory.
• A back-diffusion model has been introduced for a better prediction of the
material properties as a function of the cooling rate (see the "Thermodynamic
Databases" section for more details).
• The Ni database was improved. A thermodynamic description of the gamma''
phase was developed. Many ternary systems, such as Ni-Al-Co, Ni-Co-Cr,
Ni-Al-Ta, Ni-Al-W, Ni-Co-Re and Ni-Co-Ta were modified in order to have
better properties prediction, including a better gamma'' solvus prediction. The
thermodynamic description of the eta phase was modified based upon
exprimental data.
• The Ti database was improved. Ternary systems, such as Ti-Al-V, Ti-Al-Mo,
Ti-Al-Sn, Ti-Al-Cr, Ti-Cr-Nb or Ti-Al-O were modified in order to better
predict commercial alloys. The thermodynamic description of the Ti-Si-Zr
system was developed including the ternary S2 phase.
• A document entitled "CompuTherm_Release_Notes_2006.pdf", available in
the software installation in the dat/manuals/PDF directory, is describing the
improvements of the databases.
• The prediction of the Youngs modulus for cast iron was improved in order to
take into account the effect of the magnesium on the microstructure.
• A problem encountered in the calculation of alloy properties (with
Computherm) on Windows when the software was installed in an other
location than the C:/ drive was corrected.
• A problem when calculating properties of Al alloys with Mn on Linux was
corrected.
• The Computherm databases on Linux were updated (the update was not well
done on version 2006.0).
• Problems encountered when erasing an element from the chemical
composition and re-calculating material properties have been corrected.
ViewCAST - VisualCAST
• A new post-processor called VisualCAST is introduced in this version in
addition to ViewCAST (see the separate VisualCAST DVD for more
information). A VisualCAST 3.0 Tutorial is available in PDF format in the
dat/manuals/PDF directory of the installation.
• The PATRAN export format of a mesh was slighty changed (a minus sign in
the material ID was removed) in order to allow a better compatibility with
other softwares.
• When the Heat flux, velocity, voids, stresses and displacement results are
exported, only the results of the active domain(s) are exported (and not the
results of all the domains).
• The export format (when using the PATRAN neutral file format) was changed
in order to account for exponents with 3 digits (in order to have a space
between values).
• A bug in the export of heat flux results has been corrected.
• The display of "Face to Group" view factors has been corrected and is also
available for enclosures.
• For centrifugal casting, the relative velocities can be calculated (in order to
remove the rotational component of the vectors) (see the "Centrifugal casting"
section for more details).
Solver
• A new option has been introduced in the micro module to allow for
coupled/uncoupled calculations (see the "Microstructures/Case set-up and
Results" section for more details).
• The Fading effect of the Magnesium treatment has been added to the SGI
micro model. A new Run parameter MGTREAT has been introduced in order
to account for the delay between the Mg treatment and the beginning of the
casting process (see the "Microstructures/Iron and Steel/Iron and Steel
models" and the "Micro Run Parameters" sections for more details).
• The density calculation for cast iron in the micro module has been improved.
• The effect of the white iron on the mechanical properties of grey iron was
introduced (when a micro calculation is performed).
• The filling algorithm was improved in the case of large ingots filling (no
change for other filling situations) (see the "Large Ingot Filling" section for
more details).
• A new algorithm (MiLE) has been introduced to address non-steady state
continuous casting problems (see the "Continuous Casting" chapter for more
details).
• The possibility to address steady state continuous casting processes has been
introduced (see the "Continuous Casting" chapter for more details).
• A bug was corrected in DataCAST in the case of time-dependant temperature
assignments to enclosures (occuring when more then 10 assignments were
made).
• Possibility to define the maximum timesteps DTMAX in a table, as well as the
maximum timestep during the filling DTMAXFILL (see the "General Run
Parameters" section for more details).
• The Freckles prediction was improved. This was done both by an
improvement of the Computherm database and by the introduction of the back
diffusion model (which both lead to a better prediction of the phases and thus
of the solidification path).
• The solver was modified in order to allow the viewing of results (in
VisualCAST only) every X% of filling or X% solidified. As a consequence,
the results calculated with ProCAST 2006.1 are not anymore fully compatible
with earlier versions of ViewCAST.
• Problems in processing datasets containing large numbers of temperature
dependant assignments have been corrected.
• A problem in the resetting of contact pressures in case of mold opening has
been corrected.
• A problem in the calculation of thixo casting has been corrected.
Manual
• A new section about recommendations for inlet velocities has been added (see
the "Tips and Traps/Inlet velocity" section for more details).
• The TPROF and WSHEAR Run parameters have been described in details
(see the "Flow Run Parameters" section for more details).
• The Tutorials are now stored in a separate document.
VERSION 2006.0
This section is describing the main news of ProCAST v2006.0.
The links are referring to the corresponding section of the manual for more details.
General / Manager
• In the "Copy" panel, the possibility to access directly to the "Template"
directory of the installation was added (see "Software Manager/File
Manager" section for more details).
• The access to the CAFE pre-processor and post-processor were introduced in
the Manager, as well as to the Advanced Porosity post-processor (see the
"Software Manager/Module calls" and "Software Manager/Advanced module
calls" sections for more details).
• A bug in the "Copy" functionnality was corrected.
• The possibility to launch DMP calculations on Linux in Batch was introduced
(see the "Parallel Solver/Use of the Parallel solver" section for more details).
• The FlexLM libraries were upgraded from 9.2 CRO to 10.1.3 STL.
• A bug in the status window (wrong update of the number of cycles) was
corrected.
MeshCAST
• A link between GEOMESH and MeshCAST was created in order to speed-up
the repair and surface meshing process (especially to account for the more
efficient readers of GEOMESH) (see the MeshCAST manual for more
details).
• Surface mesh Assembly (see the MeshCAST manual for more details).
• Boolean Assembly (see the MeshCAST manual for more details).
• The Surface mesh algorithm was improved in order to have a better quality
surface mesh. Especially, cylinders defined by only one surface can now be
meshed automatically. Thin cylinders (e.g. cooling chanels) are not collapsed
anymore when the mesh size is coarse.
• The Autofix is now fully automatic (no more need to specify a length)
• A functionnality to remove automatically filets was introduced (see the
MeshCAST manual for more details).
• The Layered shell mesher was re-introduced in MeshCAST (see the
MeshCAST manual for more details).
• The format of the volume mesh has been adapted to allow the handling of files
with more than 10 millions elements.
• When a Volume mesh is loaded, new buttons "Select All" and "Deselect All"
were introduced in order to select/deselect automatically all the domains.
• The "Partial layer" option was corrected (it was giving the same results are the
Full layer option).
• The "*.bstl" files can now be read on Linux.
PreCAST
• Time-dependant Inlet boundary conditions were introduced (see the
"Boundary Conditions Database" section for more details).
• In the case of shot piston, it is possible now to define the position of the piston
in three different ways : position vs time, velocity vs time, velocity vs position
(see the "Process Database" section for more details).
• Wall BC and Velocity BC are now exclusive (i.e. if a Wall BC with a "Select
All" is set, the previously defined Velocity BC's will be preserved).
• Now, it is not anymore possible to set BC's on Periodic BC faces (the Periodic
BC faces are "protected", like Symmetry faces).
• A new "Pick" function has been introduced, to pick the coordinates and the
node number of a given point selected interactively on the model (see the
"Geometry Manipulation" section for more details).
• Stress properties (Youngs modulus, Poisson Ratio, Expansion coefficient) are
now automatically calculated, based upon the Thermodynamic databases (see
"Calculation of Stress Properties" section for more details).
• When defining the interfaces, the "Wireframe" mode has been reactivated (in
order to see a domain which is inside an other one).
• The Plastic stress properties (i.e. the Hardening) can now also be defined as a
set of tabulated tensile tests curves in an ASCII file (see "Digitized Hardening"
section for more details).
• In the case of Microstructure modelling, the default values of the Run
parameters were removed (see the "Micro Run Parameters" section for more
details).
• The mesh can be now optimized also in the case of Periodic BC.
• If a non-coincident mesh is used, a warning prevents the definition of a virtual
mould.
• New User functions have been introduced (see "User Functions" section for
more details).
• A problem in the initialization of CORE materials during cycling has been
corrected.
• A problem of Extract of temperatures with the prefix_t.unf file larger then 2
GB has been corrected.
• Enclosures can now again be defined by QUAD elements
• A new Run Parameter (GATEFS) was introduced for the handling of the third
stage pressure in HPDC (see the "Thermal Run Parameters" and "Active
Feeding" sections for more details).
• The set-up of Centrifugal cases was changed for more accuracy and the
RELVEL Run Parameter was introduced (see the "Centrifugal casting" and the
"Flow Run Parameters" sections for more details).
• Due to the changes in PreCAST and DataCAST, it is advised to reload the
d.dat and p.dat files generated on previous versions and save them before
re-running the cases with the version 2006.0.
• The Run Parameters Recommendations have been updated in order to account
for the changes in the algorithm (see the "Run Parameters Recommendations"
section for more details).
Thermodynamic databases
• Stress properties (Youngs modulus, Poisson Ratio, Expansion coefficient) are
now automatically calculated, based upon the Thermodynamic databases (see
"Calculation of Stress Properties" section for more details).
• The latest available Computherm databases are included in this version. The
material properties calculated with these databases will be more accurate than
the previous ones.
• The Computherm manual (from Computherm LCC) is added in the Software
installation (in the dat/manuals/PDF direcrtory). This manual describes for
each alloying system the phases which are calculated, the limitations as well as
the validations which have been made.
ViewCAST
• A new "Pick" function has been introduced, to pick the coordinates, node
number and values (e.g. Temperature, Pressure, ...) of a given point selected
interactively on the model (see the "Geometry Manipulation" section for more
details).
• The liquidus and solidus temperature are now shown on the Temperature scale
(see the "Results Viewing/Display Parameters" section for more details).
• The previously defined scales (i.e. the scales defined manually during a
previous session) are now automatically stored and loaded (see the "Results
Viewing/Display Parameters" section for more details).
• Stored views were introduced (see the "Results Viewing/Geometry
Manipulation" section for more details).
• When a model is loaded for the first time (or when no view was stored), the
model is displayed in isometric view, with the gravity pointing downwards.
• In case of a Stress calculation, it is now possible to calculate all the Stress
results (e.g. Effective Stress, Principal Stress, SigmaX, ...) all at once before
viewing them (see the "Results Viewing/Field selection" section for more
details).
• In the case of stress calculation, the possibility to view the total displacement
was introduced (see the "Results Viewing/Field selection" section for more
details).
• The displacements relative to a plane were introduced for the visualization of
deformed geometries (stress calculation) (see the "Results Viewing/Field
selection" section for more details).
• When a stress calculation is performed, it is possible to display the
underformed geometry in wireframe together with the deformed contours (see
the "Results Viewing/Display Parameters" section for more details).
• The calculation of the solidification time was improved in order to account for
remelting (if any).
• The calculation of the Filling time was changed in order to ignore the piping
area.
• The pressures calculated by the "Advanced Porosity module" can now be
visualized in ViewCAST (see the "Advanced Porosity Post-processing"
section for more details).
• In order to better visualize pockets of air, a new display mode of the free
surface ("Foreground") was added (see the "Results Viewing/Display
Parameters" section for more details).
• Stress results can be viewed also when the directory is in Read-only mode (on
Windows only). This is especially useful to view stress cases stored on a CD
or a DVD. The stress results are computed and stored in a temporary file on
the local disk of the computer (see the "Results Viewing/Field selection"
section for more details).
• If more than one symmetry is defined in the model, it can be automatically
retrieved (it was previously working only with one symmetry).
• In the case of a Tilt casting, it was mandatory to have the p.dat file present in
the directory to view the case. This obligation was removed.
• In the XYPlot window, when nodes were selected interactively, the selection
window (with the cursors) was disappearing each time the model had to be
rotated or moved. This limitation is now removed.
Solver
• The filling algorithm was improved. A new Run Parameter FREESFOPT was
introduced (see the "Flow Run Parameters" section for more details).
• The filling solver was improved in order to better handle Tilt pouring models.
A new Run parameter "TILT=1" was introduced in the case of Tilt models
(see the "Flow Run Parameters" section for more details).
• For centrifugal casting, the algorithm was improved by the introduction of the
RELVEL Run Parameter (see the "Centrifugal casting" and the "Flow Run
Parameters" sections for more details).
• The convergence, the accuracy and the CPU time of the Stress module was
significantly improved.
• Two new models (Norton law and Strain Hardening Creep) were introduced in
the Stress module to account for viscoplasticity and creep (see "Stress models
and Properties" section for more details).
• The Plastic stress properties (i.e. the Hardening) can now also be defined as a
set of tabulated tensile tests curves in an ASCII file (see "Digitized Hardening"
section for more details).
• In the case of Stress calculations with Vacant or Rigid materials, the memory
management was changed in order to reduce the amount of memory needed (in
the scalar version only).
• Stress calculations can now be run with piping (i.e. it is now allowed to set
PIPEFS<1 in the case of stress calculations).
• The POROS=1 model was extended to allow the treatment of the Graphite
expansion in the case of SGI (see "Cast Iron Porosity model" section for more
details).
• The POROS=1 model was improved for the treatment of the third phase in
HPDC. A new Run Parameter (GATEFS) was introduced (see the "Thermal
Run Parameters"and "Active Feeding" sections for more details).
• A bug in the POROS=1 model was corrected (in the identification of the
independant liquid pockets).
• A problem in the POROS=1 model with FREESF=2 filling was corrected.
CAFE
• The CAFE pre-processor and post-processor were introduced in the Manager
(it is no longer needed to have the calcosoft-3D installation to set-up and
visualize CAFE calculations) (see the "CAFE-3D" chapter for more details).
• The CAFE solver (in post-processing mode) was introduced in the ProCAST
2006.0 distribution (it is no longer needed to have the calcosoft-3D installation
to run CAFE calculations) (see the "CAFE-3D" chapter for more details).
Manual
(The following sections of the manual were added or changed)
• The Stress properties recommendations were modified. One should note that it
is very important to set stress properties according to the "rules" described in
the manual. Othewise, one could have convergence problems (see the "Stress
Models and Properties" section for more details - end of the section).
VERSION 2005.0
This section is describing the main news of ProCAST v2005.0.
The links are referring to the corresponding section of the manual for more details.
General / Manager
• The manager was modified in order to allow the launch of Parallel
calculations, of the Advanced Porosity module and of the Inverse module (see
the "Software Manager/Advanced module calls" section for more details).
• A "Run list" functionality was added in order to launch automatically a serie
of calculation (batch processing) (see the "Software Manager/Run list" section
for more details).
• A Graphical status was introduced in the Manager (see the "Software
Manager/Module calls" section for more details).
• The F2 key can be used to scroll between the different case names which are
present in the same directory (see the "Software Manager/File manager"
section for more details).
• The case name was added in the top bar of all the modules (MeshCAST,
PreCAST, ViewCAST).
• The suffix ".mesh" has been added to the Filter of the "File Manger".
• The FlexLM locking system was upgraded to FlexLM 9.2 CRO, for a better
security. This version will work with 2005 CRO license keys.
• LMtools can be accessed directly from the Manager (see the "Software
Manager/Software configuration" section for more details).
MeshCAST
• The call of the Volume mesher of MeshCAST in batch was re-introduced (see
the end of the "Software Manager/Module calls" section for more details).
• The possibility to read Parasolid files with the .x_t suffix was added.
• The automatic "zoom-out" was suppressed when the ADDLINE operation was
performed.
• A units problem in MeshCAST-2D was corrected.
PreCAST
• A GATENODE Run parameter was introduced in order to handle "Gate
feeding" with a shot piston (see the "Active Feeding" and "Thermal Run
Parameters" sections for more details).
• The computation algorithm of the Virtual mold was modified (to avoid odd
behavior in some cases).
• The virtual mold size can be defined automatically (see the "Virtual mold"
section for more details).
• The possibility to activate User functions was introduced in PreCAST (see the
"User Functions" section for more details).
Thermodynamic databases
• The automatic calculation of the "Thermal conductivity" from the
thermodynamic databases was introduced for all databases (see the
"Thermodynamic Databases" section for more details)
• The content of the Computherm databases was improved. This will lead to
more accurate properties predictions. Moreover, the computation of some
alloys (which was not possible with previous versions) was improved. This
means that the properties calculated with this version will be different (but
more accurate) than the previous ones (e.g. the solidus temperature of
Ni-based alloys is now much more realistic than before).
• A problem in the Mg database was fixed, giving now the right results.
• The liquid density of Pure Aluminium was changed in order to give more
accurate density predictions of Al-alloys (e.g. in the previous version, the
solidification shrinkage of an A356 alloy was about 9.15%, whereas it is now
5.45%). This will affect the density calculations for all Al alloys.
ViewCAST
• ViewCAST was adapted in order to allow the viewing of result files coming
from a different platform (e.g. viewing on Windows results computed on
UNIX) (see the "Software Manager/Software configuration" section for more
details).
• Automatic creation of AVI animations was introduced (see the
"Resultsviewing/Introduction" section for more details).
• The GIF/AVI capture action can also be started from the "File" menu.
• Export of the "Shrinkage porosity" and of the "Advanced porosity" fields has
been introduced.
• An automatic pre-defined setting for "Shrinkage porosity" was introduced (see
the end of the "Results viewing/Display types" section for more details).
• Short cuts for the different display modes were introduced (see the "Results
viewing/Display types" section for more details).
• The export of "Reversed displacements" was corrected, as well as the X-Y
plots of displacements.
• The maximum number of timesteps which could be exported was increased
from 200 to 5000.
• In Tilt casting, the particles are now "tilting" with the part. In addition, a new
"Tilt" option in the "Parameters" menu was introduced in order to deactivate
the tilt during the viewing (see the "Display Parameters" section for more
details).
• The "Fill time" plot was introduced.
• A problem with the calculation of RGL on Linux was fixed.
Solver
• User functions were introduced for "interface heat transfer coefficients" and
"Heat boundary conditions" (see the "User Functions" chapter for more
details).
• A bug with the TCLOSE (DIE COMBO cycling) was corrected.
• A problem in the cementite formation of cast iron (microstructure module)
was fixed.
• The interpolation of the linear thermal expansion coefficient (stress module)
was changed from linear by intervals to piecewise linear.
• A problem with the calculation of stress in the presence of piping (stress
module) was corrected. In such case, one should set PIPEFS = 0 (see the
"Thermal Run Parameters" section, under PIPEFS for more details).
• A new possibility was included in order to handle the "Gate feeding" in hpdc
in the presence of a shot piston (see the "Active Feeding" and "Thermal Run
Parameters" sections for more details about the GATENODE Run Parameter).
• A bug in the POROS 1 model was corrected (to avoid unexpected behaviors in
some specific cases). Moreover, the display of the macroshrinkage (i.e. holes
inside the casting) has been changed (one will have shrinkage porosity values
of 1, but no internal "Voids" will be shown. "Voids" will be used only for pipe
shrinkage at the surface of the model). Please note that the value of FVOL
has no meaning for the viewing of porosity results.
• A problem in the Surface load (stress module) which was appearing in some
cases was corrected.
• The influence of the gap width on the interface heat transfer coefficient, with
non-coincident meshes, was introduced.
• The viscoplastic model (stress module) was improved, for a better accuracy.
• The stress module was extended in order to allow the use of filters.
• DIE COMBO was extended for the use with non-coincident meshes.
• A problem in the handling of the "RESERVOIR" was corrected in order to
allow to fully empty the reservoir.
• A problem with the Freckles indicator was fixed.
• A new stop criterion (TENDFILL) was introduced. It allows to stop the
calculation N seconds after the end of filling (see the "General Run
Parameters" section for more details).
• The CAFE-3D coupled solver was introduced in the standard ProCAST solver
executable (see the "CAFE-3D" section for more details).
• Undesired "Sticking" in the pouring cup during Tilt pouring was improved.
For that purpose a new TILT Run Parameter was introduced (see the "Flow
Run Parameters" section for more details).
• A "loading increment" Run Parameter LOADSCL was introduced for more
accuracy in structure analysis type of problems (see the "Stress Run
Parameters" section for more details).
• A problem with RDEBUG=2 was fixed.
Parallel processing
• The Parallel processing solver was introduced in this version for selected
platforms (see the "Parallel Solver" chapter for more details).
Manual
(The following sections of the manual were added or changed)
• A new section was added to list the limits in the chemical compositions for the
Thermodynamic databases (see the "Databases limitations" section for more
details).
• A section on the "Influence of alloying elements" on thermophysical
properties obtained with thermodynamic databases was added.
• A new "Software Manager/Advanced module calls" section was added.
• A new "Software Manager/Run list" section was added.
• The "Software Manager/Software configuration" section was complemented.
VERSION 2004.1
The links are referring to the corresponding section of the manual for more details.
General
• On some platforms, the graphics performances were very poor (in terms of
display speed). This was corrected. However, it may still happen that on some
graphics cards, the graphical performances are not optimum (in terms of
display refresh). In such cases, the configuration of the X-Server may need to
be adjusted.
• The installation structure of the software was complemented in order to allow
the installation in a Read only location. Each user has now the possibility to
set a customized directory for his Preferences (including the MeshCAST user
customized icons) and databases. (Customized installation).
MeshCAST
• On some platforms (mainly Windows), MeshCAST was not always able to use
all the available memory (e.g. it was crashing at 1.1 GB, whereas the computer
had 2 GB of RAM). This problem was due to the way "reallocation" of
memory was done in the code and how the operating system was managing the
memory allocation. The memory "reallocation" technique was modified in
order to prevent such problems, allowing to use all (or most) of the available
memory. In such situation, temporary files will be created in the working
directory (see the MeshCAST manual for more details about this option and
the "Module calls" section for the activation of this option).
• In the case of problems with Shell meshing with symmetry, a new
functionality was added in order to allow the curing of the problem (see the
MeshCAST manual for more details about this option).
• The User defined icons can now be stored in a central location (or in a
customized user defined location) (Customized installation).
• Some corrections when reading Parasolid files were made.
• ANSYS surface mesh input is now available.
PreCAST
• The viscosity and density can be automatically calculated from the
thermodynamic databases, based upon the chemical composition
(Thermodynamic Databases).
• The thermal conductivity for Aluminium alloys can be determined
automatically from the thermodynamic databases, based upon the chemical
composition (Thermodynamic Databases).
ViewCAST
• A new "Particle tracing" has been introduced. The streamlines are
automatically computed during the calculation and thus the display in
ViewCAST is fast (see "Display parameters" for the viewing and "Flow Run
Parameters" for the set-up).
• When a case is viewed during the calculation, the timesteps can be
automatically updated during the ViewCAST session (see the Display
parameters section for more details).
• In XY-Plots, the Import/Export of the node numbers (as well as the scale) was
not working well. This was corrected. See the "Curves" section for the format
of the Import/Export.
• The GIF capture was not always working with screen resolutions different
from 1280x1024. This was corrected.
• When capturing a GIF picture/animation, if the ".gif" extension is not
specified, it will be automatically added.
Solver
• Although many improvements were made in version 2004.0 in the filling
algorithm, in order to reduce numerical diffusion (to have "less viscous"
liquid), this did create some detrimental behaviors in some situations. Thus,
the filling algorithm was further refined and tests have shown much better
results in all the geometries and processes which were calculated. A special
effort was put in the Run Parameters settings for filling simulation. Now, the
default values can be used for all processes (only WALLF should be
customized in order to account for the "friction" of the liquid at the mold wall)
(Flow Run Parameters, Run Parameters Recommendations).
• In LPDC, the stability of the filling with a pressure boundary condition was
totally reviewed and now good results are obtained also with pressure filling
(Flow Run Parameters, Run Parameters Recommendations)
• The POROS=1 model was improved and a few bugs were corrected. The
description of the POROS=1 model was also improved (see in particular the
description of the effect of FEEDLEN)
• The visco-plastic model was slightly changed in order to allow a more
straightforward definition of the input parameters from measured data and the
possibility to set a Norton law (with threshold) for creep (Stress models and
properties). One should be careful that viscoplastic models which have been
set in previous version should be modified in PreCAST in order to add the
normalization stress. Otherwise, the solver will not work.
• Porosity calculations can now be done together with cycling.
• In the Lost Foam model, there was situations where small bits of foam were
not burning. This situation was corrected.
• The stress solver was extended in order to account for moving solids.
• A bug in the Restart of Stress calculation was corrected, leading to a much
improved convergence of the restarted step.
• A bug in the GATEFEED=1 option was corrected in the case of
time-dependant pressure or inlet velocity (GATEFEED was not activated in
this cases)
• A check was introduced at the beginning of the calculation in order to prevent
a missing pressure BC when GATEFEED is set to 1 (to prevent piping in
pressure die casting, in thermal only calculations).
• A bug was corrected in the "inlet BC" algorithm when nodes at the interface
between the casting and the mold were selected. One should be careful in such
cases that the filling time may not be exactly correct (if interface nodes are
selected).
• A bug in the micro module for hyper-eutectic cast iron was corrected.
• When the microstructure module is used, the material properties definition has
slightly changed (as some data can be automatically obtained from the
thermodynamic databases) - see the "Microstructures/Case set-up and Results"
section for more details.
• A print-out bug was corrected in the d.out file.
• On some platforms, it was still not possible to generate files larger than 2 GB.
This is now corrected. One should be careful however that the file system
(kernel) should be well configured in order to accept such large files.
However, as it was not possible to generate a 64 bits version of the User
interfaces (and especially ViewCAST) on IBM (due to Tcl/Tk problems), it is
not possible to visualize such large models on IBM at this stage. Work around
: visualize these results on a different platform.
• A bug in the calculation of the permeability of a mushy zone at low fraction of
solid was corrected. This has an effect on cold shots or early solidification
during filling (in case of very large interface heat transfer coefficients).
• DataCAST execution time was reduced for large models.
Manual
(The following sections of the manual were added or changed)
• Changes in the definition of the visco-plastic and creep model (Stress models
and properties)
• Correction in the description of the FEEDLEN effect in the POROS=1
porosity model.
• New tutorial for LPDC.
VERSION 2004.0
The links are referring to the corresponding section of the manual for more details.
General
• All the user interfaces have been re-developed, using the Tcl/Tk and OpenGL
technology. Thus, NutCracker is not anymore necessary to run the software on
Windows machines
• The software is licensed using the FlexLM tool (FlexLM). This gives the
possibility to have floating licenses
• Direct access to the on-line help
Manager
• A new Software Manager was introduced (to replace "PCS")
• Directories can be created and managed (File Manager)
• Cases can be automatically copied, with prefix changes (File Manager)
• The different modules can be called interactively (Module calls)
• The calculation can be monitored with a direct status access (Module calls)
• The software configuration can be tailored in the Manager (Software
configuration)
MeshCAST
• The volume meshing algorithm is significantly improved in order to generate
less elements and nodes with a better quality
• The 2-D mesh generator is replaced by a new 2-D mesh generator based upon
the one of calcosoft-2D. It is extended in order to automatically generate 3-D
meshes of one slice for both cartesian and axisymmetric modeling
• The maximum number of surfaces which can be allocated to an enclosure was
increased to 5000
• The new user interface is available for UNIX/Linux users
PreCAST
• PreCAST was totally re-developed in a new user interface, based upon the
same fundamental principles as the previous version (Pre-processing)
• Interactive tools to manipulate the geometry (rotate, zoom, center, drag)
(Geometry manipulation)
• Different display types (wireframe, hidden, hidden mesh, shade) (Geometry
manipulation)
• New symmetry definition, with interactive pick of the coordinates (see the
Radiation section)
• Enhanced definition of the Virtual mold, with possibility to view the thermal
depth and the mold box (Virtual Mold)
• Display of the size of the model (Geometry Assignment)
• New Material database design (Material Database)
• Coherency about material properties definition (enthalpy, Cp, fs and L)
(Material Properties)
• Direct visualization of tables (Material Database)
• Direct access to material and boundary condition databases (Materials
assignment, Boundary conditions assignment)
• Search and Sort tools for each database (Material Database)
• Exothermic material definition (Exothermic)
• More explicit definition of the material domain types (casting, mold, foam,
filter, exothermic, ...) (Materials assignment)
• Definition whether a material domain is empty or not in the "Materials
assignment" menu
• Possibility to remove interfaces which have been previously created
(Interfaces assignment)
• More flexible way to define non-coincident interfaces (Interfaces assignment)
• Possibility to change interactively the non-coincident interface tolerance
(Interfaces assignment)
• In all databases, possibility to Import or Export the tables (Material Database)
• Enhanced selection of boundary conditions, automatic propagation, selection
of remainder, selection of interfaces (Boundary conditions assignment)
• Copy, Paste and Clip capabilities in the Boundary conditions selection
(Boundary conditions assignment)
• A "velocity calculator" was introduced in order to calculate the desired inlet
velocity boundary condition (Boundary Conditions Database)
• For radiation problem with an enclosure, the alignment of the surface vectors
is simplified (Radiation)
• For all databases, the database management (add, read, copy, delete) can be
done in the same window as the assignments (Material Database)
• New design of the Run Parameters layout (Run Parameters)
• Automatic settings of up to nine Run Parameters Templates (Pre-defined Run
Parameters)
ViewCAST
• ViewCAST was totally re-developed in a new user interface, based upon the
same fundamental principles as the previous version (Results viewing, Results
analysis, Results export)
• Better rendering of the models and of the free surface, including shading and
transparency
• Interactive tools to manipulate the geometry (rotate, zoom, center, drag)
(Geometry manipulation)
• Full capability "Tape player" to step forward and backwards in single steps or
animated mode (Tape Player)
• More comprehensive way to select the different view modes of the material
domains (solid, invisible, wireframe, hidden mesh, shaded, transparent)
(Geometry manipulation)
• Direct access to the different view modes (snapshot, slice, scan, cut-off)
(Display types)
• Interactive slicing is available at any time (Display types)
• Enhanced timestep display definition, with the possibility to define time
intervals for the display (in addition to step intervals) (Display parameters)
• Full export of stress results (in a Neutral format which can be imported in any
stress software) (Results Export/Stress)
• Full export of the displacements (in a case of stress calculation), with the
possibility to "reverse" the displacements in order to perform reverse
engineering of distorted molds (Results Export/Displacements)
• Automatic calculation of the solidification time (Field selection)
• Enhanced symmetries settings (same panel as in PreCAST) (Display
parameters)
• Merge of the former PostCAST capabilities into ViewCAST. This allows to
have only one post-processor for the snapshots and the time evolutions curves
(Results viewing/Curves), as well as for the generation of outputs, such as
Niyama, cooling rates, SDAS, ... (Results analysis/Criterion functions)
• The XYPlots settings have been totally reviewed (Results viewing/Curves)
• The node selection for XYPlots can now be stored, as well as the scale of the
graph for further use (Results viewing/Curves)
• Three different modes for the viewing of the free surface have been introduced
(On, Off, Only) (Display parameters)
• The definition of the scale is totally new and interactive. The scale is kept in
memory for each field (Display parameters)
• The materials can be "exploded" for a better viewing (Geometry manipulation)
• The display of vectors is improved. It is possible to define the length of the
arrows, as well as to have a unique length for all the arrows (Display
parameters). The unit vector length is now shown on the screen during
snapshots and slices
• An automatic scan of slices (scanning along the X-, Y- or Z-direction) can be
done at a given step (Display types)
• The definition of slices (cuts) is improved. It is now possible to view or not the
geometry behind (or ahead) of a slice (Display types)
• The "cut-off" capabilities were improved. It is now possible to specify two
cut-off values (Display types)
• The display in cut-off mode in the presence of a free surface is now well
treated (Display types)
• The information about the case status (prefix, displayed field), time, timestep
has been reviewed and is now shown on the bottom of the screen (Results
viewing/Introduction)
• The free faces information is now generated in DataCAST and thus the
loading of a case is much faster
• Full viewing capabilities are now available when symmetries are activated
(cut-off, free surface, ...)
Solver
• The flow solver for filling was extensively reviewed. In particular the flow
along walls was significantly enhanced (WSHEAR=2). The stability of the
solver was thus improved for filling (Flow Run Parameters)
• A new Run Parameter (WALLF) was introduced in order to better account
for the surface quality of the mold (Flow Run Parameters)
• The EDGE Run parameter was suppressed (due to the WALLF introduction)
• The end of the filling can be shortcut with the new Run parameter ENDFILL
(Flow Run Parameters)
• A new "Pool" model was introduced (RESERVOIR type) in order to facilitate
the modeling of cases with large horizontal free surfaces, such as tundishes
(Materials assignment)
• Improved lost foam model
• The inlet boundary condition was improved in order to preserve the right
amount of incoming mass
• The Porosity model (POROS=1) was significantly improved in order to give
more accurate predictions (Porosity models)
• The density definition in the case of porosity calculations was extended
(Density definition)
• The gate feeding (in the case of injection - hpdc - lpdc) was changed in order
to be fully automatic (Active feeding)
• Enhanced enthalpy formulation for more accuracy of the thermal solver
• Exothermic materials can be modeled (Exothermic)
• Large files (over 2 GB) can now be handled
• The visco-plastic model for stress was adapted to account for creep (Stress
models and properties)
• Fatigue prediction is available (Fatigue life indicator)
• Hot tearing calculations are available (Hot tearing indicator)
• Cracking indicator is introduced (Cracking indicator)
This section contains the identified bugs and problems which could not be
corrected in this version.
General
• On UNIX, the command window which is used to launch the manager is not
available for any further operations (as long as the manager is open).
• On some platforms, there may be an error message when exiting the different
modules. This message can be ignored.
Manager
• Problems may be encountered if the working directory name contains dots
(".").
• If the Status is refreshed too often (by clicking on the "Status" button), a Tcl
error may be encountered. This happens if the refresh is done exactly in the
same time than the writing of the stat file by the solver. This can also happen if
two refresh are done in a very short time interval.
PreCAST
• On some platform, the "Get coords" functionality to define symmetries is not
working properly.
• On Sun, windows are opened behind the application, which can not be
accessed.
• The display of the Virtual mold thermal depth has some problems.
• If GATEFEED is set to 1 and POROS to 0, the calculation does not start if no
pressure BC is set (as there is a mistake in the logic of such testing). Work
around : one should set GAFEFEED = 0 when POROS = 0.
• When a case is opened directly from PreCAST (and not from the Manager), if
this case is located in a different directory (from the directory where PreCAST
was launched) and if this directory contains a ".", the case can not be opened.
Work around : open the case directly from the Manager.
• The material names should not contain blank spaces (this will generate an
error in DataCAST : "An enthalpy function is referenced in 5 0 data
set 1 which has not been given". Work around : if an old "d.dat" file is
containing such a material name, one can edit the d.dat file and change the
material name in the "card 5.0" to suppress the blanks ( 5 0 1
NICKEL_IN_718 1 )
• When a translation velocity as a function of position, v(x), is defined by a
table, the velocity values should not contain any zero values.
ViewCAST
• Some problems were observed in the visualization of stresses on SGI.
• When displaying a slice with vectors, some problems may occur.
• When displaying vectors in a slice, the contour of the slice does not appear.
• It is not possible to superimpose a contour and vectors (in slices).
• Sometimes, the particle tracing is giving wrong results at the last steps of the
filling.
• In same cases, the automatic timestep update is freezing ViewCAST on SGI.
Work around : do not activate the automatic timestep update on SGI.
• The visualization of result files larger then 2 GB is not possible on IBM (as the
executable of ViewCAST is 32 bits). Due to Tcl/Tk problems on IBM 64 bits,
it was not possible to create an executable of ViewCAST 64 bits. Work around
: use an other platform to visualize such results.
• On UNIX, depending on the configuration of the X-Server, it was observed
that the performances of the graphics may be not optimum.
• When a case is opened directly from ViewCAST (and not from the Manager),
if this case is located in a different directory (from the directory where
ViewCAST was launched) and if this directory contains a ".", the case can not
be opened. Work around : open the case directly from the Manager.
• Particle tracing is not available in Tilt and Reservoir cases (i.e. when there is
no inlet velocity, inlet pressure on inlet BC).
• In Tilt casting the velocity vectors will not "rotate" with the part and thus they
will appear in the wrong direction. Work around : deactivate the tilt rotation in
ViewCAST when viewing at velocity vectors.
• It is not possible to visualize only a filter. Work around : select also an other
material.
• In Lost foam calculations, one see non-zero velocities in the mould at step
zero. Just ignore these velocities which are meaningless.
• It is not possible to view the Particle tracing with symmetries (i.e. the traces
are not replicated).
Solver
• It is not possible to fill simultaneously two separated cavities. One should fill
only one cavity at a time or set (by hand in the p.dat file) the Run parameter
COLDSHUT = 0 (this will deactivate the cold shut detection algorithm).
• In models where there are more than one RESERVOIR domain, they should
all touch each other and they should be all full at the beginning. Otherwise,
problems may be encountered.
• PIPEFS must be set to zero if it is to be followed by an Advanced Porosity
calculation.
• It is not possible to model temperatures below -273°C (below absolute zero)
with the THERMAL=2 model. Work around : use THERMAL = 1.
• When a domain defined as RESERVOIR is remaining full during the whole
filling, the Pool model will be activated at the wrong fill fraction (i.e. at
LVSURF minus the volume of the reservoir). Work around : set this domain to
CASTING instead of RESERVOIR, as if this domain remains full, it is useless
to define it as RESERVOIR.
• When the data files are stored on a remote disk (accessed through a network),
it may be possible to have problems with DataCAST on large problems
(DataCAST is in fact running well, however, the g.unf file is not written
properly and thus ProCAST will crash right at the beginning). In such case, it
is worth to run DataCAST on the local disk of the machine and then transfer
the files (or at least the g.unf file) on the remote disk. This is due to the fact
that the writting of large files (with large record length) through a network
may encounter problems. This is a system problem which has nothing to do
with the software. This was observed only when DataCAST was run on
Windows.
• If we have piping which results in a part of the casting being separated from
the main body, this will lead to problems in the stress calculation (i.e. the
stress calculation will diverge). Work around : use the PETSC_SOLVER = 0
option.
MeshCAST
• On some Windows OS, the Open in MeshCAST is not done in the current
working directory.
SOFTWARE MANAGER
FILE MANAGER
When the ProCAST software is started, the Manager is launched, as seen in the
figure below. The Manager allows to browse in the desired working directory, to
create new directories and cases, to copy files from one directory to an other and
to launch the different modules of the software.
The default starting directory can be set in the "Installation settings" (see the
"Software configuration" section).
Firstly, the working directory should be defined (with the "browse" and
"create directory" icons) and secondly, the name of the case (prefix name)
should be defined.
When several cases are present in the same working directory, the prefix of the
first one (in alphabetical order) is selected. In order to scroll through the other
cases, one can use the F2 key.
The File Manager allows to perform some actions on the files, such as "Refresh"
(to refresh the file list), "create new text file" , "Rename file" ,
"Duplicate file" and "Delete file" .
A set of files can be copied from one directory to the other, using the
The extensions "d.dat", "p.dat", "d.out" and "p.out" are well handled by the copy.
However, if it is desired to copy "unf" files, the user is prompted for the "old
prefix" and the "new prefix", as shown below :
In the "Copy" panel, the icon gives a direct access to Template files (for
the user functions and for the porosity input file).
MODULE CALLS
The different modules of ProCAST can be called from the top bar buttons :
Firstly, the FEM mesh should be generated with MeshCAST. Then the case
should be set-up with the pre-processor (PreCAST). Once all the parameters and
conditions are defined, the calculation files should be prepared with DataCAST.
The calculation can be run at this time by calling the ProCAST solver. Finally,
the results can be visualized with the post-processor VisualCAST. At any time
during and after the modeling, the status of the calculation can be monitored with
Status.
The Manager allows also to launch the "2-D mesh generator" or MeshCAST with
the -M option. To do so, a "right click" on the "MeshCAST" button should be
done. This opens a sub-menu, where one can select between MeshCAST-2D or
MeshCAST-3D and MeshCAST-3D with the -M option.
This special mode of MeshCAST should be used only in the case of large memory
requirements (see the MeshCAST manuals for more details).
Once the data are ready (after PreCAST and DataCAST), the ProCAST solver can
be launched, with the "ProCAST" button. The following window is then opened in
order to "Confirm" the start of the calculation. Moreover, it is possible to
automatically launch "DataCAST" before the ProCAST solver, by checking the
"Execute DataCAST first" box.
Upon "Run", a Command Window will be automatically opened and the solver
will be started. As this window is independent from the Manager, it is possible to
use the Manager (and the other modules, such as the post-processing for instance)
while the calculation is running. When the calculation is finished, a PAUSE will
be prompted and one has to hit any key (e.g. RETURN) to make the window
disappear. If there is a problem during the execution, a warning or an error
message may be printed in this window.
The inverse solver can also be called from the "ProCAST" button (see the
Software configuration section for more details about how to activate the Inverse
option):
Results can be viewed at any time (even during the run of the calculation), with
the VisualCAST button. If the case name is set, as well as the current working
directory, the case is automatically loaded in the post-processor, as shown below.
A right click on the "VisualCAST" button gives also access to "ViewCAST", the
"Advanced Porosity" and the "CAFE" post-processors (i.e. the calcosoft
post-processor) :
Please note that VisualCAST should be installed in order to use this button (the
installation of VisualCAST is distinct from the ProCAST installation.
To monitor the calculation, the "Status" button can be used at any time. A
graphical status of the case will be shown with the relevant information
concerning the calculation (i.e. percent filled, solid fraction, timestep, current
time, cpu time, ...) corresponding to the case name indicated in the lower orange
window. The status is refreshed every 15 seconds, unless the "Status" button is
pressed again. To exit the Status window, press the "Return to File Manager"
button. Please note that the Graphical status is working only when activated in the
working directory of the case (as the prefixp.dat file is read).
Finally, all these modules can be called manually from a Command Window. The
"Command Window" button opens automatically a Window in the working
directory.
Then, the modules can be called in the following way (or via ad-hoc aliases) :
For Windows :
Manager %ProCAST20091%\bin\manager.exe
MeshCAST %ProCAST20091%\bin\meshcast
PreCAST %ProCAST20091%\bin\precast prefix
DataCAST %ProCAST20091%\bin\datacast prefix
ProCAST %ProCAST20091%\bin\procast prefix
ViewCAST %ProCAST20091%\bin\viewcast prefix
Manager $ProCAST20091/bin/manager.exe
MeshCAST $ProCAST20091/bin/meshcast
PreCAST $ProCAST20091/bin/precast prefix
DataCAST $ProCAST20091/bin/datacast prefix
ProCAST $ProCAST20091/bin/procast prefix
ViewCAST $ProCAST20091/bin/viewcast prefix
The Volume mesher of MeshCAST can be called in batch mode with the
following command :
On Windows :
%ProCAST20091%\bin\meshcast -batch prefix.sm Flag
To do so, firstly, the Manager should be configured in the following way (in the
"Installation Settings" window.
In the "Preferences" tab, one can select which "modules" should be activated
(under "Modules display") in the Manager :
"Advanced Porosity module" will activate the possibility to launch and visualize
Advanced Porosity calculations (see the "Advanced Porosity Calculations"
chapter for more details).
"Inverse module" will activate the possibility to launch inverse calculations form
the Manager (see the "Inverse modeling" chapter for more details).
"ProCAST Parallel" button (in the "Parallel" tab) will activate the possibility to
launch the parallel solver (see the "Parallel Solver" chapter for more details).
When these modules are activated in the "Preferences", the "ProCAST" window
has the following appearance (more details about the functionnalities in the
corresponding chapters) :
RUN LIST
The "Run List" allows to launch automatically several calculations in a row, as
well as to "chain" different calculations (such as a ProCAST thermal calculation
followed by an Advanced Porosity calculation). This allows also to launch
calculations in batch mode.
When the "Run List" menu is selected the Manager window appears as follows :
If it is create a "Run List" from cases in the same directory, one should check that
the desired case name is specified in the bottom window, then the desired solver
should be selected in the left window (Available Solvers) and the "Add button"
should be pressed. The corresponding run will appear in the right window
(Solvers to be run). Then, one can add as many runs as desired (see example
hereafter where the Case1 and Case2 are launch sequentially).
If the cases to be run are not all in the same directory, one should first select the
Solver to be run (in the left window) and then press the "Folder" button (above the
"Add" button). This is opening a browser, where one can select the desired
directory and the desired case to be run. Once the "Open" button of the browser is
pressed, the corresponding line in the right window is added. With this method,
one can run sequentially a serie of calculation which are located in different
directories.
It is also possible to remove an entry from the right list, by selecting it and
pressing the "Remove" button.
Once the list is ready, the "Run Computations" button can be pressed. This will
automatically launch the calculations. Otherwise, it is also possible to save the
"Batch file", with the "Create Batch File" button. This file will automatically store
the batch file in the working directory.
One can also edit the batch file or run it with the corresponding buttons. In these
two cases a browser will open automatically, to select the desired file.
SOFTWARE CONFIGURATION
Preferences tab
The Preferences can defined, such as the Default directory which will appear
every time the Manager is launched, the preferred Text editor can be configured.
The Option "Use the 64Bits executables" shall be used on 64 bits machines (e.g.
Windows 64). This allows to use automatically the 64 bits executables of the
solver and of MeshCAST. If this option is not activated, the 32 bits executables
will be used.
The access to the "Advanced modules" (Advanced Porosity, Inverse, CAFE and
HCS) can be configured in this window (see the "Advanced module calls" section
for more details).
ViewCAST allows a full compatibility between UNIX and Windows results. This
means that it is possible to visualize on a Windows machine, results which have
been calculated on a Unix machine, or vice-versa.
Then, the user can select the desired default mode between the two above choices.
This default mode will be used when the "ViewCAST" button will be pressed.
However, it is still possible to use the other mode (i.e. Swap format), by making a
right click on the "ViewCAST" button. Then, one can chose between the "No
swap format" (which means that ViewCAST and the ProCAST run were
performed on the same platform) or the "Swap format" (which means that
ViewCAST and the ProCAST run were performed on a different platform).
Parallel tab
The Parallel Tab allows to define the DMP settings (see the "Parallel Solver"
chapter for more details).
Installation tab
In the "Installation" tab, one can check that the environment variable
ProCAST20091 is well configured. It should point towards the v2009.1 directory,
above the "bin", "dat" and "lib" directories.
The "Command window Launch Option" check box (in the bottom of this
Installation Settings window" allows to activate the Command windows when
PreCAST and ViewCAST are launched. This mode should be activated in case of
problems in order to see the possible error messages which may occur. In normal
situation, this mode should be deactivated.
The "DOS Path Converter" (on Windows only) can be useful to convert "long
directory names" into short "DOS" ones.
In ViewCAST, the time is displayed at the bottom of the screen. By default, the
time is shown in seconds. However, in the case of large casting modeling, it may
be useful to see the time in hours:min:sec, whereas for high pressure die casting,
milliseconds are preferred. The default settings can be set in the file
"manager.tcl", which is located in the "bin" directory of the installation.
At about 30 lines from the top of the file, there is the following lines :
### Viewcast ###
global viewcastOption
set viewcastOption "-T1"
CUSTOMIZED INSTALLATION
From version 2004.1, ProCAST allows to install the software in a Read only
location and to customize the location of the files which have to be modified by
the user (databases, status files, preferences).
In the same way, if a user would like to set it own preferences and status files, one
should copy the "pref" and "statpro" directories in the desired location (e.g.
/my_pref_location/pref and /my_pref_location/statpro). Then, the
ProCAST20091_USER environment variable should be set (manually) to the
/my_pref_location directory.
FLEXLM
For "Node locked" licenses, the following procedure can be used (for floating
licenses, please refer to the general FlexLM documentation).
The following is valid for Windows only. For Unix, please refer to the FlexLM
documentation.
In order to activate your license, a license file should be requested to ESI. The
first step is to identify the machine on which the software has to run. To do so, run
the "FlexLM License Manager", by a double click on the corresponding icon in
the "PAM-SYSTEM" directory which is on your desktop.
This will open the following panel. Go to the second tab "System Settings" (1)
and send this panel to your local distributor. You can also transform this
information in a text file, using the "Save HOSTID Info to a File" button (2).
Then, the licence file which is received from ESI should be called "pam_lmd.lic"
(usually, it is called "*****.cry" and you should rename it to "pam_lmd.lic" -
previously it was called "license.dat"). Then, the "pam_lmd.lic" file should be
placed in the C:\flexlm directory.
The location of the "pam_lmd.lic" file should be specified. The "Browse" button
could be used to find the "C:\flexlm\pam_lmd.lic" file :
When the "Finish" button is pressed, the software is unlocked and all the modules
can be used.
GETTING STARTED
SOFTWARE LAUNCH
To launch the software, double click on this icon and the ProCAST Manager will
start.
The "on-line help" can be opened with the "Help" button (1).
The working directory should be set with the browse tools (2) .
The different modules can be called with the upper buttons (4).
PROBLEM SET-UP
In order to get started, a very simple thermal problem will be set-up, step by step.
To do so, a very simple mesh located in the installation directory will be used.
Please note that the meshing, described in the MeshCAST manual will be
addressed in this section (please refer to the MeshCAST manual for more details).
On Windows : %ProCAST20050%\Test\Geometry\Test.mesh
On Unix / Linux : $ProCAST20050/Test/Geometry/Test.mesh
Software launch
This "Test.mesh" files should be copied in the desired working directory
(C:\Temp\Start in the following example). Then, the software manager should be
launched (see previous section) and one should browse to the working directory
and set the Case name (prefix) to "Test".
Pre-processing launch
When the "PreCAST" button is pressed (in the upper black band), the
pre-processing module opens and the mesh ("Test.mesh" file) is automatically
loaded.
A window with some information about the loaded model (number of domains, of
nodes and elements, as well as the model size) appears. Any click on the menus
will automatically close this window. The geometry can be rotated interactively
with the mouse. The icon can be pressed in order to view the geometry in
"Solid" mode.
Materials menu
The first operation to perform is to assign material properties to the two domains.
The first domain should be selected (1) and the desired material properties should
be selected in the database (2). An "H13-steel" is selected in this case. Then the
"Assign" button should be pressed in order to assign the selected material
properties to the given material domain (3). The material property assignment
should be repeated for the second material domain (let's select the
"AL-7%Si-0.3%Mg-A356" alloy.
Finally, the material type ("Mold" for the first domain and "Casting" for the
second domain) should be specified (4). This is done with a right click in the
"Type" column.
Interface menu
In the interface menu, one should create the interfaces between the different
material domains and assign the desired interface heat transfer coefficients.
The type of interface should be specified. To do so, one should click once on
"EQUIV" (1), which will change into "COINC" (for coincident interface). The
"Apply" button (2), which is turning orange should be pressed to accept the
coincident interface type. A message will appear on the screen, which should be
acknowledged. The desired interface heat transfer coefficient should be selected in
the database (3) and assigned to the corresponding interface (4).
The type of boundary condition should be selected with the "Add" button (1). The
"Heat" type should be selected in the list which is appearing. Then, this "Heat"
boundary condition appears in the BC-Type column (2). A "Heat" boundary
condition allows to define a heat transfer coefficient between the model and the
ambient air, as well as the air temperature. The corresponding values are selected
in the database (3) and the "Assign" button is pressed (4) to link the Heat
boundary condition to the database values. The database entry number is then
indicated in the "DB entry" column (5). This boundary condition should be then
applied to the model. To do so, the "Select all" icon should be pressed (6) and all
the model is colored in red (7), showing that all outside surfaces were selected.
Finally, the "Store" button, which was turned to yellow, should be pressed in order
to save the selection (8).
Process menu
The gravity should be defined in the "Process menu". To do so, click on the menu
"Process" and select the "Gravity" sub-menu. This will open the "Gravity" panel.
A gravity in the -Z direction is set and the panel is closed with the "Close" button.
Each domain should be selected (1) and the initial temperature should be entered
in the field in (2).
In this example, all the default values can be used. One could just modify the
value of TSTOP (in the "General/Standard" tab) to 500°C (in order to stop the
calculation when all the temperature are below 500°C).
The pre-processing stage is finished and one is ready to launch the calculation.
CALCULATION
The calculation can now be launched in the Manager. To do so, the "ProCAST"
button should be pressed (1), which will open the confirmation window for the
run. The "Execute DataCAST first" should be activated (2) and then one can
confirm the launch of the solver by pressing the "Run" button (3).
On Windows, a Command window will open and the DataCAST and ProCAST
are automatically launched. A PAUSE message is displayed when the calculation
is finished.
Once the calculation is finished, or anytime during the calculation, one can view
the results in the post-processor ViewCAST.
RESULTS DISPLAY
Once the ViewCAST window is opened, one can interactively rotate the mouse
(with a left click in the graphics area (1)). The material domains to be displayed
can be selected with the icon (2). The result to be displayed (i.e. Temperature,
Fraction of Solid, ...) should be selected in the "Contour" menu (3). In order to
display a snapshot, the icon (4) should be active. Then, one can display the
different timesteps with the tape player (5), either as an animation or as a
step-by-step display. The scale can be changed by a click on the scale values (6).
Slices through the model can be obtained with the Slice icon (1). When no slices
are defined, it will automatically open the slice definition window (2). Then, one
should select the "Add ->" button and the "XYZ Plane" sub-menu, which will
open the slice selection panel (3). When the desired slice is selected, the Apply
button should be pressed and the slice will be displayed. One can select more than
one slice by repeating the "Add ->" operation. Finally, one can show the different
timesteps with the Tape player (4).
Time evolutions can also be shown with the "XY Plot" menu. The desired field
(e.g. Temperature) should be selected and the desired locations to be plotted
should be defined (e.g. with the "Interval" sub-menu (a value of 50 nodes interval
was selected in the figure hereafter).
To Quit ViewCAST, the File/Exit menu should be used. The software is asking
whether you want to store the last view. If it is stored, the current position and
orientation of the model on the screen, as well as the material selection, will be
retrieved at the next launch of the post-processor.
PRE-PROCESSING
INTRODUCTION
To start the Pre-processor, the "PreCAST" button should be used. If a mesh file
(case.mesh) or a "d.dat" file (cased.dat) is present, the case will be automatically
opened.
If there is no case present in the working directory, then the browse window will
open so that the user can select the desired input files (see the Geometry import
section).
When the case is loaded, a window appears with some information about the
model, such as the number of materials, the number of nodes and elements, as
well as the model size. Then, the pre-processor is ready to set-up a case.
The top bar menu is divided in 9 menus which allow to perform all the operations
to set-up a case :
• File
• Geometry
• Materials
• Interface
• Boundary Conditions
• Process
• Initial Conditions
• Run Parameters
• Inverse
• Help
First the model should be opened or saved in the File menu. It allows also to quit
the Pre-processor.
Then, in the Geometry Menu, symmetries can be defined, as well as the virtual
mold characteristics. Moreover, some features of the FEM mesh can be checked,
such as negative Jacobians, or the volume of each domain.
In the Material menu, the characteristics of each domain (or each material) can be
defined. In addition to the material properties, one can specify the type of the
domain (casting, mold, filter, foam, ...), as well as if it will be empty or not at the
beginning of the calculation (for mold filling).
The Interface menu has no sub-menus. It opens the window which allows to
define the interactions between the different materials, such as heat transfers.
The Boundary condition menu allows to define all the interactions between the
different materials and the outside world (i.e. on the outside surfaces of the
model), such as external cooling, velocities at the surface of the model for flow
calculations, displacements or constraints for stress calculations, etc...
The Process menu gives access to the definition of the gravity, as well as the
definition of the motion of the different domains or enclosures.
The initial temperatures of each materials are defined in the Initial Conditions
menu.
The Run Parameters menu, as well as the Help menu have no sub-menus. All
the calculation parameters are defined in the Run Parameters window. The on-line
Help can be access from the Help menu.
Below the menus, icons allows to perform a number of operations linked to the
display of the model on the screen. These icons are described in the Geometry
manipulation section.
The next sections are presenting the set-up of a case, according to the following
flow chart.
GEOMETRY IMPORT
If there is no mesh file or d.dat file present in the working directory with the
corresponding prefix, the pre-processor opens with the browser window.
Then the user has the choice of the input format, through the following filter :
The Pre-processor is able to read PreCAST input files (*d.dat), also called
"Restart" files, or meshes coming from MeshCAST (*.mesh), from PATRAN
(*.out) or I-DEAS (*.unv).
THERMAL
Thermal model
The Thermal module allows to perform a heat flow calculation, by solving the
Fourier heat conduction equation, including the latent heat release during
solidification. The typical results which can be obtained are the following :
• Temperature distribution
• Fraction of solid evolution
• Heat flux and thermal gradients
• Solidification time
• Hot spots
• Porosity prediction
Flow chart
This section describes the set-up of a thermal case. It is also the opportunity to
introduce the general work flow of the pre-processor, as well as some aspects
which may be used by different modules (e.g. symmetry).
Each step described in the above flow chart is described in the following sections.
Geometry assignments
Once the geometry is loaded, the following operations can be performed on the
geometry :
Symmetries can be defined at this stage (see the "Thermal/Radiation" section for
more details). The definition of the Virtual mold is also done at this level. (see the
"Virtual Mold" section for more details).
Volumes gives access to the volume of each material domain, whereas Min-Max
indicates the dimensions of the model.
Materials assignment
Once the model is loaded (see the "Geometry import" section), the first operation
is to define the different materials with their properties and attributes. This is
performed in the Material/Assign menu.
On the right of the window, two frames are shown. The top one contains the
material list (or domain list), whereas the bottom one corresponds to the material
database.
When one clicks on the different materials in the material list, the corresponding
domains are highlighted (in the picture below, the hidden mode with mesh
was selected - see the "Geometry manipulation" section for the other display
modes).
In the lower frame ("Material database" list), the list of all available material
properties in the material database is displayed. To manage the database entries,
please refer the "Databases" sections. The {T} or {F}which are indicated before
the material name are telling wheter material properties are present in this material
for Thermal only calculations (T) or for Thermal and Fluid flow calculations {F}.
If a {*} appears, it means that the material properties definition is uncomplete and
that this material entry can not be used for a calculation at this stage.
In the top frame ("Domain list"), all the domains (or materials) present in the
mesh are listed. When a mesh is loaded, it appears as follows :
Then, one should assign a Material to each domain, to define the type of each
domain and to specify whether the domain is empty or not at the beginning of the
calculation.
To assign Materials, (1) one should select the desired domain in the upper list, (2)
select the desired material in the material database list, and (3) click on the Assign
button. This should be repeated for each domain.
Then, the "Type" of each material should be defined. To do so, make a right click
on the "CASTING" word and the available list of possible selection will appear :
Mold : the mold material should be set to "Mold". This will be used for cycling
calculation (in die casting) in order to allow the calculation of the heating of a die
during cycling (i.e. the temperature of the mold domains will not be reset to the
initial temperature at the beginning of each cycle).
Casting : the casting material should be set to "Casting". This setting is necessary
in particular for all the domains where fluid flow will occur. For a cycling
calculation, the casting domains initial temperatures will be reset at the beginning
of each cycle.
Filter : filter domains should be set with the "Filter" type (see the "Filters" section
for more details).
Foam : for lost foam calculations, the domains where the foam is present at the
beginning of the calculation should be set to "Foam". Of course, during the filling,
the casting material will replace the foam, as it burns.
Insulation : this type has no specific effect on the solver. It will correspond to a
"Mold" type of material. At this moment, this is for information purposes.
Exothermic : this will activate the Exothermic properties of the sleeve (if they are
defined in the corresponding material properties). If the material properties are
containing the exothermic information, but the "Exothermic" type is not activated,
the exothermic model will not be activated (see the "Exothermic" section for more
details).
Core : a core type material should be defined in the case of cycling, where cores
are placed into the mold at each cycle. This means that unlike mold materials, the
initial temperature of the cores will be reset at the beginning of each cycle.
Reservoir : a Reservoir type material is a domain where the free surface will
always be perpendicular to the gravity. This allows to simplify the free surface
computation and it is especially useful in the case of tundish modeling. The
"RESERVOIR" domains should have an "EQUIV" interface with the other
CASTING materials.
Please note that if there is more than one RESERVOIR domain, they should all
touch each other and they should be all full at the beginning. Otherwise, problems
may be encountered.
If one want to empty the reservoir, no special BC should be specified. If one
would like to keep the reservoir full, a pressure BC should be set at the surface of
the reservoir.
Finally the user has to specify which domains are empty at the beginning of the
calculation (for mold filling calculations). One should make a left click on the
"No" to turn it to "Yes" (which means that Yes the domain is empty). On
additional click returns to No. Of course, more than one domain may be empty (if
the casting is made out of several mesh domains).
When one makes a right click on the material name in the upper list, the lower list
is pointing on this particular material. This is very useful if one wants to see the
material properties of this material.
Interfaces assignment
Once the Materials are defined, one should define the Interfaces, with the
Interface menu.
As for the Material window, two frames appear on the right of the window. The
top one contains the list of all the possible interfaces, whereas the lower one
shows the Interface database. To manage the database entries, please refer the
"Databases" sections.
Firstly, one should define the type of interfaces in the upper right window :
On the left, the "Material Pair" are shown. "1 and 3" means that there is an
interface between material 1 and material 3. By default, the Type of the interface
is set to "COINC". By clicking on the "COINC" text, one can toggle between
"COINC", "NCOINC" and "EQUIV".
When a mesh is generated with MeshCAST (or with most common mesh
generator), the elements which are on either side of an interface (i.e. adjacent
elements which belongs to two different domains) are sharing the same nodes.
This is called a coincident mesh.
EQUIV option
When two domains are part of the same entity (i.e. they both belong to the casting
with the same material properties, but they were meshed separately for technical
reasons), one will set an "equivalenced" interface between them (EQUIV). It
means that there will be a continuum between the two domains, with a continuous
temperature profile across the interface, as well as continuous velocity field. In
such a case, the nodes at the interface (shown in orange in the figure below) are
shared by the elements on both sides. This EQUIV option can also be used if one
has different materials in the two domains, but the materials are welded together
(i.e. with a total bounding between the two materials).
COINC option
At an interface between two different materials, such as the casting and the mold,
there is usually a temperature drop. In this case, the nodes at the interface should
be doubled (for a coincident interface), in order to distinct temperature on each
side of the interface. As during the mesh generation, there is one node at the
interface, it is necessary at this stage to duplicate all the interface nodes (as shown
in green in the figure below). This duplication operation is performed when
"COINC" is selected (for "coincident nodes"). The interface, which is shown in
yellow in the figure below has in fact a zero thickness.
NCOINC option
It is also possible to generate a non-coincident mesh (i.e. where the elements on
both sides of the interface are not matching, which means that they are not sharing
the same nodes), by adding different meshes together (see the "Advanced
features" section for more details on non-coincident meshes). In this case, one has
to specify that the interface is non-coincident, with the "NCOINC" option.
When one toggles between the different options, the interface appears in red and
green. It is thus possible to well identify whether it corresponds well to the desired
interface (see figure below, which was obtained in hidden mesh mode )
An other way to view the desired interfaces is to click on the "Material Pair" and
the material on both sides will be highlighted in red and green respectively - the
first material in the list is in red and the second one in green (see figure below,
which was obtained in hidden mesh mode )
Once the desired selections (between COINC, EQUIV and NCOINC) are done for
each possible interface, the STORE button (which is highlighted in orange) should
be pressed.
Then the pre-processor will automatically create the double nodes and a message
will appear to confirm that the number of nodes of the model has increased.
Once the types of interfaces are defined, one has to apply the corresponding heat
transfer coefficients (for COINC and NCOINC only, as nothing as to be specified
for EQUIV). To assign Interface heat transfer coefficients, (1) one should select
the desired Material Pair in the upper list, (2) select the desired interface heat
transfer coefficient in the interface database list, and (3) click on the Assign
button. This should be repeated for each coincident or non-coincident interface.
After the definition of the interfaces, the Boundary conditions should be specified.
This is done in the "Boundary Conditions" menu.
Assign Surface
The principles of Boundary conditions definition are described in the following
figure.
Firstly, the desired boundary conditions should be "Added" (1) in the upper list
(2). With the "Add ->" button, one has the following choices :
Then the "location" where the boundary condition should be applied on the
geometry should be specified. The selection tools (3) allow to "paint" the desired
area on the geometry (4).
It is possible to have a quick access to the database entries in the following way.
When one makes a right click on the boundary condition entry in the upper list
(2), the lower list is pointing on this particular boundary condition entry in the
lower list (5).
For some boundary conditions, such as "Heat" or "Symmetry", faces of the Finite
Element Mesh are selected. For other boundary conditions, such as
"Temperature", "Velocity", ... the boundary conditions are applied on nodes. The
choice between faces or nodes is automatic.
Select all.
Deselect all.
Select and propagate. All the faces or nodes which have an angle with the
neighbors smaller than the specified propagation angle will be selected.
Deselect and propagate. All the faces or nodes which have an angle with
the neighbors smaller than the specified propagation angle will be deselected.
Select remainder. The remaining faces or nodes which have not yet been
selected are selected.
Select interface. For "Heat" boundary conditions, the selections are applied
on external faces only. However, for cycling, one would like sometimes to apply a
Heat BC on faces which lie at interfaces. This allows to select those interfaces
automatically. In this case, the following panel is opened. It is proposing the list
off all active materials. One could select one or more material and all the faces of
the selected materials which are lying on an interface will be selected.
Copy of selection. All the nodes or faces which are applied on the geometry
for the active boundary condition will be copied in the memory.
button is pressed). Please note that one can not copy the node selection of a
Temperature BC to a Heat BC (as faces are expected).
Assign Volume
In the "Assign Volume" menu, the following entities can be assigned :
Volumetric Heat corresponds to a "heat source" in the entire volume (see the
"Boundary Conditions Database" section for more details).
Surface Heat allows to specify how the free surface inside a given material
domain is cooling down. It is possible to assign a "Heat" boundary condition to
the free surface inside the domain. With this, it is possible to assign a convective
heat transfer coefficient and an external temperature which will drive the cooling
condition of the free surface itself. For instance, this could be important to take
into account the cooling of the free surface of a swimming pool. Please note that it
is not possible to activate the radiation model with view factors on a free surface.
Filter Heat allows to define the interface heat transfer coefficient between the
liquid metal and the filter material. Thus, the cooling of the hot liquid, when
passing through an initially cold filter can be modeled using a "Filter heat". The
interface heat transfer coefficient (which is the same as the one defined in the
"Interfaces assignment" section) is applied to the entire filter domain.
Assign Enclosure
In the case of radiation problems with View Factors, the ambiance or the furnace
can be modeled with an enclosure. See the "Pre-Processing/Radiation" secton for
more details. This menu is not used for Thermal problems without "View Factor
radiation".
For Thermal problems (as well as for flow), the gravity vector should be defined
in the "Process" menu.
Gravity
For Thermal only problems, it is important to define the gravity direction for the
calculation o of the porosity (using the POROS=1 model).
The gravity is defined in the following panel. For standard problem, a constant
gravity (1) is defined. The three components of the gravity are defined in (2). If
one clicks on the X, or the Y, or the Z letter, automatically, the gravity vector is
set in the X-, Y- or Z- direction. With two clicks, the negative direction will be
set.
The "Rotate" is used only for Fluid flow problem (see the "Fluid Flow & Filling"
section for more details).
Assign Volume
This option is used to specify a translation, a rotation or a revolution to the
material domains. As this is mainly used in the case of Radiation problems, please
look for more details in the "Pre-processing/Radiation" section.
Assign Enclosure
This option is used only for radiation problems (see the
"Pre-processing/Radiation" section for more details.
The initial temperature of each material should be defined in the initial condition
menu.
The initial temperature can be defined either as a Constant value throughout the
material domain, or as an Extracted temperature field coming from a previous
calculation.
In the case of "Constant", the list of all the Material domains is displayed and one
can enter the initial temperature (in the lower white field) for each one.
Please refer to the "Thermal Run Parameters" section for the full description of all
the Run Parameters.
• Filling behavior
• Free surface evolution
• Natural and forced convection currents
• Dynamic pressure of the liquid
• Entrapped gas
• Filter behavior
Flow chart
This section describes the additional set-up necessary for fluid flow and mold
filling. For the set-up of a thermal case, please refer to the "Thermal" section of
the Pre-processor.
Materials menu
Concerning the material assignments, the only requirement is that the domains in
which fluid flow calculation will be performed are defined as "Casting" domains
(or "Filter" or "Foam" - see the "Advanced features" section for more details).
Then, one should define whether the corresponding domain is Empty or not (in
the case of mold filling). One can change from No to Yes and vice-versa by
clicking on the text directly.
Moreover, the material properties of the fluid material should have the flow
properties (e.g. viscosity). This can be checked in the Material database list. The
Material name should be preceded by an {F} (for Fluid properties).
Interface menu
Concerning the interfaces, nothing special should be done concerning the fluid
flow. One can however notice that if the casting (i.e. the flow domain) is
non-coincident with the mold domain(s), one will need to set a zero velocity
boundary condition all around the casting domain (in order to prevent "leaks").
See the "Advanced features" section for more details on non-coincident meshes.
See the "boundary conditions database" section for more details about the
different type of boundary conditions, as well as the database management. See
the "Thermal/Boundary Conditions Assignment" section for more details about
how to assign boundary conditions.
Process menu
For fluid flow problems, the gravity has to be defined (see the "Thermal/Process
Conditions Assignment" section for more details).
The fluid flow module should be activated with FLOW = 1 or 3, as well as the
free surface model (FREESF) for mold filling and the gas model (GAS). The
storage frequency of the fluid flow results (VFREQ) should be specified. Then,
information about the reference pressure (PREF) and pressure driven inlet
(PINLET), about the final fill fraction (LVSURF) and the filling parameter
(WSHEAR, WALLF) shall be defined.
Please refer to the "Flow Run Parameters" section for the full description of all the
Run Parameters.
RADIATION
Radiation model
The radiation module allows to perform complex radiative problems (e.g.
investment casting), with the calculation of the shadowing effects (view factor
calculations). The typical results which can be obtained are the following :
As Radiation with view factors involves the calculation of the interaction of the
components (casting and mold) with the environment (furnace, castshop, ...), it is
necessary to include the environment into the model. This is done with an
"Enclosure".
If the casting is put into a furnace, the Enclosure is the furnace itself (or the inner
skin of the furnace). However, it the casting (and mold) is sitting on the floor of
the castshop, one should set an "artificial" enclosure which will surround the
casting and which will have the same effect as the environment.
An enclosure can be either a solid (represented by a solid 3-D FEM mesh, as the
casting, mold, etc...), or by a closed surface (represented by a closed FEM
surface mesh).
The figure below shows a casting within a solid enclosure (for symmetry reasons,
only a sector is modeled).
The figure below presents a casting within a surface enclosure shown in grey (for
symmetry reasons, only half of the geometry is modeled)
Flow chart
This section describes the additional set-up necessary for radiation calculations.
For the set-up of a thermal case, please refer to the "Thermal" section of the
Pre-processor.
a) the enclosure is built into MeshCAST, together with the solid mesh. In this
case, the enclosure should be "tagged" as an enclosure.
Geometry/Symmetry menu
Firstly the symmetries (if any) should be defined. ProCAST is able to deal with
symmetry implying one mirror, two orthogonal mirrors, a single rotation of n
sectors and a combination of them.
In order to illustrate the different possibility of symmetry, consider the simple but
explicit example of two concentric cylinders (the inside cylinder is the casting and
the outside cylinder is a solid enclosure). The full geometry is shown in the figure
hereafter.
Two examples of simplified geometry by a rotation (left: 8 sectors, right: 4 sectors) and
one mirror.
• A rotation of n sectors (R) associated with two orthogonal mirrors (M1 and
M2)
Then, the "Apply" button should be used to validate the symmetry definition. To
disable a symmetry, just uncheck the corresponding check box.
The "Get Co-ord" button allows to pick nodes of the FEM mesh for an interactive
definition of the mirror planes or rotational axis. To use it, first click in the X
coordinate box of the point which should be defined interactively, then click on
"Get Co-ord" and finally, click on the desired node on the geometry. The
corresponding coordinates will fill automatically the corresponding fields. Repeat
it for the other points. Please note that one should click very close to the desired
node (otherwise, it may be possible that a node "behind" is selected).
When the symmetry planes are defined in the "Geometry/Symmetry" menu (see
above), it is creating automatically the corresponding "Symmetry" Boundary
condition (in the "Boundary Condition/Assign Surface" Menu). However, if some
nodes are not lying exactly on the symmetry plane, this boundary condition could
be either empty or some nodes (and thus elements) will be missing. In the
following figure, such a case is shown :
This is due to the fact that three nodes are not exactly on the plane as shown on
the following figure (one node is inside and two are outside the plane).
If the nodes are not too much outside the symmetry plane, PreCAST is able to
move automatically those nodes in order to put them back onto the symmetry
plane.
If the nodes have a small offset and if the movements of these nodes is not
corrupting the adjacent mesh (e.g. by creating negative jacobians), the operation
will be successful and the following message (The symmetry node alignment
check is OK) will be displayed.
If the nodes have a too large offset with respect to the symmetry plane or if the
mesh is becoming corrupted (negative jacobians) during the node alignment
operation, the following message will be displayed.
In such case, the mesh should be changed in the mesh generator and the model
set-up should be started again.
The enclosure can be divided in different sets (1) (in order to apply different
temperatures and/or emissivities). One can Add new sets (2). Then, using the
selections tools (3) (see the "Pre-processing/Thermal" section for the full
description of the selection tools), the desired surfaces of the enclosure can be
"painted" (4) and stored (5). Finally, the desired temperature and emissivity can be
selected in the database (6) and assigned (7). Please note that one should assign to
each set both a Temperature entry and an Emissivity entry.
In order to compute well the View Factors of the enclosure, one should make sure
that the surfaces of the enclosure (i.e the triangles or the quadrangles) are well
oriented (i.e. the surfaces are pointing inwards). To do so, the following icons are
available next to the selection tools :
To view the face orientations, one should first select the desired "Enclosure set"
and then press the "View" icon. The arrows are drawn for the selected set as
shown in the figure hereafter.
If all the arrows are pointing outwards, the "Reverse" icon allows to reverse the
orientation of all arrows.
If only a few arrows are pointing in the wrong direction, one should use the
"Align" icon in order to have all the arrows pointing in the same direction. Then,
one may need to use the "Reverse" icon to point the arrows inwards.
Process menu
The process menu allows to define the motion (if any) of the enclosure (Assign
Enclosure) with respect to the casting or of material domains (Assign Volume).
Moreover, it allows to define the gravity vector (see the "Thermal/Process
Conditions Assignment" section for more details).
When "Assign Enclosure" is opened, the following panel appears on the upper
right corner of the window. Each enclosure set is displayed and three database
entries are possible. The first column corresponds to a "Translation", the second
one to "Rotation" and the third one to "Revolution". If a "*" appears, it means that
no motion is defined for this set. Otherwise, the number indicates the
corresponding database entry.
In the case of "Assign Volume", the same type of panel appears, but instead of the
enclosure sets, the different material domains are listed. Again in this case, it is
possible to specify a Translation, a Rotation or a Revolution. Any combination of
the three is possible, however, please note that the user should check that there is
no conflict between these motions and prevent any inter-penetration of the
different materials and/or enclosures.
On the bottom right, the motion database (Process database) appear (see the
"Process Database" section for more details).
Please refer to the "Radiation Run Parameters" section for the full description of
all the Run Parameters.
STRESS
Stress models
The stress module allows to perform a thermo-mechanical calculation. The typical
results which can be obtained are the following :
• Stress distribution
• Deformations (elastic and plastic)
• Displacements
• Gap formation
• Elastic springback
• Die fatigue
• Hot tears
• Cracks
In order to address these different aspects, six different stress models are
available.
• Linear Elastic
• Elasto-plastic
• Elasto-viscoplastic
• Visco-elastic
• Rigid
• Vacant
Flow chart
This section describes the additional set-up necessary for stress calculations. For
the set-up of a thermal case, please refer to the "Thermal" section of the
Pre-processor.
Materials menu
The stress properties should be assigned in the Materials/Stress menu.
This is opening the following window. Firstly, each domain should be selected
(1). Then, the desired properties should be chosen in the database list (2) and then
the "Assign" button (3) should be pressed in order to link the material with the
corresponding stress properties.
The stress properties, as well as the different stress models are described in the
"Stress database" section.
Interface menu
Nothing specific should be defined at the level of the interfaces.
One should be careful not to over constrain the model with too many
displacements boundary conditions. If this is the case, this may induce artificial
stresses locally, as shown in the figure hereafter.
One should be aware that the application of only a "Surface load" on the top die
(as shown below) is not enough.
This can be explained with the following sketch.. In the left case (blue), only two
loads are applied to the part. This part is not constraint and it can move in any
direction. In order to make sure that the part will not move, one has to apply in
addition the appropriate zero displacements (right case in green). In the center
case (orange), the displacement constraints are not sufficient and the part could
rotate around the bottom right point.
Finally, "Point load" can be applied locally to model the effect of a local force.
Symmetries
Symmetries can be used for stress calculations. However, one should be very
careful to have meshes which are exactly matching the symmetry plane. If this is
not the case, it will create artificial constraints which will degrade the quality of
the results as well as the convergence of the calculation. See the
"Pre-processing/Radiation/Check of the symmetry planes" section for more details
about how to check the symmetry planes.
The stress module should be activated with STRESS = 1. The storage frequency
of the stress results (SFREQ), as well as the calculation frequency of the stresses
(SCALC) should also be specified.
Please refer to the "Stress Run Parameters" section for the full description of all
the Run Parameters.
stresses are relaxed. As the elastic strain is released, the shape of the part is
changing (on the right picture, both the part and the mold are represented,
although the mold is not anymore present).
To model the Elastic Springback effect, the user just needs to define a function for
the interface heat transfer coefficient (at all the interfaces which will not be
anymore in contact when the mold opens). One should define a zero interface heat
transfer coefficient when the mold opens and automatically, the mechanical effect
of the mold will be removed in the model.
However, if one want to continue to simulate the cooling of the casting, while in
the air, the user should not forget to set a Heat boundary condition on the casting
surface (and eventually on the inside mold surface) with a non-zero value from the
opening time (the technique is the same as the one used in cycling calculations).
DATABASES
Material Database
All the material properties are stored in Databases. This section describes the
database containing the material properties for Thermal and flow calculations. The
Database containing stress properties is described in the "Databases / Stress
Database" section. Moreover, the "Thermodynamic" databases are described in
the "Databases / Thermodynamic databases" section.
In the Material Assignment menu (see "Material assignments"), the content of the
material database is shown (see below). Above the material list, the database
management buttons are present :
Read : the database entry can be read and modified (if the user
has the appropriate rights)
Add : a new entry can be created
Copy : an existing entry can be copied in order to create a new
entry
Del : an exisiting entry can be deleted
Sort : the list of materials is sorted by alphabetical order
Search : a search on the material name can be done
When an existing material is opened (with the READ button), a window appears
with the following content :
Firstly, the material name, the user, as well as the creation date of the material is
shown. Then, the material properties are organized in different tabs, with a
hierarchical structure :
Composition
Thermal
Conductivity
Density
Specific Heat
Enthalpy
Fraction Solid
Latent heat
Liquidus-Solidus
Exothermic
Fluid
Viscosity
Surface Tension
Permeability
Filter
Comments
The yellow tabs indicate that values are defined for the corresponding properties.
The white tab shows the active one. Once the "Thermal" or "Fluid" tab is selected,
a second level of tabs becomes active (and so on). The figure below shows the
definition of the thermal conductivity :
The above figure shows the standard panel for material properties definition.
Firstly, the user has to define whether the material property is defined by a
When a Constant is selected, the user can choose the units and enter the constant
material property in the white field below "Enter".
When a Table is selected, the user can input the temperature-dependant property
in the white field on the bottom left of the screen. Each time a new line is filled,
the graph on the right is updated. In this case, the units for the temperature and the
property can be selected.
Above the table, the buttons allow to erase the whole table, erase only the selected
line, import or export the table. For imports, the table should be in the form of a
text file, with X and Y values on the same line, separated by at least a blank. The
number of lines should not exceed 100. The Export format is the same.
Beside these properties, some of them are defined only by constants, such as the
latent heat, the liquidus and solidus temperature or the filter properties. In this
case, only the corresponding constant(s) should be entered (see figure below).
The following section "Material Properties" describes the different properties and
when it should be defined.
Material Properties
Good material properties are the best base for a good simulation. Properties could
be found in several locations, such as litterature, material suppliers, universities,
web, ...
Thermal problems
No phase change
For thermal problems (with or without solidification), the minimum data which
are required are the following (typically for mold materials) :
Thermal conductivity
Specific heat
Density
When solidification is present (i.e. for casting materials), one should define in
addition the following properties :
Fraction of solid
Latent heat
Liquidus and Solidus temperatures
The latent heat, liquidus and solidus temperatures are defined by constants. Please
note that the liquidus and solidus temperatures should be consistent with the
fraction of solid curve (no consistency checks are performed). The liquidus and
solidus temperatures are used for the porosity models and for the calculation of
the permeability of the mushy zone in the case of flow calculations.
where cp(T) is the specific heat as a function of temperature, L is the latent heat
and fs is the fraction of solid.
As there are two ways of defining the phase change, the software is automatically
detecting if there is a conflict in order to have either :
specific heat
Latent heat
or
enthalpy
In previous versions (v4.x.x and v3.x.x), there was no check to prevent both
definitions. Thus, if a model which was created in a previous versions is loaded
into PreCAST v2009.1, the following warning will be displayed (during the load
in PreCAST) :
The user will need to resolve the conflict, by selecting which data are to be kept
(i.e. either enthalpy or specific heat/latent heat) for each material which has this
duplicate definition.
Density
The density is used in thermal calculation (it multiplies the specific heat, the
enthalpy and the latent heat), as well as in fluid flow calculations and in porosity
calculations. Please refer to the "Porosity models" section for more details about
the density definition.
Viscosity
with :
strain rate
phase shift
n Power law coefficient
a Yasuda coefficient
The above parameters can be defined in the database (as constants or as function
of temperature) as follows :
Surface tension
This option is not described as it has not been validated at this stage. One can say
that in conventional casting processes, the surface tension effects are certainly
negligible in comparison to the simplifications made in the free surface
algorithms.
Permeability
For mold materials (in the case of lost foam), a permeability should be defined. In
this case, one can define a constant or a temperature dependant permeability.
For Filter materials, if the Permeability is defined, it will override the default
permeability calculated from the Filter tab.
Filter
The void fraction (Fv) corresponds to the amount of "porosity" or void inside the
filter. This value is dimensionless [-]. The definition of this value is mandatory in
all cases.
The Surface area (Sa) corresponds to the amount of "interface" between the filter
material and the air (when the filter is empty) per unit volume (see example
below). This value is used for the calculation of the thermal exchange between
the filter and the liquid metal going through, as well as for the automatic
permeability calculation. The units are the reversed of a distance (e.g. [1/m]). The
definition of this value is mandatory in all cases.
The Permeability of the filter (i.e. its resistance to the flow) can be calculated in
three different ways.
a) Automatic permeability calculation
From the Void Fraction (Fv) and the Surface Area (Sa) definitions, the
permeability can be automatically computed (based upon Karman-Cozeny),
according to the following relationship :
This mode is activated if the "Pressure Drop" and "Permeability" tabs are not
defined.
b) Pressure drop calculation
If the "Pressure Drop" tab is defined, then, the permeability is calculated, using
the following values (coming from simple experiments) :
If both the "Pressure Drop" and "Permeability" tabs are defined, the
"Permeability" values are ignored (and replaced by the ones obtained from the
above equation).
c) Specified permeability
See the "Filters" section for more details about the settings of cases with Filters.
Thermodynamic Databases
Material properties, such as the enthalpy curve and the solidification path (i.e. the
fraction of solid curve versus temperature), density, viscosity and thermal
conductivity can be computed automatically from thermodynamic databases.
Fe database:
Al B C Co Cr Cu Fe Mg Mn Mo N Nb Ni P S Si Ti V W
Ni database:
Al B C Co Cr Fe Hf Mo N Nb Ni Re Si Ta Ti W Zr
Ti database:
Al B C Cr Cu Fe H Mo N Nb Ni O Si Sn Ta Ti V Zr
Mg database:
Ag Al Ca Ce Cu Fe Gd Li Mg Mn Nd Sc Si Sr Y Zn Zr
Cu database:
Al B C Cr Fe Mn Ni P Pb Si Sn Ti Zn
The other alloying elements which are not present in this list are not available in
the database and will have no effect on the computed material properties. More
details about composition limitations are given in the Database limitations section.
Then the Base alloy (i.e. Al, Fe, Ni, Ti or Mg) should be entered, as well as each
alloying element with its concentration (in weight percent). Once the chemical
composition is entered, the "Apply->" button is pressed and the "Scheil" or
"Lever" option is selected to start the computation of the fraction of solid and of
the enthalpy.
When the "Back Diffusion" model is used, an average cooling rate (corresponding
to a representative cooling rate of the casting to be modeled) should be specified
in order to determine the amount of back diffusion.
Warning
Please note that the cooling rate should not be set below 0.01 K/s. If a lower value
is needed, please use the Lever Rule model.
For iron and carbon steel, the Lever rule is still recommended.
The main difference between the three models is the shape of the fraction of solid
curve at the end of solidification, as well as the solidus temperature (see figure
below).
The following figure is showing the influence of the different models on the
fraction of solid curve for an A356 alloy (Al-Si7%-Mg0.3%). The curves labeled
0.1 K/s and 1.0 K/s correspond to the "Back Diffusion" model with these cooling
rates. For such alloy, the Lever Rule model will correspond to low cooling rates
(below 0.01 K/s), whereas the Scheil model will correspond to high cooling rates
(above 20 K/s). One can see that the back diffusion has a quite important effect on
the solidus temperature.
On the following alloy (Mar-M 200 Hf), the difference in the solidus temperature
between the Scheil model and the Lever rule is 550 degrees !
In the example above, when the Scheil button is selected, the following curves
appear :
One could see the fraction of solid curve, as well as the fractions of the different
phases as a function of temperature.
In the same time, automatically, the fraction of solid curve, the liquidus and
solidus temperatures, the enthalpy curve, the density, the viscosity and the thermal
conductivity are stored in the database, as shown hereafter (of course the value are
finally stored only when the "Store" button is pressed, before exiting the
database).
Thermal conductivity
Density
Enthalpy
Fraction of solid
Liquidus-Solidus
Viscosity
For some chemical composition, it may happen that the software which extracts
the data from the Thermodynamic database is not able to find the right set of
stable phases at low temperature.
Usually, this does not affect the determination of properties of interest (which are
more near the solidification range). Thus, it is possible to use the calculated values
as such. If such situation occurs, one should check the calculated data in order to
be sure that it covers at least the temperature range of interest.
In very few cases, it is possible that the density calculation does not give relevant
results, as shown in the figure hereafter. In this case, these result should not be
used (or the wrong values should be erased).
As a general rule, if the results are not realistic, it is advised to suppress (i.e.
ignore) the elements which are present in very low concentrations. This is
especially true for traces of Sulfur (S) and Phosphorus (P) in steels, which are
sometimes "corrupting" the results.
At this stage, the Young's modulus, the Poisson's ratio and the Thermal expansion
coefficient can be calculated based upon the phases obtained from the
thermodynamic databases. For Al alloys, the Yield Strength can also be calculated
(this can be done only if the Back Diffusion model is selected).
When the Properties calculation is started in PreCAST with either the Scheil or
the Lever model, the following window appears :
The user has the choice of either not calculate the Stress properties, to create a
new entry in the Stress database or to substitute/Add the data to an existing entry.
If a new entry is created the user has to specify its name (without spaces).
If the user would like to substitute or add the calculated data to an existing entry
of the stress database, one should select the desired entry with the Browse button.
Once these choices are made, the computation can be started (of both the thermal
and stress properties) with the "Compute" button.
Please note that the other Stress properties (i.e. hardening, viscoplastic,...) can not
be calculated at this stage (for Al only, the Yield stress can be calculated). Thus
these properties will remain empty.
The following figures are showing examples of computed Stress data from an
A356 alloy.
Young's modulus
Poisson's ratio
Yield Stress (for Al systems only - available only with the "Back Diffusion"
model)
The Yield Stress can be calculated for Al systems only (it is calculated only when
the Back Diffusion model is activated). Moreover, the Yield Stress calculation is
taking into account the effect of the microstructure (mainly the SDAS) through
the cooling rate. The following figure is showing the calculated Yield Stress of the
A356 alloy, for different cooling rates :
One can see that the effect is very strong as it can change by about 100% between
low and high cooling rates.
Databases limitations
The Computherm databases can be used for the following elements and in the
following ranges. More information can be obtained on the
www.computherm.com web site.
The recommended composition ranges mentioned hereafter are not strict limits.
These are ranges which were extensively tested.
Al database
Developed for Al-rich alloys such as commercial casting and wrought alloys.
Tested with more than 40 commercial Al alloys.
20 Components :
Major alloy elements: Al, Cu, Fe, Mg, Mn, Si, Zn
Minor alloy elements: Ag, B, C, Cr, Ge, Hf, Ni, Sc, Sn, Sr,
Ti, V, Zr
Al 80 ~ 100
Cu 0 ~ 5.5
Fe 0 ~ 1.0
Mg 0 ~ 7.6
Mn 0 ~ 1.2
Si 0 ~ 17.5
Zn 0 ~ 8.1
other 0 ~ 0.5
Fe database
Developed for Fe-rich alloys.
19 Components:
Al, B, C, Co, Cr, Cu, Fe, Mg, Mn, Mo, N, Nb, Ni, P, S, Si,
Ti, V, W.
It was observed that alloy elements which are present is very small
quantities (such as P and S) may cause problems in the phase determination.
As these elements do not affect significantly the material properties
(although it may have important effects in other fields), it is recommended
to remove these elements for the computation.
Mg database
Developed for commercial Mg-rich alloys
17 components:
Mg, Ag, Al, Ca, Ce, Cu, Fe, Gd, Li, Mn, Nd, Sc, Si, Sr, Y,
Zn, Zr
Ni database
Developed for commercial Ni-rich alloys.
17 Components: Al, B, C, Co, Cr, Fe, Hf, Mo, N, Nb, Ni, Re,
Si, Ta, Ti, W, and Zr.
Ti database
Developed for commercial Ti-rich alloys such as alpha, alpha+beta, and beta
alloys.
18 Components: Al, B, C, Cr, Cu, Fe, H, Mo, N, Nb, Ni, O,
Si, Sn, Ta, Ti, V and Zr
Cr, Sn < 5
Cu, Fe, Ni < 3
B, C, H, N, O, Si < 0.5
Cu database
Developed for commercial Cu-rich alloys.
14 Components:
Major alloy elements: Cr Fe Ni P Si Sn Zn
Minor alloy elements: Al B C Mn Pb Ti
Example 1 : Al alloy
Example 2 : Ni alloys
Al alloy
When the Scheil model is used, with the above composition, 10 phases are found
(in addition to the liquid phase) :
AlSi9Cu3Mg0.3
AlSi9Cu3Mg0.3Fe1.3
(the effect of Fe is mainly visible at the liquidus)
AlSi9Cu3Mg0.3Fe1.3Mn0.55Ni0.55Zn1.2
(Mn, Ni and Zn are mainly affecting the second half of the curve)
AlSi9Cu3Mg0.3Fe1.3Mn0.55Ni0.55Zn1.2Cr0.15
(Cr is rising the liquidus temperature from 612 to 640°C)
AlSi9Cu3Mg0.3Fe1.3Mn0.55Ni0.55Zn1.2Cr0.15Ti0.15
(When Ti is added, the Al3Ti phase appears, with a very high liquidus
temperature above 760°C)
The following figure are showing the solid fraction curves for all the alloys
together.
In order to see the effect near the liquidus, the same figure is shown, with a
different vertical scale. Only the first 5% are shown.
One can see very well on this latter figure the effect of Ti. Ti is added in order to
create this Al3Ti phase which is stable at very high temperature, which is acting
as inocculant. The amount of this phase is very small (around 0.5%).
The above example is showing that one should be careful with the use of
Thermodynamic databases. In this case, for instance, it would be advisable to
ignore the Ti for the thermodynamic computation, in order to avoid this
"artificially" high liquidus temperature.
This explains also why there are differences observed between literature values
(measurements) and computed values for liquidus and solidus temperature. This is
due to the fact that such values are measured usually by Thermoanalysis and that
small amount of solid (like the few percents due to Fe, Cr and Ti near the liquidus
temperature) can not be detected. One should note that the usual literature value of
the liquidus for this alloy is 588°C which corresponds to a computed value of
about 5% of fraction of solid. The measurement of the solidus temperature is even
more difficult and thus, it is normal to observe differences.
Ni alloys
Usually, alloy composition is given with a range around the nominal value. The
following example is illustrating the effect of small composition changes within
this range on the value of the liquidus temperature.
The following table is showing the composition of three typical Ni alloys, with the
Low, Nom and High compositions.
This has to be taken into account when comparing experimental values with
computed ones, as well as during the real process, as composition fluctuations can
affect significantly the solidification conditions and thus the defects which may
appear.
Interface Database
Interface
For the definition of the Interface heat transfer coefficient, the following panel is
opened.
The structure of the above panel is described hereafter (see the numbers in the
panel below). Firstly, the type of the database entry is shown on the top of the
panel (1). Then, the entry can be labeled with a Keyword (2) which will help to
recognize it's content for later use. If a constant heat transfer coefficient is to be
defined, it's value should be entered in the corresponding field (3). Then the units
can be defined (4). The available choice of units is proposed when one clicks on
the unit box. If it is desired to define a "Temperature-" and/or a "Time-"dependant
If the interface heat transfer coefficient is non-constant, the value defined in (3)
and the table(s) defined in (5) are all multiplied to get the final value (at a given
temperature and a given time). It is allowed to define only a table without a
constant (in this case, it is like if the constant is equal to 1). If a User function is
specified, neither a constant value, nor a table can be defined.
Die Combo
The "Die Combo" definition is used only in the case of die casting, to handle
automatically the sequence of closed and opened dies (see the "Cycling" section
for more details).
It allows to define :
a) the interface heat transfer coefficient (when the die is closed) in the "Constant"
field and/or as a Temperature-dependant table ("Temperature" button).
b) when the die opens, the interface heat transfer coefficient between the die and
the air is defined by a heat transfer coefficient (Air Coeff) and by an ambient
temperature (Air Temp).
The check-box "Attached until Ejection" will tell whether the switch between the
interface heat transfer and the air cooling will be done when the dies opens or
upon ejection of the part.
The time at which the mold is opened, closed and when the spraying is starting
and finishing can be defined either for each Die Combo entry (to have different
values for each interface), in the above panel, and/or in the "Run parameters", in
the "Cycles" tab, as shown below.
The values define in the Run Parameters will apply to all the Die Combo
interfaces where no values are specified in the Die Combo panel (i.e. if all the
values are set to 0.).
Temperature
"Temperature" allows to specify (or impose) a temperature at a given point. It is
especially used to specify inlet temperatures in filling calculations.
As most of the panels of the Boundary Conditions database have the same
structure, the principles will be described for this one only.
Firstly, the type of the database entry is shown on the top of the panel (1). Then,
the entry can be labeled with a Keyword (2) which will help to recognize it's
content for later use. If a constant value of temperature is to be defined, it's value
should be entered in the corresponding field (3). Then the units can be defined (4).
The available choice of units is proposed when one clicks on the unit box. If it is
desired to define a "Time-"dependant Temperature, the corresponding table can be
defined in the"Time" button (5). Finally, the data are saved with the "Store" button
(6).
The following figure shows the choice of available units (when one clicks on the
unit box).
Finally, when the table is well defined, one should SAVE it (5), before Storing (6)
the database entry.
Heat
The "Heat" boundary condition allows to define the heat transfer between the
outside faces of a given domain and the outside world (air, water, ...). The
following panel is available for Heat database entry definition.
The following equation shows the three possible contributions to the "Heat"
boundary condition (all the contributions are added, although they may not be all
active in the same time or in the same cases).
The first term corresponds to a specified flux. It can be used if a given heat flux
was measured for instance. The second term corresponds to Convective cooling.
This is the most common definition of the cooling of an external face. It is defined
by a heat transfer coefficient with the ambiance and by an external temperature (of
the ambiance). The third term is used at high temperature, when radiation
becomes important. In this case, the transfer is proportional to the
Stefan-Bolzmann constant and the emissivity and the difference of the fourth
power of the temperatures (surface temperature and ambiance temperature).
The above equation contains the four values which are shown in the panel below
(1), such as the "Film Coeff", the "Emissivity", the "Ambient Temp" and the
"Heat Flux". These values can be also "Time" and/or "Temperature" dependant
(2). In this case, the value of the constant (if any) is multiplied by the
corresponding table(s). It is also possible to define the different parameters by the
corresponding "User Functions" (3).
Velocity
The definition of the velocity is made in the following panel. In this case, the three
components of the velocity vector should be defined. If a Time- or a
Pressure-dependant velocity is set, each component of the vector will be
multiplied by the corresponding table(s).
If a gate should be active until a given filling fraction is reached, the "Fill Limit"
slider can be used for that purpose. This Fill limit has not the same effect as the
Run parameter LVSURF (see the "Run parameters" section for more details).
LVSURF allows to stop the filling, but more important it allows to switch off the
Fluid flow solver when the filling is finished. On the other hand, the Fill limit
defined in this boundary condition should be used if multiple gates are present. It
allows to switch off automatically one gate when a given fill fraction is reached an
the other gate will continue until LVSURF is reached.
When an inlet velocity is set to fill a casting, there are situations where the user
would like to know which value of velocity should be set in order to achieve a
given filling time. To do so, a "Velocity calculator" is available.
To use the velocity calculator, the following operations should be done (see the
numbers in the upper figure). Firstly, one should "Add" a new velocity boundary
condition (1) and assign the corresponding nodes on the geometry. Then, open the
"Velocity" boundary condition entry (2) and click on "Velocity calculator" (3).
The "Calculator" window will appear (4) and one should select all the domains
which correspond to the casting (5) - please note that the selected volumes should
be set as EMPTY, otherwise, the volume taken into account for the calculation
will be zero. In the above example, the casting is made of four empty domains.
Then, the user has to select whether he wants to compute the inlet velocity
magnitude or the Filling time (as a function of the other one) with the
corresponding radio button (6). In the case where the Fill time should be
calculated, the user has to enter the inlet velocity (7) - if it is specified in the BC
entry, this value will be automatically used. The Inlet area is automatically taken
as the one corresponding the boundary selection on the model. However, if this
area is different, the user can change it directly in the corresponding field (7).
When the "Calculate" button is pressed (8), the Fill time is automatically
calculated (9).
If the inlet velocity is defined by a time function, the calculated filling time will
take it into account.
If a Fill limit is set in the velocity BC, it will be taken into account in the filling
time calculation.
When the inlet velocity is calculated (based upon the inlet surface and the filling
time, the computed value should then be transferred "manually" in the velocity
BC panel.
One should be aware that the velocity calculator will give accurate results if all the
nodes on the inlet surface are selected. If only a few nodes are selected, the inlet
surface will "overflow" to the neighboring elements and the filling time will be
shorter than the one specified. In order to prevent that, it is advised to design a
small additional volume, corresponding to the jet of liquid entering the casting.
Pressure
The pressure is defined in the following panel, which has the same principles as
the "Temperature" panel described above.
Inlet
Inlet boundary conditions are used to specify in the same time an "equivalent"
inlet velocity together with an inlet temperature. The inlet Flow rate can be time
dependant.
Please note that the direction of the equivalent inlet velocity vector is taken as the
average of the normals over all the nodes of the inlet. One should be careful if
nodes at edges of the geometry are selected as it will have an influence on the inlet
velocity direction.
The Fill time calculator can be used as described in the "velocity BC" section (see
above). In this case, the user has the choice to compute either the Flow rate or the
Fill time (as function of the other value). If the Flow rate is defined by a time
function, it is taken into account in the Fill time computation.
As the density is needed to compute a Flow rate in mass per unit time (e.g. kg/s),
the user has to specify a temperature at which the density is evaluated. This
temperature should correspond to the inlet temperature (if this temperature is
specified in the Inlet BC panel, it is automatically transferred to the calculator).
Turbulence
In the case of a Turbulent model, one can specify the "Turbulent" state at the entry
of the liquid in the calculation domain.
Vent
If the "Gas" model is activated, the air can escape through vents. Vents are applied
on nodes of the casting domains. Each vent is represented by an "equivalent tube"
by which the air can escape. Thus, a vent is described by the diameter and the
length of the "equivalent tube", as well as a roughness and the exit pressure.
Inject
In the case of low pressure die casting (lpdc), it is possible to model the liquid
bath of metal and to apply a pressure on it (in order to drive the metal in the mold)
via injected air. The inject BC allows to specify the amount of injected air. Please
note that this method may be delicate to converge.
Displacement
In the case of stress calculations, the displacement of the model should be
constraint. The Displacement BC allows to define these constraints. Please note
that if a field is not filled, the displacement is not constraint in this direction (e.g.
it is possible to specify only a "0" in the x-field which will mean that the node can
not move in the X-direction, but it is free to move in the Y and Z-directions.
Point Load
For stress calculations, it is possible to apply a load (or a force) at a given
location. The three components of the load should be defined.
Surface Load
In the case of stress calculations, one can define a load applied on a given surface.
In this case, it corresponds to a pressure. The direction of the load is specified by
the three components of the load vector.
Volumetric Heat
In some thermal problem, it may be necessary to define the generation of heat
inside a given material (e.g. to model the effect of an electrical resistance). In this
case, the thermal power which is generated per unit volume should be defined.
This heat source will be applied on the entire specified material.
Momentum Source
In the case of flow problems, one would like to model a source of momentum in a
given material. This could be the case to take into account electromagnetic forces
or the effect of a propeller. In this case, the momentum force vector should be
defined. Please note that this force will be applied in the whole specified material.
Mass Source
In the case of filling calculations, one would like to define the amount of metal
which enters into the material with a Mass source (instead of a velocity or an inlet
BC). In this case, one can specify that a given amount of metal (Flow rate) is
"appearing" at the given location (X, Y and Z coordinates) at the specified
temperature. The metal will "appear" with a zero velocity.
Emissivity
For radiation calculations, the emissivity of the mold, the casting top surface, the
furnace or the enclosure should be specified.
The emissivity should be defined between zero and one.
Process Database
Translation
The translation of a given material domain or of an enclosure is defined with the
following panel. The translation can be defined in three different ways :
x(t) - Translation vector or position (i.e. position vs time)
v(t) - Translation velocity as a function of time (i.e. velocity vs time)
v(x) - Translation velocity as a function of position (i.e. velocity vs position)
When x(t) is defined, it is possible to define the translation vector with user
functions.
The velocity of the domain vs time can be defined with the panel hereafter. The
direction of the velocity vector is defined by the U, V and W components. This
velocity vector can be either constant or it can be multiplied by a Time function.
When v(t) is used, the software is transforming it in x(t) automatically in order to
have a translation position which is independant from the timestep.
The velocity of the domain vs Position can be defined with the panel hereafter.
The direction of the velocity vector is defined by the U, V and W components (it
should be a "unit vector"). This velocity vector can be either constant or it can be
multiplied by a "Distance" function. The "Distance" is the relative distance with
respect to the initial location of the domain which moves.
When v(x) is used, it is mandatory to have a non-zero velocity for the initial
position (i.e. the zero position). Please note that it is also mandatory to have
always non-zero velocities defined (i.e. the velocity should never be zero).
Rotation
The following panel allows to define the rotation of a material domain or an
enclosure. The axis of rotation should be defined by two points.
The rotation is defined in the "Time" button, which opens the following panel, for
the definition of the angle as a function of time.
Revolution
The "Rotation" definition is used when less than one turn is done during the entire
process. When one has several turns, the "Revolution" definition is used. In this
case, the axis of revolution should be defined by two points.
Stress Database
Then, the list of available materials (stress properties) is shown in a window in the
lower right corner (as for the standard material database. The database can be
managed in the same way with the "Read", "Add", "Copy" and "Del" capabilities.
When a new stress database entry is created, the following window appears.
The first thing to do is to select the desired model among the five following
options :
Vacant
Rigid
Linear-Elastic
Elasto-Plastic
Elasto-ViscoPlastic
"Vacant" is used to specify that the domain will not participate to the stress
calculation. Thus, no stress and no strain will be calculated and the domain will
not participate to the contact algorithm (i.e. the domain will not create any
resistance to the neighboring domains). No properties should be defined for a
Vacant domain.
No stress calculation will be done in a "Rigid" domain, however, the domain will
participate to the contact algorithm (i.e. the neighboring domains will not be
allowed to penetrate in the Rigid ones). No properties should be defined for a
Rigid domain.
For the three other models, the following data are required :
The three models, as well as the meaning of the corresponding properties, are
described in the "Stress Models and Properties" section.
Warning
For stress calculations, the thermal material properties must include the phase
change (i.e. fraction of solid curve). Otherwise, no stresses will be calculated in
this material (as the fraction of solid will not be defined, it will be considered as
"zero" and no stress will be calculated).
Beside the Young's modulus, one should define the Poisson's Ratio and the
Thermal Expansion coefficients. The value of the Poisson's ratio is usually around
0.3 for metals.
The figure above shows how to transform a measured "Strain curve" (shown in
blue) in a Secant thermal expansion coefficient. To do so, one needs to define a
Reference Temperature (Tref), shown in the above figure as To, which
corresponds to a zero strain. The "Secant" thermal expansion coefficient at
temperature T2 corresponds to the slope S2 of the line between a zero strain (To)
and the strain at the temperature T2.
The Reference Temperature is defined in the lower part of the "Secant" panel (see
above).
Elasto-Plastic model
For the Elasto-Plastic model, the properties described above for the Elastic model
(i.e. the Young's modulus, the Poisson's ratio and the Thermal Expansion
coefficients) should also be defined.
In addition, one should define the Yield stress and the Hardening coefficient.
The Yield stress corresponds to the stress at which plastic deformation starts. It
can be temperature dependant.
The hardening coefficient corresponds to the slope of the stress-strain curve in the
plastic range. Four different models of hardening are available in ProCAST :
The Hardening can also be defined freely in an ASCII file where one can enter
any "digitized" curve. More information about this possibility is in the "Digitized
Hardening" section.
In order to take into account for the "Kinematic" non-isotropic hardening behavior
(Bauschinger effect), the Amstrong-Frederick model is available :
x is called the back stress. It corresponds to the "movement" of the center of the
Yield surface.
Elasto-ViscoPlastic model
For the Elasto-ViscoPlastic model, the properties described above for both the
Elastic model (i.e. the Young's modulus, the Poisson's ratio and the Thermal
Expansion coefficients) and the Elasto-Plastic model (i.e. the Young's modulus
and the Hardening coefficient) should also be defined.
a) Perzyna
b) Norton
Perzyna model
This model allows to describe the secondary (steady-state) creep with threshold.
Norton model
As there is no threshold value, the data for the Yield Stress and the Hardening will
be ignored.
This model allows to describe both the primary (strain hardening) creep and
secondary (steady-state) creep regimes with a possible threshold.
For the three models, one have (for the plastic model) :
or
Finally, one should however notice that in most cases, the use of an elastic-plastic
model gives very realistic results, without the burden of finding appropriate data
for the visco-plastic behavior.
Annealing
During a stress calculations, at high temperature, the plastic deformation is not
contributing to the hardening. The stress solver was modified in order to ignore
the contribution of the plastic deformation to the hardening, above a critical
temperature (called "Annealing temperature"). As a result, only the plastic
deformation which has occured below the "Annealing temperature" will
contribute to the hardening.
Please note that the "Plastic strain" (which can be viewed in the post-processing)
will be the result of the accumulation of all the plastic deformation, below and
above the Annealing temperature (without reset).
The Annealing model can be used with both the Elasto-Plastic or the
Elasto-ViscoPlastic models.
Please note that the definition of these parameters in the mushy zone will
influence the hot tearing indicator prediction.
The value of the Young's modulus in the mushy zone (below a fraction of solid of
20%) should be set as a constant value corresponding to the value of fs = 20%.
Usually, this corresponds to values between 50 and 500 MPa.
The value of the Yield stress in the mushy zone (below a fraction of solid of 50%)
should be set as a constant value corresponding to 5-10 MPa. One should not set
values below 5 MPa.
If no data are available, the Hardening (in case of linear hardening) can be set to
about 1/20 of the Young's modulus. In all cases, the Hardening should be set to 0
MPa for fraction of solid smaller than 50%.
Remark :
If a tensile test is made on a sample which will be fully mushy at a given time, a
non-zero value of the Hardening should be set for fraction of solid below 50%.
However, this case is never occurring in usual casting processes.
Together with these data, the corresponding Run parameters are recommended
(see the "Stress Run Parameters" section for more details) :
CRITFS = 0.5
CONVS = 0.01 (for more accuracy, it is advised to use a smaller value like 0.001)
PENALTY = 0.01
AVEPEN = 0.1 mm (for large casting, this value can be increased)
SCALC = 5
Digitized Hardening
To activate this mode, one should select the "Table" tab in the "Hardening" tab :
Then, the number of tables specified in the ASCII file should be set (i.e. the
number of different temperatures at which a curve is digitized - two in the
example below).
CURVE 2
POINTS 3
TEMPERATURE 1 823.
0. 9.838
0.01 3.89105e+01
0.05 4.73871e+01
The stress units are defined first (the unit code corresponds to the one of the
ProCAST d.dat file - e.g. 4 corresponds to MPa).
Then, the number of the curve should be specified, followed by the number of
points in the curve.
The temperature at which the curve is defined (with the unit code before it - e.g. 1
corresponds to degrees Kelvin, 2 to Centigrade and 3 to Fahrenheit), followed by
the curve itself (strain - stress).
Please note that only the plastic part of the curve (i.e. the hardening) is defined by
the tables. This means that each curve should start by a zero strain (this is
mandatory) and that the corresponding stress value is the Yield stress at this
temperature. Thus, the Yield stress defined in the "Yield Stress" tab in PreCAST
will not be used in the case of Tables (the values will be ignored).
Each curve can have a different number of points. Above the last point, the Stress
is extrapolated as a constant (i.e. perfect plasticity).
Remark :
Each domain (of the mesh) for which we would like to use the digitized hardening
should have its corresponding prefixssN.dat file. This means that if it should be
used in domains 2, 5 and 8, one should duplicate three times the same file with the
names prefixss2.dat, prefixss5.dat and prefixss8.dat.
These tensile tests should be performed for different temperatures using various
strain rates. To do so, the tensile test should be done at constant RAM speed (in
order to pull at specified strain rates). The stress and strain should be recorded
during these tests. This is usually obtained with a Gleeble machine.
The figure below shows the results of such measurements : at low temperature (T1
- green curve), the curve is independent from the strain rate. Thus, the behavior is
elastic-plastic. At high temperature (T3 - red curves), one can see that the stress
level is depending upon the strain rate and that there is no hardening
(viscoplasticity only), as the curve is horizontal at a given stress level. At
"intermediate" temperature (T2 - blue curves), one has both plasticity (with
hardening) and viscoplasticity (as the curve depends upon the strain rate).
Low Temperature
At low temperature (usually T1 < Tm/3), the tensile test curve is independent
from the strain rate. This means that different tests made at different RAM speeds
are giving the same curve (see the green curve in the figure below).
From this curve, one can get the Yield stress (σo) and the Hardening coefficient
(H).
The value of the viscous parameter (η) should be set to zero in order to "disable"
the viscoplasticity. The Power (n) can be set to any value as it will not be used.
High Temperature
At high temperature (usually T3 > Tm/2), the tensile test curve is dependent from
the strain rate and the stress level is constant (no hardening) after the transition
stage (see the red curves in the figure below). As there is no hardening, the value
of H should be set to zero. Therefore the threshold (σy) is equal to the Yield stress
σo.
In order to obtain the viscoplastic parameters (η and n), one should first determine
the value of the Yield stress (σo), which can be equal to zero. To do so, one
should perform a tensile test at the lowest possible strain rate (possibly zero !) and
see the value of the stress on the plateau. If this value is very low, one can
consider that the Yield stress (σo) is zero. In order to determine this value, one
can also make loading-unloading tests, with increasing loads until plastic
deformation is obtained.
Then, for each measured tensile test curve, the value of (σ- σY)/σ* (which is the
stress level of the plateau minus the Flow stress, σY, which is equal to σo in this
case, as H = 0) and dε/dt should be plotted in a log-log graph.
The value of η and n can be deduced from this graph (see below).
Intermediate Temperature
At intermediate temperature (usually Tm/3 < T3 < Tm/2), the tensile test curve is
dependent from the strain rate and the stress level is not constant (hardening) after
the transition stage (see the blue curves in the figure below).
The hardening (H) corresponds to the slope of the curves (after the transition
stage). In this model, the hardening is the same for all strain rates.
In order to obtain the viscoplastic parameters (η and n), one should first determine
the value of the Yield stress (σo), which can be equal to zero. To do so, one
should perform a tensile test at the lowest possible strain rate (possibly zero !) and
see the value of the stress zero plastic strain. If this value is very low, one can
consider that the Yield stress (σo) is zero. In order to determine this value, one
can also make loading-unloading tests, with increasing loads until plastic
deformation is obtained.
Then, for each measured tensile test curve, the value of (σ- σY)/σ* (which is the
extrapolated stress level at zero strain minus the Flow stress, σY. In this case, σY
The creep tests should be performed for different temperatures using various loads
(stress).
The figure below shows the results of such measurements : at low temperature (T1
- green curves), the curves are independent from time (plateau). Thus, the
behavior is elastic-plastic. At high temperature (T3 - red curves), one can see that
the plastic strain is increasing linearly with time (viscoplasticity only). At
"intermediate" temperature (T2 - blue curves), one has both plasticity (with
hardening) and viscoplasticity.
Low Temperature
At low temperature (usually T1 < Tm/3), the following curves are measured at
different loads.
From these curves, one can get the Yield stress (σo) and the Hardening coefficient
(H). To do so, the values (shown in the graph below) of ε and σ (at the different
loads) should be plotted in a σ-ε graph. A linear or Power Law Hardening can be
fitted on those points.
Beside the above, the value of the viscous parameter (η) should be set to zero in
order to "disable" the viscoplasticity. The Power (n) can be set to any value as it
will not be used.
High Temperature
At high temperature (usually T3 > Tm/2), the slope of each curve represents the
strain rate, dε/dt (see the red curves in the figure below).
In order to obtain the viscoplastic parameters (η and n), one should first determine
the value of the Yield stress (σo), which can be equal to zero. To do so, one
should perform a creep test at decreasing loads. The Yield stress is reached when
there is no more plastic strain. If this value is very low, one can consider that the
Then, for each measured creep curve (shown above), the value of strain rate
(dε/dt) should be plotted as a function of (σ- σY)/σ* (which is the stress level of
the plateau minus the Flow stress, σY, which is equal to σo in this case, as H = 0)
in a log-log graph (see below).
The value of η and n can be deduced from this graph (see below).
Beside the above values, as there is no hardening, the value of H should be set to
zero. Therefore the threshold (σy) is equal to the Yield stress σo.
Intermediate Temperature
At intermediate temperature (usually Tm/3 < T3 < Tm/2), the creep test curves are
dependent from both the load (stress level) and time (see the blue curves in the
figure below).
From these curves, one can get the Yield stress (σo) and the Hardening coefficient
(H) as described above in the "Low Temperature" section. To do so, the values
(shown in the graph below) of ε and σ (at the different loads) should be plotted in
a σ-ε graph. A linear or Power Law Hardening can be fitted on those points.
In order to obtain the viscoplastic parameters (η and n), one should first determine
the value of the Yield stress (σo), which can be equal to zero. To do so, one
should perform a creep test at decreasing loads. The Yield stress is reached when
there is no more plastic strain. If this value is very low, one can consider that the
Then, for each measured creep curve (see blue curves in the zoom below), the
values of the INITIAL strain rate (dε/dt) should be plotted as a function of (σ-
σY)/σ* (which is the load, σi, minus the Flow stress, σY. In this case, σY is
The Perzyna model is defined as follows (see the "Stress Models and Properties"
section for more details) :
The second term of the right hand side can be called the "Viscoplastic stress"
( ).
Thus, we have :
with
If we perform a "numerical isothermal tensile test" at various strain rate, one gets
the following curves. The first graph is with a zero hardening, whereas the second
graph is with hardening.
One can well see how the "Viscoplastic stress" is increasing with the strain rate.
As the "Viscoplastic stress" is a function of the inverse power of the "Power" term
"p", it is very sensitive to this parameter. Thus, the combination of the "Power"
term and the "Viscous parameter" term should lead to consistent values of the
"Viscoplastic stress", and that at each temperature.
The following figures are showing the temperature dependence of the Viscous
parameter and of the Power for an A356 Aluminium alloy. Please note that the
Viscous parameter is displayed in a logarithmic scale.
Based upon the Viscous parameter and of the Power of the above graphs, the
"Viscoplastic stress" has been calculated for different strain rates, as shown
hereafter.
One can see on the above graph that the "Viscoplastic stress" is higher at
intermediate temperature, however, when this "Viscoplastic stress" is normalized
by the Yield Stress (see next figure)
The following figure is showing the same curves as the previous section, but all
grouped together.
In ProCAST, all the properties defined by tables are linearly interpolated. In the
case of the Viscous parameter, as the values are changing over so many orders of
magnitude, the linear interpolation is not anymore appropriate if the curve is
defined by too few points.
In the next example, the properties (corresponding to the A356 alloy of the above
figure) between 300 and 500°C are defined only by two points. Thus, the software
will make a linear interpolation of the Viscous Parameter and the Power between
these two temperature. The following figures are showing the "interpolated"
values of the calculated "Viscoplastic Stress" at 320, 400 and 480°C (see red
dots). This is showing that this linear interpolation is not appropriate and leads to
"wrong" values of the "Viscoplastic Stress" (especially at 480°C).
In order to prevent such odd behavior, one should make sure that the Viscous
Parameter and the Power are tabulated with enough points (for instance with a
20°C interval).
Visco-elasticity
This section describes briefly the models which are implemented in the software,
followed by a short description of the different inputs, in relationship with the
above mentioned models.
For the Stress material properties definition, select the Visco-elastic properties.
The specific data for the visco-elastic properties should be defined in the "Shear
Modulus" tab.
Second, the relaxation shear moduli table should be defined in the second tab (see
equation above).
Otherwise, all the other conditions (e.g. displacement BC and Run parameters)
should be defined in a standard way.
• Removal of domains in order to run a new calculation with less domains than a
previous run
Please note that all the functionalities described in this chapter can be used as well
with the Scalar and the DMP versions.
One should note that if a model (named "model_B") was created by an Add or
Delete of "model_A" and then a "model_C" is created as an exact copy of
"model_B", a special care should be taken if it is desired to Extract results in
"model_C" coming either from "model_A" or "model_B" :
Extraction in "model_C" of results from "model_A" : the "model_B.nnc" and
"model_B.enc" files should be copied in the "model_C" directory as
"model_C.nnc" and "model_C.enc" before the Extract operation
Extraction in "modelC" of results from "model_B" : there should be no *.nnc and
*.enc files present (as there is no change of mesh between these two cases).
One should note that if a model (named "model_B") was created by an Add or
Delete of "model_A" and then a "model_C" was created by an Add or Delete of
"model_B" ("cascade Delete/Add operation"), a special care should also be taken
if it is desired to Extract results in "model_C" coming either from "model_A" or
"model_B" :
Extraction in "model_C" of results from "model_A" : one should make sure that
the *.nnc and *.enc files are well present in the "model_B" and "model_C"
directories respectively. Then, the extraction will work well as such.
Extraction in "model_C" of results from "model_B" : if this is the desired
operation, it is required that before the Add/Delete operation of "model_C", from
"model_B", the "model_B.nnc" and "model_B.enc" files must be removed (or
renamed). This will suppress the "link" between model_B and "model_A", thus
allowing the Extract from "model_B" to "model_C".
In any case, it is strongly advised to isolate each case in a separate directory and to
well document the "parents" and "child" in the "Delete/Add" operations.
Warnings
It is not possible to optimize a mesh which is the result of a "Delete/Add"
operation if an "Extract" has been made. This will mess up the node and element
numbering and thus the Extract information.
Results extraction
The Results extraction allows to use results from a previous calculation as initial
conditions for a new calculation.
• Temperature
• Stress fields
• Free surface information (FVOL)
• Porosity
The Stress fields correspond to the 6 components of the stress tensor, the plastic
strain and the displacements (thus the deformed geometry is also extracted).
The Free surface information corresponds to the value of FVOL. This means that
the locations which are full and which are empty will be extracted. When a
thermal only calculation is following an extract of a partial filling, it is necessary
to activate POROS = 1. This is due to a fact that in a thermal only calculation with
POROS = 0, the FVOL values are not stored. Thus, the extract of FVOL would be
lost.
One should note that in version 2009.0 the Microstructure results are not
extracted.
To extract results, the following procedure should be used (see next figure).
The above window appears when the "Initial Conditions/Extract" menu (1) is
called. The list of the available domains appears on the top right of the screen.
First the desired domain(s )which will correspond to the same extraction (i.e.
results extracted from the same case and at the same timestep) should be selected
(2) - more than one domain can be selected at the same time. A simple click on the
domain name is selecting it (and is highlighting it in red). An additional click on a
selected domain will unselect it. Then, the "Extract" button should be pressed (3),
which will open the "Extract panel" (4). The case from which the results should be
extracted must be selected with the "Browse" button (4). The file "prefixt.unf"
should be selected (to select the case named "prefix"). Then the desired timestep
of the extraction should be specified (5). In all cases, the temperature are
extracted. The user can select in addition the Free surface information (FVOL),
the Stress fields and/or the Porosity. Finally, the "Apply" button (6) will confirm
the extract definition.
Once the Extract is confirmed (with the Apply button), the extracted temperature
can be displayed on one or more domains. To do so, select the desired domain(s)
and press on the "Display" button. The temperature will be displayed as shown in
the following figure. Please note that the scale can not be changed in this view.
When an Extract of FVOL, of Stress Fields and/or of Porosity are defined, the
Run Parameter INILEV is automatically set to -1 in the p.dat file. The calculation
will however start at the step 0 (see the "General Run Parameters" section for
more details). If a "pre-defined" set of Run Parameters is selected, automatically,
the value of INILEV will be set to -1.
Warning
When an Extract is performed with an original case having time functions, one
should be careful that the time zero of the extracted calculation is not anymore the
same as the original one. Thus, the time functions have to be shifted in the second
calculation, with respect to the first one.
Examples illustrating the possiblities offered by the Extraction of the free surface
information and the Stress Fields are described in the "Extract examples" section.
Results mapping
The results mapping is working exactly in the same way as the "Extract" (see the
"Results extraction" section for more details).
The only difference is that one should select the "Mapping" button instead of the
"Extract" one. All the rest of the procedure is identical.
The principle of the mapping is the following for fields defined at nodes (like
Temperature, FVOL or porosity) : for each node of the target mesh, the
corresponding element of the source mesh is searched. Then, the value of the
target node is interpolated from the values of the vertices of the source element.
A special care should be taken if the target node is outside the source mesh. If the
mismatch is small (i.e. a fraction of the average size of the neighboring element of
the outside node), the value is calculated as an interpolation of the nodes of the
closest face of the source mesh from the target node.
If the target node is outside the source mesh with a larger mismatch, the average
of the neighboring target nodes which have already been calculated is propagated
to the node. This allows to take care of situation where the target mesh has an
extra volume (like an added riser or a removed cooling channel) with respect to
the source mesh, with reasonable mapping values.
When multiple domains mapping is performed, the mapping is done between the
same "Material Type" of domains (i.e. MOLD, CASTING, CORE, ...). If the
source calculation has a CORE and then it is transformed in CASTING in the
target calculation, there will be no mapping between the CORE temperature of the
source calculation and the CASTING temperature of the target calculation. This
volume will be considered as a hole and the propagation described above will be
applied (if it is desired to map the CORE temperature onto the CASTING volume,
one should change the CORE to CASTING type of this domain and run "datacast
-u" before doing the Mapping).
Warning
The mesh from which the results will be extracted must be in tetrahedra only (it is
not allowed to have hex or wedges). However the target mesh (i.e. the mesh on
which the mapping will be applied) can have any type of elements.
The two following images are showing the result of a mapping of an HPDC
casting from a coarse mesh (left) to a fine mesh (right).
One can see that the match between the two meshes is very good. The quality of
the mapping is as good when going from a fine mesh to a coarse one.
Domain removal
When a mesh contains several domains, it is possible in PreCAST to remove one
or more domains from this mesh. This can be useful in two kinds of situations :
• When one domain of the original mesh is not anymore interesting for the
calculation (e.g. when several scenario are done based upon the same mesh).
• When a mold must be removed at the mold knock-out or a gate is cut.
This is opening a Delete/Add panel on the right of your screen. Select the
domain(s) to be deleted (by toggling between No and Yes).
Then press the "Delete" button and the corresponding domain(s) will be removed.
The main advantage of this "Delete" procedure, with respect to doing it in the
Mesh Generator is that the set-up of the case is remaining after the delete
operation (at least all the features which are applied to the remaining domains).
Thus, it is only necessary to complement the set-up by the missing or new
features and the calculation can be run again.
Remark 1
Please note that when a Domain is "Delete", the Run parameters are automatically
reset (even if a p.dat file was present in the working directory). It is thus necessary
to redefine the Run parameters after a "Delete".
Remark 2
In the case of stress extraction with domain(s) removal, one should make sure that
all the remaining domain(s) are well defined as "CASTING". For instance, if the
original model has a CASTING, a MOLD and a CORE and only the MOLD has
to be deleted first (i.e. to run a stress calculation with the CASTING and the
CORE), the CORE must be switched to CASTING (ignore the Warning message
saying that two different materials are assigned to the different CASTING
domains) in order to have a successful Extract. This is due to the "renumbering"
algorithm applied to the Extract which is currently made only on the CASTING
domains.
Domain addition
With the "Domain addition" functionality, it is possible to Add one or more
domain(s) to an existing mesh or case. It is also possible to add an Enclosure.
To illustrate that, one shall take the example of an HPDC case where we will add
the outer mold to the model.
As it can be shown in the next figure, only the "inner mold" is taken into account
in this first model.
The goal is now to "Add" the "Outer mold" to this model. This outer mold has
been mesh separately as shown in the next figure. Please note that of course, the
coordinate system of the two meshes should coincide.
In order to Add this "outer mold", the following steps should be performed :
A Browser will open. Select the right filter (d.dat or .mesh file) and browse the
desired mesh of the outer mold. Select it and press the "Open" button. This will
open the "Second model" window, as shown hereafter.
Then, the material properties and boundary conditions corresponding to this new
added domain(s) should be specified.
In particular, one should not forget to create the non-coincident interfaces between
this new domain(s) and the one of the first original model (see figure hereafter).
As we can see on the following figure, the inner and outer molds are well taken
into account in this new model.
Remarks
It is possible to use the "Add" functionality to add an enclosure to an existing case
(where the enclosure would not been present already).
Please note that when a Domain is "Added", the Run parameters are automatically
reset (even if a p.dat file was present in the working directory). It is thus necessary
to redefine the Run parameters after an "Add".
Extract examples
This section is giving a few examples of the different possibilities offered by the
extended Extractions possibilities :
The following example is showing a very simple case where the casting (and
gating) is first solidified in a Rigid mold (the mold can be as well elastic-plastic).
Then the mold is opened and thus the shape of the casting can change due to the
fact that the mold constraints are disappearing. Later on, the gate is removed,
allowing the part to take its final shape. This can be done now in three stages in
the following way.
The stress calculation should be activated in the casting and gating (it is optional
in the mold). Then the calculation should be run until the time of the mold
opening (mold knock-out).
The following figures are showing the results of these three calculations. The first
set of pictures shows the results without an increased deformation, whereas the
second set is showing the same results, with a deformation magnitude of 10. In
both sets, the total displacement is shown (with the same scale for all figures),
allowing to well see the effect of the mold knock-out and the gate cutting on the
shape of the casting.
View of the total displacement (same scale for all the figures, from 0 to 18 mm) at
different stages of the casting process : 1. Initial geometry (calculation 1), 2. Just
before mold knock-out (calculation 1), 3. Just after mold knock-out (calculation 2), 4.
Just before gate cutting (calculation 2), 5. Just after gate cutting (calculation 3), 6.
Room temperature (calculation 3). The pictures are taken with a "Displacement
magnitude" of 0 (no visible deformation of the geometry).
View of the total displacement (same scale for all the figures, from 0 to 18 mm) at
different stages of the casting process : 1. Initial geometry (calculation 1), 2. Just
before mold knock-out (calculation 1), 3. Just after mold knock-out (calculation 2), 4.
Just before gate cutting (calculation 2), 5. Just after gate cutting (calculation 3), 6.
Room temperature (calculation 3). The pictures are taken with a "Displacement
magnitude" of 10 (increased deformation).
Quite often in Investment or gravity casting, the cavity is not completely filled
(i.e. the pouring cup is only partially filled). If one wants to perform a stress
calculation on such partially filled casting, but without having to perform the
stress calculation during the filling, it is possible to do so with the extraction of
the Free surface information.
Moreover, the stress calculation can be continued after the mold removal (mold
knock-out).
The different stages are the following (illustrated by a very simple test geometry) :
Calculation 1 (filling) :
Filling + Thermal calculation, with the usual set-up and LVSURF = 0.70 (to stop
the filling at 70%)
The calculation can be stopped just after the end of the filling.
The following figure is showing the result of the stress calculation (the total
displacements are shown, with a "Displacement magnitude" of 20).
The following figure is showing the results just before and just after the mold
knock-out.
The following figure is showing the temperature and casting shapes just after the
mold knock-out (step 0 of calculation 3) and at the end of the cooling (last step of
calculation 3).
The procedure to do that is the following. Please note that on this simple example,
it may look not optimum to use this procedure to achieve this goal, however, if the
mold is very complex and leads to many interface definitions, this can be quite
powerful. Moreover, this could allow a quite easy change of shot piston geometry
without remeshing the mold and without redefining all the mold properties,
boundary conditions and interfaces in PreCAST.
The above figure is showing the temperature contour during the last cycle,
whereas the following figure shows the temperature evolutions along the 5
calculated cycles.
2. Load this d.dat file into PreCAST and remove the casting domain (using the
Geometry -> Delete/Add Materials menu). The geometry will appear like the
following figure.
3. In the "Geometry -> Delete/Add Materials" menu, select "Add" and Browse in
the directory where the mesh containing the casting and the shot piston is located.
Please note that this can be either in a mesh file (*.mesh) or in a data file (*d.dat).
4. Open this second file and a new window called "Second model" will open
below the "First model window" (see hereafter). The list of all available domains
of he second model is shown in the "Second model" window.
5. Select the domains (Materials) that you want to import from the Second model
to the First one (by toggling No to Yes) and then press the "Move up" button (left
window above).
6. The selected domains of the Second model are automatically imported in the
First model (see right window above).
7. Now that the geometry is complete, all the missing assignments should be
made. The first one is the Material properties assignment (see below).
8. If the "Added" geometry was coming from a mesh file, some interfaces may
need to be created. If the "Added" geometry was coming from a d.dat file, the
already present interfaces will be imported as well and thus do not need to be
created again.
9. Between the original model (of Calculation 1) and the imported domains
(coming from Calculation 2), it is necessary to create "Non-coincident" interfaces.
It is recalled that if temperature dependant interface heat transfer coefficient will
be used, the Master must be the casting.
10. As we would like to use the temperature of the mold at the end of the last
cycle, we should extract it from Calculation 1. To do so, select all the mold
domains (which will be highlighted in red) and click on "Extract". A window will
open, which allows to Browse the Calculation 1 case and to specify the timestep
corresponding to the end of the last cycle.
11. When the Browsed model is selected (select the prefixt.unf file), the Open
button can be pressed, in order to execute the extraction.
12. When the "Display" button is pressed, the extracted temperature can be
viewed.
The next three figures are showing the results of the calculation. The first figure
shows the initial step with the extracted temperature of the mold.
Important Remark
As described at point 9) above, at interfaces between the original domains of
Calculation 1 and the imported (added) domains of Calculation 2, there must be a
"Non-coincident" (NCOINC) interface.
In the case of large castings, sometimes, the filling is done is several stages. As
the time between two filling sequences may be significant, it is now possible to
run a "thermal only" calculation in between these fillings. This is saving a lot of
CPU time.
The above figure is showing the cooling in between the two filling stages. As it is
a Thermal only calculation, it can be done with large timesteps (much larger than
when the flow is activated).
RUN PARAMETERS
This section contains the comprehensive description of all Run parameters. The
ones shown in blue are either new or have been modified or further described,
with respect to previous versions.
The different categories of Run Parameters are organized in Tabs. The first Tab
layer (1) corresponds to the different modules of ProCAST. For each modules, the
Run Parameters are divided in a "Standard" tab (2), with the most often used
parameters and one or two "Advanced" tabs (2) with less often used Run
Parameters. The values are then defined in (3).
The values of the Run Parameters are stored in a file named "prefixp.dat" (often
called the "p.dat" file). The Run Parameters can thus be changed either in the
PreCAST interface, or directly "by hand" in the "p.dat" file (see example below).
The default values are not anymore specified in the manual, as they are set in the
"default_p.dat" file (see the "Pre-defined Run Parameters" section for more
details).
The Run Parameters which are used by any module are grouped in the "General"
tab. It is also possible to defined the units which will be used for the display of the
results (it will not affect the results themselves).
The Run Parameters of the "General/Advanced" tab are usually not changed from
the default values, except in very specific cases. It corresponds mainly to the
parameters of the different general solver algorithms.
General
AVEPROP
specifies the method to be used in calculating the properties for each element.
ProCAST will calculate the properties at each Gauss point or you may specify that
the properties be calculated only at the element center and that this value will be
used as an average for the whole element. This averaging reduces, somewhat, the
finite element integration time required. This averaging does not apply to the
specific heat or enthalpy calculations.
Choose from:
0 to calculate at each point, or
1 to use the average
CGSQ
specifies the Conjugate Gradient Squared solver flag. The values specified in this
parameter may be added together. This allows you to "build" a customized solver
approach for your simulation.
Choose from:
CONVTOL
specifies the convergence tolerance which will be used in conjunction with the
default non-symmetric iterative solver.
Enter a floating (real) value.
DIAG
specifies the diagonal preconditioning flag for the symmetric solver.
Choose from:
0 = use partial Cholesky preconditioning for everything,
8 = use diagonal preconditioning for pressure,
16 = use diagonal preconditioning for energy, and
16384 = use diagonal preconditioning for radiosity
DT
specifies the initial time step size. Setting DT to zero when INILEV > 0 will cause
ProCAST to use the DT at step INILEV.
Enter a floating (real) value. The default is 1.0000e-03.
Select the units of time from: {sec | min}
DTMAX
specifies the maximum time step size.
Enter a floating (real) value.
Select the units of time from: {sec | min}
The time table is stored in the d.dat file. The value which appears in the DTMAX
field, as well as in the p.dat file should be zero in order to activate the table. If one
would like to deactivate the table, a non-zero value should be set.
Please note that if the value of DTMAX is reduced at a given time (e.g. reduction
from 5 to 1 s. at 50 s.) the timestep may be reduced only after about 55 s. (as the
previous timestep may have been at 49.9 s. and thus the next timestep will still be
done with DTMAX = 5 s.).
DTMAXFILL
specifies the maximum time step size which will be used during the filling stage
only. Once the filling is finished, the DTMAX value will be used. If
DTMAXFILL is not defined (i.e. it is set to zero), the value of DTMAX will be
used for the whole calculation.
Enter a floating (real) value.
Select the units of time from: {sec | min}
INILEV
specifies the initial time level. When an analysis is first started, INILEV should be
equal to zero. When you are resuming an analysis, INILEV should be set to the
time step from which you would like to continue. Note: You must have results for
that time step.
If a larger value than the last stored step is specified, the last stored timestep will
be used. The last complete step corresponding to TFREQ, VFREQ and SFREQ
will be selected and the calculation will be restarted with this last time step. It is
assumed that MFREQ=TFREQ.
LUFAC
specifies the preconditioning parameter for the CGSQ solver. This parameter may
speed-up calculations of large models
Choose from:
0 to use diagonal preconditioning, or
1 to use partial LU factorization preconditioning
NEWTONR
turns on the NEWTON Raphson technique for the energy equation.
Choose from:
0 to turn off the Newton Raphson technique,
1 to turn the Newton Raphson technique on, or
2 to turn on the Newton Raphson technique and use bsplines
The default is 0.
Option 2 results in using b-splines instead of linear line segments in the
representation of the thermal properties. It is suggested that all thermal input data
be smoothed before attempting to use b-splines.
Enter an integer value.
NPRFR
specifies the printout frequency. This controls the time step interval at which
results are output to the prefixp.out file.
Enter an integer value.
NRSTAR
specifies the number of allowable restarts before the entire run is abandoned. A
restart occurs when the maximum number of corrections is reached. If too many
restarts are taking place, it could indicate problems with the model setup.
Enter an integer value.
NSTEP
specifies the number of time steps to take in the current run and is used in
conjunction with TFINAL. ProCAST will terminate the run when it reaches this
limit or the TFINAL value, whichever occurs first.
Enter an integer value.
PRNLEV
specifies the level of nodal results to be printed out. The values specified in this
parameter may be added together. This allows you to collect combinations of
nodal information in a single run.
Choose from:
0 = no printout,
1 = nodal velocities,
8 = nodal pressures,
16 = nodal temperatures,
64 = nodal turbulence intensities,
128 = nodal turbulence dissipation rates,
1024 = nodal displacements,
8192 = surface heat fluxes, and
32768 = nodal magnetic potentials
SDEBUG
specifies the level of solution debugging messages to be captured. These messages
are written to the p.out file.
Choose from:
0 to capture no solution debugging messages, or 1 to obtain information
concerning, solver performance, time step control, and the free surface model
TENDFILL
specifies the delay after the end of the filling at which to terminate a ProCAST
analysis. If this parameter is zero, this parameter will not be active. If one wants to
stop the calculation right after the end of the filling, one should set a small value,
different from 0.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TFINAL
specifies the simulated time at which to terminate a ProCAST analysis. If this
parameter is zero, the run will be stopped by the TSTOP or NSTEP parameter. If
NSTEP and TSTOP and TFINAL parameters are set, the simulation will be
terminated based upon which parameter is reached first.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TSTOP
specifies the temperature at which to terminate a ProCAST analysis (i.e. when all
the temperatures at all nodes are below the TSTOP temperature). If this parameter
is zero, the run will be stopped by the TFINAL or NSTEP parameter. If NSTEP
and TSTOP and TFINAL parameters are set, the simulation will be terminated
based upon which parameter is reached first.
Enter a floating (real) value.
Select the units of Temperature from: {C | K | F}
TMODR
specifies the time step modification factor for restarts. If MAXCOR correction
steps are taken without convergence, the time step is multiplied by TMODR.
Therefore, this number should be less than 1.
Enter a floating (real) value.
TMODS
specifies the time step modification factor for normal stepping. If the number of
correction steps is less than or equal to NCORL, the subsequent time step is
multiplied by TMODS.
If the number of correction steps is greater than or equal to NCORU, the
subsequent time step is divided by TMODS.
Enter a floating (real) value.
Units
PUNITS
specifies the pressure units to be used in the outputs.
Choose from: {N/m**2 | Pa | KPa | MPa | bar | dyne/cm**2 | atm | psia | Ksi |
lb/ft**2}
QUNITS
specifies the heat flux units to be used in the outputs.
Choose from: { W/m**2 | cal/cm**2/sec | cal/mm**2/sec | Btu/ft**2/sec |
Btu/in**2/sec | cal/cm**2/min | cal/mm**2/min | Btu/ft**2/min | Btu/in**2/min}
TCUNITS
specifies the thermocouple units to be used in the outputs and is only used for
inverse modeling.
Choose from: {C | F | R | K}
TUNITS
specifies the temperature units to be used in the outputs.
Choose from: {C | F | R | K}
VUNITS
specifies the velocity units to be used in the outputs.
Choose from: {m/sec | cm/sec | mm/sec | ft/sec | in /sec | m/min | cm/min |
mm/min | ft/min | in/min}
The choice of the units does not have any effect on the calculated results.
The Thermal model activation, the Temperature results storage frequency, as well
as the Porosity model and parameters are the main parameters to be defined in this
tab.
Thermal
CINIT
Reserved for future use.
CLUMP
specifies the capacitance matrix lumping factor.
Enter:
0 to use consistent matrix, or 1 to use diagonal matrix
CONVT
specifies the convergence criterion for temperature. A value of around one degree
is generally appropriate. Values larger than the mushy (liquidus--solidus) zone
range are not recommended.
Enter a floating (real) value.
Select the units of temperature from: {C | F | R | K}
DRAINFS
For LPDC, it is possible to "drain" the remaining liquid in the inlet sprue region
which is still liquid when the pressure is released. DRAINFS corresponds to the
critical solid fraction below which the metal is drained. This means that all the
metal which has a solid fraction below DRAINFS which is connected to the
pressure or inlet boundary condition location will be drained (i.e the metal will be
removed). This value should be used in conjunction with DRAINTIME.
Enter a floating (real) value between 0 and 1. This Run parameter has to be added
manually in the p.dat file. The default value is 0.7.
DRAINTIME
For LPDC, it is possible to "drain" the remaining liquid in the inlet sprue region
which is still liquid when the pressure is released. The value of DRAINTIME
corresponds to the time at which the pressure is released and thus the time at
which draining will occur. The unit code of time should be specified before the
time (e.g. 1 for seconds). This value should be used in conjunction with
DRAINFS.
Enter a floating (real) value after the unit code of time (e.g. DRAINTIME 1
25.). This Run parameter has to be added manually in the p.dat file.
FEEDLEN
"Feeding length" (distance), at which macroporosity can occur, beyond the
MACROFS isosurface.
Enter a floating (real) value.
GATEFEED
allows to specify whether liquid can be fed at the ingate or not. In the case of
injection (i.e. hpdc or lpdc), the shrinkage at the gate is compensated by the liquid
pushed by the piston (for hpdc). On the other hand, in gravity casting, there is no
feeding at the top of the risers and piping will occur. GATEFEED=1 tells the
software to activate the feeding at the ingate, thus leading to no piping at this
location. GATEFEED=1 will automatically activate the "Active feeding" where
inlet velocities or pressures are set (see the "Active Feeding" section for more
details).
Enter an integer value.
GATEFS
GATEFS allows to control the critical fraction of solid above which there is no
more feeding during the third stage pressure in HPDC (see the "Active Feeding"
section for more details). This Run Parameter has to be added or modified
manually in the p.dat file (it does not appear in PreCAST). A higher value of
GATEFS will involve less porosity.
Enter a value between zero and one. The default value is 0.95.
GATENODE
GATENODE is an alternative to GATEFEED. When a gate feeding should be
applied, but there is no external surface of the model where an inlet surface or a
pressure can be applied. This is the case for instance when a shot piston is
modeled. In such case, one can define a location in the volume where this gate
feeding is applied (typically in the middle of the biscuit). In order to define this
location, the corresponding node number should be specified. In such case, no
pressure or inlet velocity BC needs to be specified. To find the desired node
number, it is advised to use the "pick" selection in ViewCAST (Run DataCAST
first) (see the "Active Feeding" section for more details).
Enter an integer value.
LINSRC
specifies the source term linearization parameter for micromodels. This parameter
may be used in conjunction with micromodels that control the evolution of the
solid fraction and thus the release of latent heat. The default value of zero
indicates that the heat generation will only appear in the right hand side source
term. A value of one will give some contribution to the diagonal terms of the left
hand side matrix.
This improves numerical stability, but does require that the LHS be factored,
which would normally happen anyway.
Enter:
0 = no linearization, or
1 = for linearization of the source term
MACROFS
Parameter for the macroporosity calculation. It sets the limiting fraction of solid
between the macroporosity and the microporosity formation. The value should be
set between 0 and 1.
Enter a floating (real) value.
MOBILE
The mobility limit, MOBILE, determines the critical liquid fraction at which the
fluid is no longer able to flow.
When considering the development of pipes in risers, higher mobility limits will
produce deeper pipes.
MOBILE will also control the motion of pistons in shot sleeve (HPDC). The
piston will not move anymore when the higher solid fraction of an element in
contact with the piston is higher than (1-MOBILE).
Enter a floating (real) value.
MOLDRIG
The rigidity of the mold has an influence on the amount of porosity in the case of
expanding alloys. If the mold is totally rigid, the casting can not expand and thus
the alloy expansion will be "available" for the "refill" of the existing porosity. On
the other hand, if the mold is very soft (or weak), the casting will expand and thus
there will be no "refill" of the porosity (of course, the reality is more complex as
the solid shell is thick enough it will act as a "rigid" mold, even if the sand mold is
weak.
In order to take the mold rigidity into account, the Run Parameter MOLDRIG is
used (see the "Cast Iron Porosity model" section for more details). MOLDRIG
should be defined by a value between 0 and 1. All the net expansion is multiplied
by MOLDRIG. Thus, with MOLDRIG=1, corresponding to a rigid mole, the
expansion will be fully accounted. On the other hand, no expansion will be taken
into account if MOLDRIG=0. The expansion will be compensated by the mold
movement because the mold is too weak to hold the expansion in this case. For
real situations, the value of MOLDRIG should be set somewhere between 0 and 1
depending upon the casting processes. The default value is 1.
PIPEFS
Parameter for the piping calculation (during a porosity calculation). It sets the
limiting fraction of solid for piping formation. The value should be set between 0
and 1.
In order to disable the porosity calculation, one should set POROS = 0. However,
this may still lead to some piping calculation. In order to disable also the piping,
one should set in addition PIPEFS = 0.
For stress calculations, PIPEFS must be set to 0. This is needed in order to prevent
the unexepected influence of the pipe on the stress calculation. One should note
that if in the physical sitution, piping must occur, this will be taken into account in
the calculation as "macropores" when PIPEFS = 0.
Enter a floating (real) value.
POROS
specifies the porosity calculations to be performed (see the porosity models
description).
Choose from:
0 for no porosity calculation,
1 - most advanced porosity model
4 - evolution of the POROS=8 model, which allows to handle multiple free
surfaces
8 - old porosity model from version 3.2.0
In order to disable the porosity calculation, one should set POROS = 0. However,
this may still lead to some piping calculation. In order to disable also the piping,
one should set in addition PIPEFS = 0.
QFREQ
specifies the time step interval for writing heat flux data to the unformatted results
file. This parameter can be used to reduce the size of the prefixq.unf file. Heat flux
results may not be of interest to everyone, so it may be desirable to minimize the
size of this file.
Enter an integer value.
TFREQ
specifies the time step interval for writing temperature data to the unformatted
results file. This parameter can be used to reduce the size of the prefixt.unf file,
which can become quite large for problems with many nodes and time steps. Note
that it is only possible to restart a run from one of the time steps that was written
out.
THERMAL
specifies the thermal analysis to be performed.
Choose from:
0 for no thermal analysis. Solve flow equations alone,
1 to perform thermal analysis, using temperature as the primary variable, or
2 to perform thermal analysis, using enthalpy as the primary variable
TRELAX
specifies the temperature relaxation parameter. This is used for computing the
initial guess for the temperature field in the predictor step. TRELAX should be
greater than or equal to zero and less than or equal to one.
Enter a floating (real) value.
USERHO
specifies the how the density in the mushy zone is calculated.
Choose from:
0 - automatic density calculation, by extrapolation of the densities at liquidus and
solidus, weighted by the fraction of solid.
1 - the density table, which is defined in the database is used.
When cycling is modeled, one should define the corresponding Run Parameters in
this tab (number of cycle and cycling time). In addition, when "Die Combo"
interface definitions are used, additional parameters should be set.
CYCLEF
CYCLEF allows to perform a filling calculation at the last cycle. This means that
the N-1 cycles are run with a thermal only calculation and for the last cycle (N) a
filling-thermal calculation is performed.
The casting should be set as empty and the inlet boundary conditions (thermal and
velocity/inlet) should be defined (these will be applied only during the last cycle
and will be ignored during the first N-1 thermal cycles).
When CYCLEF is ON, FLOW should be set to 9 (instead of 3).
Please note that if a value of 1 is used for FLOW, the flow (i.e. natural
convection) will be activated during all cycles (and the filling will be applied only
to the last one). Although this is possible, this is not really useful and it is thus not
advised to do so.
Enter an integer value. CYCLEF OFF corresponds to a value of 0 (default),
whereas a value of 1 corresponds to ON.
NCYCLE
specifies the number of casting cycles to be simulated in a continuous mode. This
parameter is used along with TCYCLE. Both NCYCLE and TCYCLE must be
set. This parameter is typically used in die casting and permanent mold problems.
Enter an integer value.
TCYCLE
specifies the time of casting cycle to be simulated in a continuous mode (duration
of a cycle). This parameter is used along with NCYCLE. Both NCYCLE and
TCYCLE must be set.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TOPEN
specifies the time at which the mold opens, during one casting cycle. This is used
in conjunction with the "Die Combo" interface definition.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TEJECT
specifies the time at which the part is ejected from the mold, during one casting
cycle. This is used in conjunction with the "Die Combo" interface definition.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TBSPRAY
specifies the time of the beginning of the spray sequence, during one casting
cycle. This is used in conjunction with the "Die Combo" interface definition.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TESPRAY
specifies the time of the end of the spray sequence, during one casting cycle. This
is used in conjunction with the "Die Combo" interface definition.
Enter a floating (real) value.
Select the units of time from: {sec | min}
TCLOSE
specifies the time at which the mold closes (before the start of the next filling),
during one casting cycle. This is used in conjunction with the "Die Combo"
interface definition.
Enter a floating (real) value.
Select the units of time from: {sec | min}
When a calculation with Radiation, including View Factors is run, the following
parameters should be defined. If the "View factor OFF" option is selected, it is not
needed to define these Run parameters.
ANGTOL
specifies the angle tolerance to be used with VFLIM. Radiation faces which are
grouped using VFLIM tolerance are further differentiated by their solid angle.
Enter a floating (real) value.
ENCLID
specifies an enclosure identification number. This parameter is used in
combination with VFDISP for updating view factors by a displacement interval.
ENCLID indicates which enclosure set is to be tracked, in case all the enclosure
elements are not moving at the same rate.
Enter an integer value.
EPTOL
specifies the emissive power tolerance to be used with VFLIM. Radiation faces
which are grouped using VFLIM tolerance are further differentiated by their solid
angle.
Enter a floating (real) value.
RDEBUG
specifies the user debug parameter for printing detailed view factor information.
Various combinations of these files may be obtained by adding together these
numbers. For example, RDEBUG = 7 gives all three files. Note that these files can
be quite large, especially the prefix.vf.
Enter an integer value based upon the following:
1 for face to face view factors after symmetrization, in the prefix.vf file,
2 for face to group view factors after symmetrization, in the prepfix.view file
(necessary to see FACE TO GROUP in ViewCAST), or
4 for row sum errors before symmetrization, in the prefix.serr file (necessary to
see ROW SUM ERRORS in ViewCAST
RFREQ
specifies the radiation update frequency. This provides a mechanism for
recomputing the radiosities at some time step interval other than one. This is
particularly useful if you are performing a filling transient along with the view
factor radiation model. In this case, the time step size may be small due to the
filling whereas the mold temperature may not be changing very rapidly. You can
save some computational time by recomputing the radiosities at every tenth step,
for example.
Enter an integer value.
TRI2QUAD
specifies the option to group or not triangles into quadrangles for the radiation
view factor calculation. When TRI2QUAD is set to 1, adjacent triangles are
grouped in order to obtain quadrangles (providing the angle between the triangles
is not too large). This has the effect of reducing the number of radiative face (by
about 50%) which is cutting down the CPU time significantly (by about 75%).
VFDISP
specifies the displacement interval for updating view factors in the radiation
model if there are moving relative surfaces. This is used in conjunction with
ENCLID and will be used in preference to VFTIME if both are specified.
Enter a floating (real) value.
Choose the units of length from: {m | cm | mm | ft | in}. The default is m.
VFLIM
specifies the view factor limit. This parameter is used to agglomerate faces in the
view factor calculations. This reduces the size of the radiosity matrix and speeds
up the radiation calculations.
VFLIM
can be set to a fraction between zero and one. If one face occupies less than this
fraction of the total view space, as seen from another face, the first face is
combined with some others. A value of 0.01 is a good starting point.
Enter a floating (real) value.
VFTIME
specifies the time interval for updating view factors in the radiation model if there
are moving relative surfaces.
Enter a floating (real) value.
Choose the units of time from: {sec | min}. The default is sec.
When fluid flow and/or filling models are activated, the "Flow" Run Parameters
should be defined. The main parameters are present in the "Standard" tab. In the
"Advanced 1" tab, one should mainly define the WSHEAR and the WALLF
parameters. In most cases, all the other parameters can remain as the default
values.
ADVECTW
specifies the weighting of advection velocities and controls the degree of
non-linearity of the momentum equations. ADVECTW can take on values
between zero and one. Velocities at the last time step are used as the advecting
velocities if a value of zero is used. Velocities at the current time step are used as
the advecting velocities if a value of one is used. Numerical experience has shown
that the accuracy of natural circulation flows can be enhanced by using a factor of
0.5. For most filling analyses, a value of zero works fine and requires much less
computational time.
Enter a floating (real) value.
COLDSHUT
Disable the cold shut algorithm detection. If a "cold shut switch off" flag is
activated (which is stopping the calculation) for an undesired reason, it is possible
to disable this detection (and restart the calculation), by setting COLDSHUT = 0.
The default value is 1.
A cold shut will happen if for any given node in the fluid domain there is no valid
connection all the way to the inlets. This path can be
broken by 2 conditions:
- solid fraction is above 1-MOBILE
- FVOL < 0.5 (this can be due to a physical boundary, e.g. if there are two
separate fluid cavities, or a bad mesh)
The second condition may happen if the mesh has an anomaly, like for instance if
an element is connected to the neighbors only through edges and not through at
least one face. If this situation occurs, even for an isothermal case, the "cold shut
switch off" will happen right at the beginning of the calculation.
COMPRES
specifies whether this is an incompressible flow problem or a compressible flow
problem.
Choose from:
0 to specify an incompressible flow problem, or
1 to specify a compressible flow problem
CONVV
specifies the convergence criterion for velocity. The value given here is a fraction
of the maximum velocity calculated at each step. Generally, .05 or 5% is
appropriate.
Enter a floating (real) value.
COUPLED
specifies whether the energy and fluid solutions should be coupled or decoupled
within a time step. When the analysis is decoupled, the momentum and pressure
equations are solved repeatedly until convergence. Subsequently, the energy
equation is solved until convergence, assuming the flow field is fixed. With a
coupled analysis, the energy equation is solved in the same loop with momentum
and pressure. Both the momentum and temperature convergence criteria have to
be met to terminate the loop. This method is more accurate, but usually takes
more computational time.
Choose from:
0 to decouple energy and fluid solutions within a time step, or
1 to fully couple energy and fluid solutions within a timestep
COURANT
specifies the courant limit on time step size. This parameter is only used for filling
problems. If COURANT is set to 1.0, the time step will be adjusted so that the
fluid will advance no more than one element length. This is a fairly severe limit
on time step size, but will give the most accurate results for filling transients.
Acceptable results can usually be obtained with values around 100.
Enter a floating (real) value.
DETACHTOP
allows to have a better treatment of the detachment of liquid which is in contact
with horizontal top walls (like "roofs"). A value of 1 allows a better detachment of
such liquid regions, however it takes some more CPU time. This Run parameter
should be added/changed manually in the p.dat file.
EDGE
This Run parameter is not anymore available (however it should be set to 0 if
present in the p.out file)
ENDFILL
Sometimes, the user may not be interested by the filling of the last percents (e.g.
the end of the filling of a riser or an overflow). If ENDFILL = 0.98, once 98%
will be reached, the remaining 2% will be filled in one timestep.
FFREQ
specifies the flow update frequency. This provides a mechanism for re-computing
the velocities at some time step interval other than one. This might come into play
if you were solving a conjugate heat transfer problem where the velocity field is
changing on a longer time scale than the temperatures. This option is not
appropriate for free surface problems.
Enter an integer value.
FGROUP
The mass balance correction algorithm at the free surface in the case of multiple
free surface (e.g. multiple inlets) is governed by FGROUP. With FGROUP = 0,
the free surface is considered as a single entity and the mass balance correction is
performed globally on the whole free surface. This algorithm is not working very
well in the case of multiple free surface (e.g. with multiple inlets). In order to
obtain a well balanced filling in the case of multiple inlets one should activate a
more sophisticated algorithm with FGROUP=2. FGROUP = 2 is the default
value for both the Scalar and the DMP solvers (please note that FGROUP = 1 is
not anymore recommended).
Enter an integer value. Default value : 2.
FLOW
controls the use of fluid equations.
Choose from:
0 do not solve fluid equations,
1 to solve fluid equations,
3 to solve fluid equations during filling, but switch over to thermal only analysis
when the LVSURF fill limit is reached and NCYCLE = 1,
9 to solve fluid equations during filling, but switch over to thermal only analysis
when the LVSURF fill limit is reached and NCYCLE > 1. FLOW must be set to 9
when CYCLEF is activated (see the "Cycling Run Parameters" sectoin for more
details).
The default is 0 if there are no "Casting" materials. If "Casting" materials exist,
the default is 1.
FLOWDEL
specifies the delay time between the end of fill and a switch to a thermal only, in
the case of FLOW = 3 simulation. This option is used in conjunction with velocity
boundary conditions with active fill limits. The time delay corresponds to the time
for the fluid to completely settle down in the casting before the thermal only phase
begins.
Enter a floating (real) value.
Select the units of time from: {sec | min} The default is sec.
FREESF
specifies the free surface model number to be used.
Choose from:
0 = no free surface model activated
1 = use the momentum dominated movement of free surface, rapid filling model,
2 = use the gravity dominated movement of free surface, slow filling model.
FREESFBAL
In the case of large horizontal free surfaces (e.g. bottom filling of large ingots,
filling of a "swimming pool"), the free surface may be not as calm as it should be.
In order to stabilize the free surface (and to avoid unexpected waves), the
calculation of the mass balance of partially filled elements at the free surface was
refined. This can be activated with FREESFBAL = 2. This should be activated
only for gravity filling with large horizontal free surfaces and not in other cases.
The Run parameter FREESFBAL should be added manually in the p.dat file.
0 = default value (no special treatment of the partially filled elements)
2 = improved algorithm for the calculation of the mass balance of partially filled
elements at the free surface (to be added manually in the p.dat file)
FREESFOPT
The filling algorithm (in the case of FREESF = 1) was significantly improved in
version 2006.0. From now on, three filling algorithms are available :
FREESFOPT = 0 : it corresponds to the filling algorithm of version 2005.0. It is
less precise than the two other algorithms, however it is the most robust algorithm.
FREESFOPT = 1 and 2 corresponds to two improved filling algorithms. The
difference between these two algorithm is the "numerical balance" between the
mass conservation contribution to the free surface and the momentum
contribution. With FREESFOPT = 1, the mass conservation contribution is more
important than the momentum contribution. This algorithm is more robust, but
may lead to slightly less precise results. With FREESFOPT = 2, the momentum
contribution is predominant over the mass conservation contribution. This model
is supposed to be the most precise, however it is more sensitive to the quality of
the mesh. In order to have good results with FREESFOPT = 2, the mesh should be
well adapted to the local free surface (e.g. one should have enough nodes in the
stream of the liquid in the case of a thin jet of liquid). This means that the
FREESFOPT = 2 algorithm is less robust as the 1 (or the 0).
As a conclusion, it is strongly advised to use FREESFOPT = 1 in most cases. If
more precision is required, the mesh should be refined and well tuned so that
FREESFOPT = 2 could be used. The FREESFOPT = 0 option is provided for
backwards compatibility purposes, however, it is not advised to be used.
The default value is 1.
GAS
specifies whether or not to consider the trapped gas effects. If the option to
consider trapped gas effects is chosen, trapped gas effects will be considered even
when the model contains no vents, gas injection, or gas diffusion through the
mold. When features normally found in a gas problem (vents, injection, or
gas diffusion through the mold ) are present in a model, GAS will be set
automatically.
Choose from:
0 to not consider trapped gas effects, or
1 to consider trapped gas effects
HEAD_ON
specifies the approach to be used when calculating gravitational term in the
momentum equation for flow problems without free surfaces.
Choose from:
0 = calculate as rho - rho_ref, or
1 = calculate as rho * g
HIVISC
specifies different solution methods for viscosity in the flow problem.
Choose from:
0 = normal flow problem,
1= high viscous flow problem. To be used when the Reynolds number is less the
one. This method only
works for viscosity less than 104 poise. In this case, the advection terms are
neglected, symmetric solvers are employed on the momentum equations, and
large degrees of pressure relaxation are utilized, or
2 = very high viscous flow problem. To be used when the Reynolds number is less
the one. This method is always preferred. In this case, the advection terms are
neglected and momentum effect on implicitly included within a Poisson pressure
equation. This option usually allows for much larger time steps than HIVISC = 1
JUNCTION
JUNCTION is activating different algorithm for the calculation of the "metal front
tracking" and metal movement during the filling. This allows to track where the
particles and impurities are most likely to be accumulated and to view the flow
junction lines. As this algorithm is taking an amount of CPU time which may not
be negligible, by default this calculation is not activated (JUNCTION = 0). To
activate it, one should set JUNCTION to 1, 2, 3, 4, 11, 12, 13 or 14 in the p.dat
file (this Run parameter should be added manually in the p.dat file). For the
description of the meaning of the different values, see the "Results analysis/Fluid
Front Tracking" section for more details.
Enter an integer value of 0, 1, 2, 3, 4, 11, 12, 13 or 14 (default : 0).
LVSURF
provides a way to switch from the filling transient to a mode where advection is
due to buoyancy and shrinkage. LVSURF turns all inlets off. It is assumed
thereafter that the free surface is perpendicular to the gravity vector. This allows
the time step to increase significantly. The number represents the fraction of the
total casting volume which is to be filled before changing modes.
Enter a floating (real) value.
MLDUPDT
specifies the material properties update frequency (in the mold only). For large
casting (when a mold is present), it the calculation can be slightly speed-up by
reducing the frequency of update of the material properites (reducing the number
of matrix operations). The integer which is specified after corresponds to the
update frequency. By default the value is 1. See the "Large Ingot filling" section
for more details.
Enter an integer value.
MLUMP
specifies the mass matrix lumping factor.
Choose from:
0.0 to use a consistent matrix, or
1.0 to use a diagonal matrix
NNEWTON
specifies whether the flow is newtonian or non-newtonian.
Choose from:
0 to indicate Newtonian flow, or
1 to indicate non-newtonian flow, where viscosity is a function of shear rate
PENETRATE
Flag to activate the algorithm of inter-penetrating meshes (with a value of 1). This
is used especially for the modeling of a shot piston in hpdc.
Enter an integer value.
PFREQ
specifies the "Particle tracing" launch frequency in the solver. Particles are
launched at each node of the inlet (defined by a velocity BC, an inlet pressure BC
or an Inlet BC), every PFREQ steps. A value of 50 is recommended (see "Display
parameters" for the description of Particle tracing)
Enter an integer value.
PINLET
specifies a pressure driven inflow. Setting PINLET to 1 indicates that all the
pressure boundary conditions are also inflow boundary conditions. Use of this
option allows one to avoid using thin filled regions at the inlets of pressure driven
problems. It allows for filling of metal without having an initial layer of fluid.
Enter an integer value of 0 (off) or 1 (on).
PLIMIT
specifies the pressure cutoff limit. You can use this parameter to turn off an inlet
velocity when the back pressure exceeds the given value. This is useful
particularly in cases where cold shuts are occurring. Otherwise, the program will
keep trying to force more mass into the fluid region, even though there is no place
for it to go, and the pressure will continue to rise.
Enter a floating (real) value.
Choose the pressure units from: {N/m**2 | Pa | KPa | MPa | bar | dyne/cm**2 |
atm | psia | Ksi | lb/ft**2}
PREF
specifies the pressure which is to be subtracted from any boundary condition
pressure in order to convert an absolute pressure into a gauge pressure. This
parameter comes into play when: (1) there is trapped gas, (2) a pressure boundary
condition drives the flow, (3) there are vents, and/or (4) there is gas injected. For
example, if the pressure boundary condition drives the flow at a gauge of 1
atmosphere, the boundary condition is set to 2 atm. PREF should be set to 1 atm.
Enter a floating (real) value.
Choose the pressure units from: {N/m**2 | Pa | KPa | MPa | bar | dyne/cm**2 |
atm | psia | Ksi | lb/ft**2}
PRELAX
specifies the pressure relaxation factor. PRELAX, to have an effect, should be
greater than zero and less than one. If it is left to the default value of one,
ProCAST will automatically compute an appropriate relaxation factor.
Enter a floating (real) value.
RELVEL
For centrifugal casting, the fluid flow should be solved in a Relative velocity
reference frame (i.e. in a rotating velocity reference frame). Thus, for centrifugal
casting cases, one should set RELVEL to 1. This Run parameter should be added
manually in the p.dat file (it does not appear in PreCAST).
0 = standard case - no centrifugal (default value)
1 = activation of the relative velocity reference frame - to be used for centrifugal
casting.
Enter an integer value.
TILT
For Tilt pouring problems, one should activate the "Tilt" mode, by setting TILT to
1. This improves the appearance of undesired sticking in the pouring cup.
This Run parameter should be added manually in the p.dat file (it does not appear
in PreCAST).
0 = standard filling mode (default value)
1 = activation of the Tilt mode
Enter an integer value.
TOFRSF2
TOFRSF2 allows to switch the filling algorithm from FREESF = 1 to FREESF =
2 automatically when the specified fill fraction is reached. This option should be
used for Large ingot filling (see the "Large Ingot filling" section for more details).
The fill fraction for the transition should be specified as a real value (e.g.
TOPFILL
TOPFILL is used for the top filling of large ingots (see the "Large Ingot filling"
section for more details). TOPFILL should be used in conjuction with a mass
source.
0 = deactivated (default value)
1 = activated
TPROF
This parameter defines the type of algorithms which are used to calculate the
temperature in the case of advection along a mold wall (i.e. in the case of fluid
flow along a wall). The two following situations are considered :
a) In order to account for thermal boundary layers at the mold wall (in the case of
fluid flow), the temperature of the nodes which are close to the wall (like Node C
in the figure below) can be computed in two different ways.
The temperature of Node C is a function of the temperature of the nodes
backwards (i.e. Nodes A and B which are behind - look at the direction of the
flow indicated by the green arrows). The temperature of Node C can be either a
linear interpolation between the temperatures of Nodes A and B (corresponding to
T0 in the figure below). This is the case with TPROF = 0.
In order to better account for the thermal boundary layer (corresponding to the red
temperature profile in the figure below), the temperature of Node C will be taken
as T1 (which is larger than T0 and thus closer to the temperature of Node A). This
is the case with TPROF = 1.
It is highly recommended to use TPROF = 1 which is much closer to the reality,
especially when the mesh is coarse. This algorithm is valid for all liquid nodes
which are in the elements next to the wall, but are not themselves on the wall.
b) Beside the algorithm described above, one should also compute the temperature
of the wall nodes which are just beyond the free surface (see Node D in the figure
below which is in the air at time t1 and which is at the free surface at time t2). The
green arrows are showing the direction of the velocity.
When TPROF = 0, the temperature of Node D is taken as the one of the closest
node, which may be either a core or wall node, (e.g. the one of Node C in the
example of the figure below).
When TPROF = 2, the temperature of Node D is taken as the closest surface node
(i.e. node B).
This algorithm is active only for the wall nodes which are at the free surface.
As a consequence, the free surface temperature at the mold wall will be "colder"
in the case of TPROF = 2 than with TPROF = 0.
One should also be aware that when TPROF=2 is used, the TPROF=1 algorithm
described above in a) will be deactivated. In order to have both algorithms active,
one should use TPROF = 3.
TSOFF
This parameter specifies the time at which to switch off the flow solution. For
example, TSOFF 1 42, indicates that the flow solution will be turned off 42
seconds into the simulation. If a cyclic analysis is being performed, then the flow
solution will be turned off 42 seconds into each cycle.
Choose from:
0 = turns this option off, or
a real value sets the time
Enter a floating (real) value.
Choose the time units from: {sec | min}.
VFREQ
specifies the time step interval for writing velocity and pressure results to the
unformatted files. This parameter can be used to reduce the size of these results
files, which can become quite large for problems with many nodes and time steps.
Note that it is only possible to restart a run from one of the time steps that was
written. Only the steps that are written can be viewed with post-processing.
Enter an integer value.
VPROF
This parameter indicates that a flow boundary layer profile is used at the wall for
the momentum equation with advection. This has been found to reduce false
diffusion errors, although it is not very useful with WSHEAR=2.
Choose from:
0 = do not use boundary layer profile, or
1 = use boundary layer profile
Enter an integer value.
WALLF
WALLF is used to compute the velocity of the free surface at the mold wall (not
used away from the free surface). In order to compute the velocity of the free
surface at the mold wall, one takes the core velocity of the closest core node (i.e.
closest node in the volume). The surface velocity at the free surface is equal to the
core velocity of this closest node multiplied by WALLF. This algorithm is applied
for any value of WSHEAR.
A value of 0.99 corresponds to more slip along the wall, whereas a value of 0.8
will act as if the mold surface is rougher (more friction). It is advised to use a
value of 0.8 for sand gravity casting and LPDC casting, a value of 0.9 for gravity
die casting and a value of 0.99 for high pressure die casting (HPDC).
The default value is 0.9.
WSHEAR
The WSHEAR algorithm (wall shear) allows to take into account a velocity
boundary layer along mold wall. It allows to have non-zero velocities at the mold
walls, which is more representative of the reality (slip of the liquid along walls).
This is valid at all liquid-mold interfaces and not only at the free surface (whereas
the WALLF algorithm is valid only at the free surface).
Physically, the velocity at a mold wall is zero (in the figure below, the velocity of
Node B should be zero, represented by the green dot). However, a velocity
boundary layer (see the red velocity profile) is present and the velocity close from
the wall is very quickly not zero anymore. As the mesh size is most of the time
larger (or much larger) than the velocity boundary layer thickness, the wall
velocity is computed in order to reflect the velocity just away from the wall (the
velocity will correspond to v1 shown with the blue dot in the figure below).
When WSHEAR = 0, the velocities at the mold wall are always zero. With
WSHEAR = 1 or 2, the velocities at the mold walls are computed in order to
account for the boundary layer. The difference between WSHEAR 1 and 2
corresponds to laminar or turbulent boundary layers (please note that the thickness
of the boundary layer is influenced by the viscosity of the alloy).
The figure below is showing a zoom on the boundary layer with the wall
velocities corresponding to different values of WSHEAR.
CMU
specifies the proportionality constant used in the turbulent viscosity equation.
Enter a floating (real) value.
CONE
specifies the proportionality constant used in the production of turbulent energy
dissipation.
Enter a floating (real) value.
CTWO
specifies the proportionality constant used in the destruction of turbulent energy
dissipation.
Enter a floating (real) value.
KAPPA
specifies the Von Karman's constant, usually taken as 0.4
Enter a floating (real) value.
SIGMAE
specifies the diffusivity modifier used in the turbulent energy dissipation transport
equation.
Enter a floating (real) value.
SIGMAK
specifies the diffusivity modifier used in the turbulent kinetic energy transport
equation.
Enter a floating (real) value.
TBRELAX
specifies the turbulence relaxation parameter.
Enter a floating (real) value.
TURB
specifies whether the turbulent flow model is turned on or off. A model started
with TURB = 1 can be restarted at a later time with TURB = 0. This allows
laminar conditions to be considered during mushy or natural circulation flows.
Once TURB has been set to zero, the turbulence model can not be restarted at a
later time. Setting TURB to one for a flow problem which has no turbulence
boundary conditions assigned is OK; the software will automatically define them.
Enter:
0 to turn the turbulent flow model off, or
1 to turn the turbulent flow model on
AVEPEN
AVEPEN corresponds to the "Average Penetration". This corresponds to the
maximum average penetration which is allowed during the calculation.
During the calculation, the PENALTY is automatically changed in order that the
penetration is not larger than AVEPEN. The goal is to have the lowest possible
PENALTY number to speed-up convergence, and AVEPEN allows to set the
upper limit.
The default value of AVEPEN is 0.1 mm. For large casting, it is advised to
increase the value of AVEPEN for faster convergence.
CLAYERS
CLAYERS operates in conjunction with GLOAD = 1. CLAYERS controls the
number of layers of nodes, starting from the wall, that have a zero displacement
constraint applied while the nodal fraction solid is less than CRITFS. If
CLAYERS = 1, for example, then once the wall nodes have solidified (fs >=
CRITFS), there will be no more zero displacement constraints applied
automatically to the casting. (Displacement BCs applied by the user will still be
active.) By contrast, if GLOAD = 0, then all the unsolidified nodes of the casting
will be constrained, which limits the effect of gravitational load.
Default value : 3
CRACK
The cracking indicator model is activated with CRACK = 1 or 3. By default, the
value of CRACK is set to zero. This Run parameter is not available in the Run
Parameters menu and it should be added manually in the p.dat file.
CRACK = 1 activates the cracking model, without feedback on the stress
properties
CRACK = 3 activates the cracking model, with a feedback on the stress
properties. In this case, the hardening properties are changed, according to the
cracking model.
CRITFS
This corresponds to the critical fraction of solid where the stress calculation starts.
By default, it is set to 0.5 (i.e. 50% fraction of solid). This critical value is also
used for the computation of the hot tearing.
CONVS
specifies the convergence criterion for the stress calculation.
Enter a floating (real) value.
CYCL_ALGM
During a stress calculation, together with cycling, "CYCLE_ALGM=2" is
resetting the original mesh at the beginning of each cycle. In such case, the
deformation is still accumulated. This method may prevent the creation of highly
distorted elements, (i.e. Negative Jacobians) during successive cycles, due to too
high accumulated distortions. If the calculation is able to go through the first
cycle, the calculation of the other cycles should be then possible without
problems. However, in the post-processing, we will still be able to see a mesh
which is distorting more and more along the cycles. The default value is 1, which
correspond to no special action and thus no re-alignment.
Enter an integer value (default : 1).
GAPMOD
specifies the treatment of the interface heat transfer coefficient. With a value of 1,
the interface heat transfer coefficient is automatically modified to account for the
air gap formation (an additional thermal resistance is computed as a function of
the gap width, taking into account conduction and radiation through the air or
vacuum). With a value of 0, the interface heat transfer coefficient which is defined
in PreCAST will not be modified during the calculation.
GLOAD
In the case of stress calculations, one should make sure that the casting is always
in contact with the mold in the direction of the gravity. Moreover, especially for
large heavy casting, it is better to take into account the casting weight on the
mold. The Run Parameter GLOAD allows to take these two effects into account :
GLOAD = 1 : in order to take into account the casting weight, the static pressure
head is calculated and applied as an internal boundary conditions in the mushy
zone (on the isosurface corresponding to the solid fraction defined by the Run
Parameter GLOADFS - default value .8). This option is useful only for heavy
castings. See the description of CLAYERS which is active in the case of GLOAD
= 1.
GLOAD = 2 : In order to guarantee that the casting is always in contact with the
mold, in the direction of gravity, GLOAD should be set to 2. The algorithm will
find the smallest gap in the direction of gravity between the casting and mold, and
then move the casting down by that amount. Once the casting is in contact or
penetrates the mold, this algorithm is deactivated. It is strongly advised to set
GLOAD = 2 in all cases where the casting is not fully surrounded by a mold.
GLOAD = 3 : activation of both models 1 and 2 described above.
Remark : Please note that even with a casting weighing several tons, the pressure
on the mold will be only a few MPa, to be compared with thermally induced
stresses which can go for instance up to a few hundreds MPa.
This Run parameter has to be added manually in the p.dat file.
Default value : 0
GLOADFS
As described above for GLOAD = 1, the static pressure of the casting is applied
on the GLOADFS isosurface. If GLOADFS < CRITFS then, GLOADFS is
automatically set to CRITFS.
Default value : 0.8
LOADSCL
When a load is applied in non-linear problem, it has to be applied incrementaly
(within a timestep). LOADSCL is the number of increments for this loading. The
higher the value of LOADSCL, the more accurate is the result (and the higher is
the CPU time). This has to be used mainly for structure analysis type of problems
(e.g. tensile test). For usual casting, as the loading is progressive (due to gradual
temperature changes), it is not necessary to apply such increments. The default
value of LOADSCL is 1. This value has to be added manually in the p.dat file.
PENALTY
PENALTY controls the level of "penetration" allowed by the "contact algorithm".
As the displacements are computed numerically at interfaces, there is always some
penetration between two bodies (as long as they are touching).
High values of PENALTY means that the allowed penetration is very small and
this leads to a more difficult convergence of the algorithm. Small values of the
PENALTY allows to have more penetration (which means a more easy
convergence).
The default value for Penalty is 1, but for thin sections, it could be advised to set it
to 0.01.
During the calculation, the PENALTY is automatically changed in order to
optimize the CPU time and in order to be within the AVEPEN limit.
When a calculation is restarted, it is advised to set the PENALTY to zero. In this
case, the last PENALTY which was used in the calculation will be automatically
set. If a non-zero value is defined, it will be used for the restart step. One should
note that if the PENALTY was decreased during the calculation, to leave the
default value may lead to a very long convergence at the restart step. Thus, it is
strongly advised to set it to zero for a restart. The PENALTY at each stress
timestep is indicated in the p.out file.
PENMOD
Parameter which allows to scale the penalty number according to the local
properties.
With PENMOD = 1, the local contact penalty number is scaled according to the
Young's modulus. Thus, in the mushy zone where this value is quite low, the
penalty number will also scale down. The penalty value of the p.dat file will be
used at Room temperature. Then, at high temperature, it will be scaled according
to Youngs modulus.
This results in a better balance between the contact force and the strength of the
material to sustain it at all temperature. With PENMOD = 1, it allows to increase
the global penalty number from the 1.e-2 (standard usual value) to 0.1 or even 1.0.
With PENMOD = 0, the standard model, with a global PENALTY number is
used. In such case PENALTY should be set to 1e-2 to 1e-3.
Enter an integer value (Default value : 0).
SCALC
specifies the time step interval the stress calculation is performed. It is thus
possible to perform the stress calculation only every 10 thermal steps. SCALC and
SFREQ are independent values and one will not affect the other.
Enter an integer value.
SFREQ
specifies the time step interval for writing stress results to the unformatted files.
This parameter can be used to reduce the size of these files, which can become
quite large for problems with many nodes and time steps. Note that it is only
possible to restart a run from one of the time steps that was written. SCALC and
SFREQ are independent values and one will not affect the other. One should
however be careful that SFREQ is a multiple of SCALC.
Enter an integer value.
STRESS
specifies whether the stress calculation is turned on or off.
Enter:
0 to turn the stress calculation off, or
1 to turn the stress calculation on
VACUUM
When GAPMOD = 1, the interface heat transfer coefficient depends upon the gap
width. With VACUUM = 0, air conduction is taken into account (i.e. a heat
resistance corresponding to heat conduction through air is taken into account).
With VACUUM = 1, no heat conduction is taken into account. In both cases,
radiative transfer through the interface is taken into account (when there is a gap).
EQNMAX
First nucleation parameter of the dendritic primary phase. Maximum density of
nuclei of the Gaussian distribution.
EQSTD
Second nucleation parameter of the dendritic primary phase. Standard deviation of
the Gaussian distribution.
EQUNDER
Third nucleation parameter of the dendritic primary phase. Average undercooling
of the Gaussian distribution.
EUNUCL
First nucleation parameter of the eutectic phase. Nucleation factor.
EUPOWER
Second nucleation parameter of the eutectic phase. Nucleation exponent.
EUGROW
Eutectic growth kinetics constant.
FADING
In the case of SGI, the melt is treated with Magnesium. Depending upon the
volume of the melt (in ladle) or the type of treatment (e.g. "in mold" treatment),
the Mg content can decrease with time (Fading effect), which will affect the
graphite nodules. This is leading to a decrease of the expansion. The FADING
Run parameter allows to calibrate the Fading effect. A value of 1 corresponds to a
maximum Fading (i.e. the expansion potential disappears totally after about 20
minutes). With a value of 0, no Fading is occurring (i.e. we have always the same
expansion potential). Between 0 and 1, we will have a reduced Fading effect (i.e.
the smaller the value of FADING, the longer the expansion will be occurring). See
the "Microstructures/Iron and Steel/Iron and Steel models" section for more
details).
FADING should be defined between 0 and 1. Default value : 1.0
GRAPHITE
In the micro model for cast iron, the amount of "graphitisation" can be adjusted
with the GRAPHITE Run parameter. A value of 1 corresponds to a maximum of
graphitisation, whereas a value of 0 corresponds to no graphitisation. The larger
the value of GRAPHITE, the larger the expansion of the cast iron will be. See the
"Microstructures/Iron and Steel/Iron and Steel models" section and the "Cast
Iron Porosity model" section for more details).
GRAPHITE should be defined between 0 and 1. Default value : 1.0
MGTREAT
For SGI, the graphite formation tends to diminish as the interval between the
inocculation and the beginning of the solidification increases. This is accounted in
the SGI micro model by the "fading" model (see the "Microstructures/Iron and
Steel/Iron and Steel models" section and the "Cast Iron Porosity model" section
for more details). In order to have an accurate "fading" model, it is necessary to
account for the right amount of time between the inocculation and the beginning
of the solidification. The value of MGTREAT corresponds to the time between
the inocculation and the beginning of the calculation (which corresponds to the
beginning of the filling or when the mold is full).
Default value : 0 seconds.
MICCPL
The Microstructure model can be run in "Coupled" or "Uncoupled" modes (see
the "Microstructures/Case set-up and Results" section for more details). This is
activated in the Micro Run parameter panel by the button on the right of the
MICRO Run parameter. This choice corresponds to the MICCPL Run parameter
in the p.dat file.
MICCPL = 0 (default value) corresponds to the "Uncoupled" mode.
MICCPL = 1 corresponds to the "Coupled" mode.
MICRO
Activation of the microstructure module. The module is activated with a value of
1.
PERGROW
Pre-factor of the Growth kinetics of the pearlite. The default value is 0.0168. This
value can be changed in order to calibrate the calculated amount of perlite
(respectively of ferrite) with experimental observations. If the value is decreased,
the amount of pearlite will also decrease, as the growth kinetics will be slower.
Please note that the pearlite/ferrite fractions also depend upon the nodule count.
Thus, it is advised to first calibrate the nodule count (EUNUCL, EUPOWER and
EUGROW) first and then to calibrate PERGROW (together with PERNUCL).
Default value : 0.0168
PERNUCL
Nucleation parameter of the pearlite.Pre-factor of the Nucleation model of the
pearlite. The default value is 5e6. This value can be changed in order to calibrate
the calculated amount of perlite (respectively of ferrite) with experimental
observations. If the value is decreased, the amount of pearlite will also decrease,
as the number of pearlite nodules will be lower. Please note that the pearlite/ferrite
fractions also depend upon the nodule count. Thus, it is advised to first calibrate
the nodule count (EUNUCL, EUPOWER and EUGROW) first and then to
calibrate PERNUCL (together with PERGROW).
Default value : 5e6
See the "Microstructures" chapter for more details about the models.
It is possible to select the default values for a given system with the "Select
Default Values" button.
When one clicks on the "Select Pre-defined Set" button, a list of the available sets
are displayed. Then, the user has to select the desired one. In the case below, six
sets are available, among ten possible sets.
The values contained in the Pre-defined sets correspond to the Recommended Run
Parameters for each process. It is thus strongly advised to use these pre-defined
sets or to customize them according to your specific processes (see below).
The configuration of the Pre-defined sets can be easily performed by the user. One
should just save the a "d.dat" file, with the desired Run Parameters in the
"dat/pref" directory of the installation, under the name "predefined_x_p.dat",
where x is a number between 1 and 10 (see below).
The label which appears in the Run Parameter window above should be included
in the predefined file as follows (TITLE line) :
ProCAST provides the access to many Run parameters, in order to allow the
treatment of all kind of situations.
However, for an everyday use, only a few Run parameters have to be set or
modified. This section is presenting the most "popular" Run Parameters that
should be set, with proposed values, for each main family of processes.
These recommended Run parameters correspond to the one which are pre-defined
in the "Pre-defined Run Parameters" window. It is thus advised to activate the
"pre-defined" set corresponding to the process and then to set the appropriate
stopping criteria.
Gravity casting
Timestep handling
DT = 1e-3
DTMAXFILL = 1e-1
DTMAX = 0.5 - 5 (depending upon the size of the model and thus the
solidification time)
Porosity
PIPEFS = 0.3
GATEFEED = 0
Filling
WSHEAR = 2
FREESFOPT = 1
WALLF = 0.8
LVSURF = 0.98
Tilt casting
Timestep handling
DT = 1e-3
DTMAXFILL = 1e-1
DTMAX = 0.5 - 5 (depending upon the size of the model and thus the
solidification time)
Porosity
PIPEFS = 0.3
GATEFEED = 0
Filling
WSHEAR = 2
FREESFOPT = 1
WALLF = 0.8
TILT = 1
DETACHTOP = 1
Centrifugal casting
Timestep handling
DT = 1e-3
DTMAXFILL = 1e-1
DTMAX = 0.5 - 5 (depending upon the size of the model and thus the
solidification time)
Porosity
PIPEFS = 0.3
GATEFEED = 0
Filling
WSHEAR = 2
FREESFOPT = 1
WALLF = 0.8
RELVEL = 1
POROSITY MODELS
POROS = 1
The POROS = 1 model corresponds to the latest porosity model of ProCAST. It
accounts for coupled micro and macroporosity, as well as pipe shrinkage. It can be
applied to both gravity casting and injection (either hpdc or lpdc).
POROS = 4
The POROS = 4 model corresponds to the same model as POROS=8 (see below),
with the additional treatment of multiple free surfaces for piping.
POROS = 8
The POROS = 8 model corresponds to the porosity model which was available in
version 3.2.0 of ProCAST (using the POROS=1 Run Parameter of v3.2.0).
Although this model is less sophisticated than the current POROS = 1 model, it
was kept in this version for the users who have calibrated the model to their
casting and who obtained good results in the past.
The only difference between the current POROS=8 model and the one available in
version 3.2.0 is the method used to compute liquid pockets. Thus, this may lead to
slight differences between the versions.
One should notice that even if POROS = 0 is set, piping will be still calculated
and displayed. In order to disable the piping calculation (in the case of a
THERMAL only calculation), one should set FREESF to zero.
In order to disable the porosity calculation, one should set POROS = 0. However,
this may still lead to some piping calculation. In order to disable also the piping,
one should set in addition PIPEFS = 0.
POROS=1
It is necessary to define the gravity vector in order to have meaningful results with
POROS=1.
Model
fraction equal to or lower than MACROFS, plus those nodes between the
MACROFS and MACROFS + FEEDLEN isosurfaces, is computed. This
shrinkage is then distributed according to the following scenarios:
a) Find the highest point of the region that is on a free surface and has a
fraction of solid lower than PIPEFS. In this case, piping occurs and the free
surface of the casting (usually the riser) goes down by the amount corresponding
to the shrinkage
(For display purposes, in the pipe, the shrinkage porosity value is set to 1 and
FVOL is set to a value below 0.5, so that it will exhibit piping, i.e. "empty
nodes").
b) Find the highest point of the region that is on a free surface and has a
fraction of solid higher than PIPEFS. In this case, macroshrinkage occurs at that
point. This will also correspond to piping (same result as a) above). However,
instead of showing an empty volume, it will have a "Shrinkage porosity" value of
1.
(For display purposes, in the macroshrinkage, the shrinkage porosity value is set
to 1 and FVOL is set to a value above 0.5, so that it will not exhibit piping, i.e.
"empty nodes", but shrinkage).
c) No nodes of the "region" are on a free surface with a fraction solid lower
than PIPEFS. In this case, no more piping can occur and macroshrinkage in the
bulk of the casting will appear. The macroporosity will appear at the highest most
liquid point of the region (e.g. if there is a pocket of liquid which is surrounded by
a mushy zone, itself surrounded by a solid shell, the macroporosity will start at the
highest point of the liquid pocket).
(For display purposes, in the macroporosity region, the shrinkage porosity value is
set to any value and FVOL is set to a value above 0.5, so that it will not exhibit
piping, i.e. "empty nodes", but shrinkage).
a) microporosity can appear only in the zone where the fraction of solid is in
between MACROFS and 1.
d) if FEEDLEN is set to a very large value (larger than the casting side), the
region between MACROFS+FEEDLEN is not existing anymore. Thus, no
microporosity is created until the whole pocket is above MACROFS (i.e. no
Symmetry planes are taken into account in the liquid pocket calculation (i.e.
symmetry planes "close" the pockets, even if there is liquid at the symmetry wall.
See the "Density definition" section for more details about how to define the
density.
See the "Active feeding" section for more details about feeding from the piston or
the ingate in the case of hpdc and lpdc.
POROS=4
Model
This model is based upon the same algorithm as the POROS=8 model (see
below). In addition, multiple piping can be considered. This means that piping
occurs at the highest free surface of each region. Thus, one can have piping at the
top of risers which are at different levels (this was not the case in version 3.2.0
and thus does not occur with POROS=8)
See the "Density definition" section for more details about how to define the
density.
POROS=8
Model
When the casting solidifies, pockets of liquid are created, surrounded by a mushy
zone and then a solid shell. As soon as a pocket of liquid is surrounded by a zone
which has a solid fraction higher than MACROFS, the density of each node inside
the pocket (from fs=1 to fs=MACROFS) is recorded (as a "critical density").
Then, the amount of porosity of each of these nodes is equal to the density
variation between this "critical density" and the density of the solid.
Piping is occurring at the highest free surface of the model. Thus, if there are two
separate regions with risers at different heights, only the highest one will exhibit
piping, even if the lower one is in an isolated region. POROS=4 corrects this
situation (see above).
Symmetry planes are taken into account in the liquid pocket calculation (i.e.
symmetry planes are "closing" the pockets, even if there is liquid at the symmetry
wall.
See the "Density definition" section for more details about how to define the
density.
Density definition
For most alloys, the density at the liquidus is lower than the density at the solidus,
thus leading to porosity. By default, the porosity module of ProCAST is using the
density curve in the mushy zone which is defined in the material database (see the
green curve in the Graph A below). However, if this density change in the mushy
zone is not well known, it is possible to automatically calculate the density in the
mushy zone, as an average of the liquid and solid densities, weighted by the
fraction of solid. The density of the liquid and of the solid is calculated as a
function of temperature, by extrapolating the density slopes at the liquidus and
solidus respectively.
The user has the possibility to activate this automatic density computation with
the Run parameter USERHO. If USERHO is set to 0 (instead of 1 which is the
default value), the density curve which is defined in the material properties,
between the solidus and liquidus, will be ignored and the automatic density
computation will be activated. (Please note that in version 4.x.x, the default value
was 0).
Active Feeding
In the case of injection (e.g. high or low pressure die casting), for a while, the
shrinkage is compensated by the piston in the case of hpdc and by the liquid bath
for lpdc, thus leading to no piping.
ProCAST can account for such "active feeding", by setting the Run parameter
GATEFEED=1. In this case, no piping will occur, but liquid will feed the ingate,
as long as the fraction of solid is lower than GATEFS (i.e. there is feeding in all
regions which are within the same GAFEFS isosurface as the gate). The gate is
defined as being the region where in inlet velocity is applied, or where a pressure
is applied (or where GATENODE is applied).
One should note that the porosity level will NOT depend upon the value of the
pressure. The Active feeding is there only to compensate for the shrinkage of the
region in contact with the piston, as long as the fraction of solid is lower then
GATEFS (i.e. within the same GATEFS isosurface as the gate). This does not
correspond fully to a real third stage pressure, however, the value of GATEFS
could be adjusted with pressure if desired).
For HPDC and LPDC, it is advised to set PIPEFS = 0.0 in order to prevent piping
at the top of the casting, if the gate is closing too early.
In the case of a thermal only calculation (for HPDC or LPDC), one should set a
Pressure boundary condition at the ingate (in order to activate the active feeding)
and one should disable the FLOW run parameter (as normally a Pressure BC
would automatically switch ON the flow solver). Of course, GATEFEED should
also be set to 1 in this case.
Warning :
In the case of a thermal only calculation, a Pressure BC should be set at the ingate
in order to switch ON the active feeding. In such a case, the Pre-processor
automatically is activating the flow solver (FLOW=3). As a consequence, one
should set FLOW=0 manually in the d.dat file in order to deactivate the flow
solver (but the porosity calculation will be performed with active feeding).
In order to account for such expansion, the POROS=1 model was adapted in the
way describe hereafter.
In the "pockets" of liquid and mushy zone which have a solid fraction lower than
MACROFS (+FEEDLEN), the integral of the density change at each timestep is
performed. In the case of expanding material, some location will expand and some
will shrink. If the total density change corresponds to a net shrinkage, the
"regular" model applies (i.e. this macroshrinkage will occur as the highest most
liquid point). If it corresponds to a net expansion, two scenario may occur :
a) Find the highest point of the region that is on a free surface and has a fraction
of solid lower than PIPEFS. In this case, the free surface of the casting
(usually in the riser) will go up by the amount corresponding to the net
expansion.
b) Find the highest point of the region that is on a free surface and has a fraction
of solid higher than PIPEFS. The amount of expansion is applied
proportionally to all of the nodes in the region that have pre-existing porosity
(thus, the macroporosity which appeared earlier, when there was a net
shrinkage, will be "refilled"). For example, if the amount of expansion is
enough to "refill" 50% of the total porosity in the region, then the porosity of
each node is reduced by 50%.
c) No nodes of the "region" are on a free surface with a fraction solid lower than
PIPEFS. The expansion is distributed to all the nodes in the region as
described in b).
d) If FEEDLEN is set to a very large value (larger than the casting size), the
region between MACROFS+FEEDLEN is not existing anymore. Thus, no
microporosity is created or refilled until the whole pocket is above MACROFS
(i.e. no microporosity created or refilled according to b1) above). Then,
microporosity can start to occur or disappear, according to b2) above.
Mold rigidity
The rigidity of the mold has an influence on the amount of porosity in the case of
expanding alloys. If the mold is totally rigid, the casting can not expand and thus
the alloy expansion will be "available" for the "refill" of the existing porosity. On
the other hand, if the mold is very soft (or weak), the casting will expand and thus
there will be no "refill" of the porosity (of course, the reality is more complex as
the solid shell is thick enough it will act as a "rigid" mold, even if the sand mold is
weak.
In order to take the mold rigidity into account, the Run Parameter MOLDRIG is
introduced. MOLDRIG should be defined by a value between 0 and 1. All the net
expansion is multiplied by MOLDRIG. Thus, with MOLDRIG=1, corresponding
to a rigid mole, the expansion will be fully accounted. On the other hand, no
expansion will be taken into account if MOLDRIG=0. The expansion will be
compensated by the mold movement because the mold is too weak to hold the
expansion in this case. For real situations, the value of MOLDRIG should be set
somewhere between 0 and 1 depending upon the casting processes. MOLDRIG
should be added in the "Thermal Run Parameters" tab. The default value is 1.
Density curve
For expanding materials, the density defined in PreCAST should be not anymore
monotonic. The density can increase (locally) with increasing temperature.
One should note that such density curve will not be obtained when the material
properties are computed with the Computherm database. This is because the
expansion is depending upon the microstructure which itself depends upon the
cooling rate (see next section). Thus, such curve should be defined manually by
the user, based upon experiments.
The following figures are showing the solidification of a simple SGI casting
calculated with the MICRO model. The corresponding cooling curves and
densities (calculated by the micro model) are also shown. One can well see that
depending upon the cooling rate, one can have full expansion to full shrinkage,
with intermediate behavior in between.
Please refer to the "Microstructures/Iron and Steel/Iron and Steel models" section
of the manual for the description of the influence of GRAPHITE, FADING and
MGTREAT on the densities and thus on the porosity.
Example
The figures hereafter are showing an example of an expanding alloy. One can see
very well that the level of the metal in the left riser (which has an insulating
sleeve) is first going down (overall shrinkage) and then it is going back up
(overall expansion).
The following figure is showing an enlargement on the riser where the liquid level
is going down and then up.
VIRTUAL MOLD
ProCAST offers the capability of modeling a mold without meshing it, with the
Virtual Mold option. This is especially useful in the case of large sand casting. It
can also be used in permanent mold casting, if one is mainly interested in the
filling behavior.
When a Virtual mold is used, one should define the dimension of the mold (which
is an orthogonal box, aligned with X, Y and Z, the material properties of the mold
and the interface heat transfer coefficient between the different part of the casting
and the mold.
The Virtual mold model considers that the thermal diffusion is occurring in the
mold, according to "half diffusion distances". Thus, the model calculates for each
face of the casting what is the half diffusion distance. This distance is either the
one between the face and the limit of the mold or half the distance between the
face and an other face in front. At the limit of the mold, an adiabatic boundary
condition (i.e. no flux) is considered. Thus, it is advised to defined a "box" which
is large enough, in order not to "saturate" the virtual mold.
A good start for the Virtual Mold definition is to look at the minimum and
maximum size of the model in the "Geometry/Check Geom/Min-Max" menu.
Then, one should open the "Geometry/Virtual Mold" menu, to get the following :
The size of the Virtual Mold box can be defined in the following fields. One has
the possibility to define automatically the virtual mold box size, using the "Default
Size" button. One should first select on the right, whether the virtual mold box
should be 1, 2, 3, 4 or 5 times larger than the part, in the X, Y and Z directions
respectively. For small components, it is advised to use a virtual mold 5 times the
component size. For large castings, a virtual mold with a size of 2 o r 3 times the
component size will be enough. In case there is any doubt, it is advisable to take a
large virtual mold rather than the reverse, in order to avoid saturation.
Please note that one should NOT use the real size of the mold for the Virtual mold
computation. As the model considers that there is no cooling of the sides of the
virtual mold, the box should be large enough in order to avoid satuaration of the
mold (and thus too slow cooling).
The size of the Mold box can be visualized with the "Show Mold" button (in the
case of the figure hereafter, the virtual mold of a size 1 time larger than the
component was selected, for viewing purposes).
When the Mold size is good, one can "compute" the Virtual Mold, using the
"Compute Mold" button. Please note that the computation may take sometime,
depending upon the size (i.e. the number of surface elements) of the model. The
computation time is independent upon the size of the mold box itself (in the case
of the figure hereafter, the virtual mold of a size 1 time larger than the component
was selected, for viewing purposes).
Once the Virtual Mold is computed, the "Thermal depth" can be visualized with
the "Show Depth" button. The color scale can be changed with the "Set Scale"
button, which opens the following panel :
Finally, if needed, it is possible to erase the Virtual mold, using the "Remove
Mold" button.
If the mesh is made out of several material domains, it is mandatory to define first
the interfaces between the different domains and then to generate the Virtual
mold. If interfaces are changes after the Virtual mold creation, it will be
destroyed.
Once the Virtual Mold is generated, it appears in the Material list in the Material
properties assignment. One should assign the desired material properties to this
Virtual mold domain.
In the same way, the interfaces between the Virtual Mold and the different
material domains are automatically generated. These are labeled "Virtual". One
should assign an interface heat transfer coefficient to each "Virtual" interface.
FILTERS
In the Interface menu, the interfaces between the casting and the filter material (on
both sides) should be kept as "EQUIV" (i.e. no interface should be created
between the casting material and the filter).
Finally, the interface heat transfer coefficient between the liquid metal (casting
material) and the filter material can be defined in the "Boundary
Conditions/Assign Volume/Filter Heat" menu. The list of the Filter materials is
shown, as well as the interface heat transfer coefficient database, ready for an
assignment.
EXOTHERMIC
The "standard" properties (thermal conductivity, density and specific heat) should
be defined as usual (1) for the sleeve material. In addition, the exothermic
properties should be defined in the "Exothermic" tab (2). The Exothermic energy
(3) corresponds to the amount of energy which is generated during the burning of
the sleeve. The Ignition Temperature (3) corresponds to the temperature at which
the exothermic reaction is initiated.
The "burning kinetics" is defined in the table (4-5), as a fraction of burning, which
is a function of time. Once the exothermic reaction is started (i.e. when the
temperature is going above the ignition temperature), the exothermic energy will
be released according to the burnt fraction.
CYCLING
ProCAST allows to model the cycling sequence in the case of die casting (i.e. to
account for the heating of the die mold during the first cycles of casting). The
figure below shows the principles of cycling, as well as the different sequences.
In order to model cycling, one should be able to take into account the fact that
when the mold is closed, there is an interface heat transfer between the casting and
the mold and when the mold is opened, one should consider a cooling of the die
and of the casting with the ambiance. One should thus switch between an interface
heat transfer coefficient and two "Heat" boundary conditions.
The above diagrams are showing in green the interface heat transfer coefficients
as a function of time for both the interface between the casting and the fixed die
(solid green line) and between the casting and the mobile die (dashed green line).
One could see that the interface heat transfer coefficient goes to zero when the
casting and the mold are not anymore in contact. On the reverse, as soon as the
mold opens, a "Heat" boundary condition should be set (the red curve on the
above figure corresponds to the heat transfer coefficient and the dashed blue curve
corresponds to the ambient (or external) temperature. One could see on the above
figure that during the spray sequence, the heat transfer coefficient increases and
the spray temperature is also taken into account.
When the die is closed, before the next filling, the interface heat transfer
coefficient between the dies is again activated and the interfaces between the
casting and the die(s) are set to zero.
One should note that the Pre-defined Run parameters for HPDC-cycling are not
designed in order to peform a Porosity calculation during the cycling calculation.
If this is desired, one should set to :
POROS 1
PIPEFS 0.
GATEFEED 1
Then, a pressure BC should be set at the ingate in order to trigger Gate feeding.
LOST FOAM
The cavity is filled with foam and is surrounded by a sand mold. During the
filling, the hot liquid metal is heating up the foam which is burning, leaving the
space for the liquid metal.
In the Lost Foam process, the filling is controlled by the rate of burning of the
foam, as well as the escape of the gas through the ingate or through the permeable
sand mold.
Firstly, the geometry should contain at least three components : a part of the
downsprue which is empty, the cavity filled with the foam and a sand mold (see
figure below).
The following figure shows how the materials should be set. The downsprue
should be assigned with the casting material and it should be set as EMPTY=YES.
The cavity, which is filled with foam should be assigned with a "Foam" material
(EMPTY=NO) and the mold should be set with a sand material (EMPTY=NO).
There are no special requirements for the material properties definition of the
casting material.
For the Foam material, the thermal conductivity, the density, the specific heat and
the latent heat (of burning) of the foam should be specified. Moreover, under
Tsolidus and Tliquidus, one should specify the temperature at which the foam
starts and ends to burn respectively. One should note that the burning kinetics
(and thus the filling rate) is influenced by the density, the specific heat, the latent
heat and the burning temperature. The larger these values, the slower the burning
kinetics will be. This is due to the fact that the foam should be heated up by the
liquid metal and that if these quantities are larger the amount of heat required to
burn the foam will be larger and thus it will take more time.
For the sand, in addition to the usual thermal properties, it is necessary to define
it's permeability (in the "Fluid/Permeability" tab). Typical values range from 1e-6
to 1e-7 cm**2.
The interface between the casting material and the foam material should be set as
"EQUIV" (i.e. no interface), whereas the other interfaces should be set as
"COINC".
In addition the usual "Heat" boundary condition (to cool down the outside of the
mold) and the inlet Temperature, one should set two Pressure BC. One for the top
of the down sprue (on the whole surface) and the other for the outside of the mold.
It is recommended to set a value of 1 atm. on the outside of the mold and a value
slightly larger (e.g. 1.05 atm) on the top of the downsprue.
Additional Run parameters should be set in the p.dat file for Lost Foam :
FOAMHTC 0.02
FOAMHTCMAX 0.25
BURNZONE 1.0
GASFRAC 0.1
DIAG 262144
These parameters are regulating the rate of transfer of heat between the liquid
metal and the foam. When the liquid metal front is at a distance of "BURNZONE"
from the foam, the "interface heat transfer" between them is equal to the value of
FOAMHTC/BURNZONE. This heat transfer is increasing as the liquid get closer
from the foam. When the liquid is "touching" the foam, a maximum heat transfer
coefficient (equal to FOAMHTCMAX) is set. FOAMHTC and FOAMHTCMAX
are defined in the p.dat file in CGS units and BURNZONE is in "cm".
BURNZONE corresponds to the "usual" distance between the foam and the liquid
metal. One should be careful that the mesh size must be finer than the value of
BURNZONE. One should note that the burning kinetics will change when the
value of BURNZONE is changed (as the interface heat transfer coefficient is
equal to FOAMHTC divided by BURNZONE).
As the rate of heat transfer between the foam and the liquid metal is not well
known, it may be needed to calibrate the filling time with the values of
FOAMHTC and FOAMHTCMAX. To shorten the filling time, one should
increase these values.
In order to allow for smooth Restarts, the Run Parameter DIAG should be set to
262144. This setting is not necessary if no Restart is performed.
THIXO CASTING
ProCAST has a dedicated model for thixo casting (or semi-solid casting).
In order to account for such a behavior, a specific "Power cut-off" model was
designed.
In order to set-up a Thixo casting case, one should define the appropriate material
properties of the casting material and define an additional simple input file
(prefixg0.dat).
In the Material properties definition, all the thermal properties should be defined
as usual. In the Fluid tab, the Viscosity should be defined with the "Power-Cutoff"
tab.
The values of the "Zero viscosity", the "K Factor" and the "Power", which can be
Temperature dependant, should be calibrated with experiments. Please note that
the value of n should be negative (in order to have a decreasing viscosity with an
increasing shear rate).
The principle of the Thixo casting model is to divide the mesh in different regions.
For each region, it is possible to define the critical "cut-off" value of the shear rate
The format of this ASCII file is the following (one line per domain of the mesh)
:
Domain_number critical_shear_rate
The following example shows a g0.dat file for a case with 4 domains. One should
note that domain 2 is the mould and a dummy value (of 1) should be set. The units
of this critical cut-off shear rate is [s].
The principle for the definition of the critical cut-off shear rates is described
hereafter. It is based upon the fact that dendrites are broken in locations where the
shear rate is high and then, even if the shear rate is decreasing afterwards (in an
opened cavity), the viscosity will not increase anymore (as the dendrites have
been broken). This means that the viscosity in a cavity which is following a gate
will remain at a rather low level, corresponding to the shear rate of that gate. One
can illustrate this principle with the following figure.
In domain 3, the viscosity of the semi-solid material is about the one which
corresponds to the shear rate that was encountered in the gate (i.e. domain 2).
Thus, one should set a "cut-off" value in domain 3 which corresponds to the mean
shear rate of domain 2. In the same way, the "cut-off" value of domain 5 should
To determine the shear rate in a gate, one can view the "Non-Newtonian Shear
Rate" in ViewCAST.
Finally, in order to activate the Power cutoff model, the two following Run
parameters should be set :
HIVISC = 2
NNEWTON = 2
CENTRIFUGAL CASTING
The centrifugal casting model of ProCAST is designed for vertical centrifugal
casting only. The inlet should be centered, with inlet velocities parallel to the
rotation axis.
Starting from version 2006.0, centrifugal casting cases should be set-up in the
following way.
Only the CASTING domains should be set with a revolution velocity (i.e. not the
mold materials). This must be defined in the "Process/Assign Volume" menu,
with a "Revolution".
The Run Parameter RELVEL should be set to 1. This Run parameter is activating
the resolution of the fluid equations in a "Relative velocity reference frame" to
handle the rotation.
The rest of the set-up should be done as for a standard gravity casting (i.e.
WSHEAR = 2, WALLF = 0.8, FREESFOPT = 1 (or 2)).
When a Centrifugal casting case is run with the RELVEL algorithm (relative
rotational frame), the velocity vectors are including the rotational component. As
a consequence, it is not convenient to view these velocities. In order to view the
velocity vectors as if the observer would be attached to the rotating casting, one
should calculate the "Relative velocities" with the "Action/Relative velocity"
menu :
This will create a set of relative velocity files, named prefixru.unf, prefixrv.unf
and prefixrw.unf. In order to view these results, it is necessary to change manually
the name of these files to prefixu.unf, prefixv.unf and prefixw.unf (which will
substitute the existing velocity files having the same names - of course, it is
advised to save these files before this substitution, especially if a restart is
planned). In future version, it will be possible to automatically view both type of
results.
The figure below is showing the "standard" velocity vectors (including the
rotational component).
The figure hereafter is showing the same result as above, but with the relative
velocities.
On the figure below, a zoom on one branch of the above geometry is shown. On
the left, the "standard" velocities (including the rotational component) are shown,
whereas on the right, the "relative velocities" are displayed.
In order to reduce as much as possible the CPU time, the following procedures are
proposed.
Top Filling
The recommended procedure for Top filled ingot is the following :
A "Mass source" should be set at the bottom of the casting cavity, with the
appropriate amount of inlet metal and the desired inlet temperature. The
coordinate of the mass source should be approximately the bottom of the ingot,
but it may be a few millimeters or centimeters above.
Then, the Run parameter TOPFILL = 1 should be added manually in the p.dat file,
together with FREESF = 2. To Deactivate this Top filling mode, TOPFILL should
be set to 0.
This setting will have the effect that, instead of having a point mass source (as
usual), the incoming mass will be evenly distributed just below the whole free
surface (which is moving upwards). This will have the advantage of not having a
large velocity at the mass source location, which will require the use of small
timesteps.
Bottom Filling
The case of Bottom filling is more complicated as one needs to fill first the
downsprue(s) and then to bring the hot metal from the bottom.
In order to do so, one should set the Run Parameter TOFRSF2 followed by a fill
fraction (e.g. TOFRSF2 0.01), together with FREESF = 1. The filling will start in
FREESF = 1 mode (i.e the normal filling mode). When the critical fill fraction
defined by TOFRSF2 is reached, the calculation will switch automatically from
FREESF = 1 to FREESF = 2 for the rest of the filling. However, the hot metal will
well be brought at the bottom of the ingot through the downsprue(s). The
TOFRSF2 Run parameter should be added manually in the p.dat file.
The switch to FREESF = 2 will allow larger timesteps and thus it will reduce the
CPU time. One should however be aware that if the downsprue diameter is rather
small the velocities can be quite large (as well as at the exit of the downsprue on
the casting) and this will limit the size of the timestep, thus leading to important
CPU times.
General rules
MLDUPDT
When a mold is present, it is possible to reduce slightly the CPU time by updating
the material properties in the mold every N step. To do that the MLDUPDT Run
parameter should be added manually in the p.dat file, followed by an integer. This
number corresponds to the material properties update frequency. A value of 50
may be recommended. The MLDUPDT Run parameter should be added manually
in the p.dat file.
Boundary Layers
When large castings are modeled, the mesh size becomes very large with respect
to the thermal gradients, especially in the early stages of cooling. In order to have
appropriate answers (i.e. more accurate temperatures), it is advised to generate a
few layers (of a few mm in thickness) inside the casting, as well as inside the
mold.
ProCAST has the capability to handle non-coincident meshes (see figure below).
If the different domains are meshed separately, they can be loaded in PreCAST
using the "File/Multiple Meshes" menu.
Once the meshes are loaded, the set-up of the case in PreCAST should be done as
for conventional meshes, except for the definition of the interfaces (see figure
below). Firstly, all the non-coincident interfaces, where at least one element is
coincident on each side, are automatically listed in the interface list. One should
change the "Type" from EQUIV to NOCOIN (1). Then, for the non-coincident
interfaces which are not listed, one should "create" manually these interfaces, with
the Add button (2). This will open the"Add Interface Pair" window, where the
material numbers of each side can be defined (3). The "Master-Slave" concept is
used in the calculation of the heat transfer across the non-coincident interface. It is
advised to set the casting as the Master and the mold as the Slave (although it does
have only a minor importance on the accuracy of the results). Then, one should
check the "Non-coincident Interface" button (4). Once all the interfaces are
defined, one can "Apply" the whole selection (5), which will create the
appropriate interfaces.
The "In-plane Tolerance" corresponds to the maximum distance between the two
surfaces in order to have a contact (distance normal to the plane of the interface).
By default, the In-plane and Perimeter tolerances are taken as a fraction of the
smallest edge of the whole mesh. Thus, if the mesh has a quite heterogeneous
mesh size, these tolerances may be too small (and thus, there will be "no contact"
and thus no heat transfer at these non-coincident interfaces). If such a case occurs,
one can change (i.e. increase) these tolerances. One should however be careful not
to use too large tolerances so that nodes beyond the opposite surface will be taken
into account. A good tolerance value should be about half of the mesh size of the
corresponding surface.
GEOMETRY MANIPULATION
The geometry can be manipulated and can be displayed with different modes,
using the following icons.
Manual rotation of the model (it opens the Rotate window). Otherwise,
the model can be rotated interactively with the mouse at any time (as long as the
Center or Drag icon is not activated). If the model is rotated interactively with the
mouse, while the "Shift" key is pressed in the same time, the model is rotating
along the axis perpendicular to the screen (the mouse should be rotated
horizontally).
Center of the model (the location of the model which is selected will
move to the center of the screen)
MESH OPTIMIZATION
The speed of a FEM calculation depends upon the structure of the mathematical
matrices which have to be solved. This depends upon the way the FEM mesh is
numbered (i.e. which nodes belong to which element and vice-versa).
It is possible to optimize the FEM mesh numbering in order to minimize the CPU
time. This was done previously in MeshCAST. However, as the addition of
interfaces is creating new nodes, it was necessary to know already in MeshCAST
which interfaces had to be created.
As the optimization operation may be rather long for large meshes, it is not done
by default at every save of the model. In order to activate the optimization, one
should set it in the "File" menu :
This optimization will be done before exiting PreCAST (it is not performed with a
"Save" or "Save as" operation, but in this case a warning is displayed upon Exit).
One should note that if an optimized model is loaded again in PreCAST, it is not
necessary to optimize it again, as long as the interface settings are not changed.
USER FUNCTIONS
ProCAST allows the use of User Functions. User Functions can be used to define
boundary conditions in a very flexible way, such as combined time, temperature
and/or space dependant heat transfer coefficients.
Please note that the User functions usage requires the corresponding license (not
included in the base package).
• Time
• Local temperature
• Local fraction of solid
• Local coordinates
• Material number
• Boundary condition ID
This allows to define conditions as a function of any of the above arguments (e.g.
a time- and space-dependant interface heat transfer coefficient).
In addition, "external functions" allow to get the value of the main fields (e.g.
temperature, fraction of solid, velocities, ...) at any location of the model. This
allows to define for instance a heat transfer coefficient as a function of a "control
temperature", corresponding to a given location, somewhere else ("remote
control").
In addition, it is possible to get the node number by giving the coordinates, using
the following function :
where xin, yin and zin are the coordinates specified by the user, xout, yout and
zout the coordinates of the closest nodes which corresponds to the returned node
number. domain# is the domain number in which the search node should be
found.
On Windows, the compiler and linker of the Microsoft Visual C++ 6.0 package
has been used to create the executables. An "reduced" version of this compiler and
linker, called "Microsoft Visual C++ 2008 Express Edition" is available for free
on the web (information valid in January 2009). It can be downloaded from the
following link and then installed :
If this link is not anymore valid, search in the Microsoft official site.
In order to use "Microsoft Visual C++ 2008 Express Edition", one should
"manually" set the following :
At about 10 lines from the top of the file, there is a line for the tool kit variables
initialisation :
@call "C:\Program Files (x86)\Microsoft Visual Studio
8\VC\vcvarsall.bat" %2
On Windows, the compiler and linker of the Microsoft Visual C++ 6.0 package
has been used to create the executables. An "reduced" version of this compiler and
linker, called "Microsoft Visual C++ 2008 Express Edition", together with
"Windows SDK for Windows Server 2008" and ".NET Framework 3.5", are
available for free on the web (information valid in January 2009). They can be
downloaded from the following link and then installed :
Windows SDK for Windows Server 2008 and .NET Framework 3.5
http://www.microsoft.com/downloads/details.aspx?FamilyID=e
6e1c3df-a74f-4207-8586-711ebe331cdc&displaylang=en
If these links are not anymore valid, search in the Microsoft official site.
In order to use "Microsoft Visual C++ 2008 Express Edition", one should
"manually" set the following :
At about 10 lines from the top of the file, there is a line for the tool kit variables
initialisation :
@call "C:\Program Files (x86)\Microsoft Visual Studio
8\VC\vcvarsall.bat" %2
:amd64
if not exist "%~dp0bin\amd64\vcvarsamd64.bat" goto missing
call "%~dp0bin\amd64\vcvarsamd64.bat"
goto :eof
to :
:amd64
if not exist "%~dp0bin\vcvars64.bat" goto missing
call "%~dp0bin\vcvars64.bat"
goto :eof
Linux
Then, when the "procast" executable is launched (either from the Manager, or
"manually" in a Command window), the User Functions are automatically
compiled and linked. A local DLL is created, as well as a local executable. Then,
this executable is automatically launched and the calculation starts. This DLL and
local executables are automatically deleted at the end of the execution (if the
calculation is stopped manually or if it crashes, these files will remain in the local
directory).
Units
ProCAST allows to define the inputs with almost any kind of units. In order to
define which units should be taken into account into the software, two possibilities
are provided to the user.
1. Default units can be specified in an installation file (either the main central
installation or in the local user preference file (see the "customized
installation" section for more details). These units will be used in all the cases
run by the user. This ASCII file is called "UserFunctions_units.dat" and is
located in the "dat/pref" directory.
2. Specific units can be used for a given case. These specific units should be
specified in a text file in the local execution directory. This ASCII file is
called "prefix_units.dat".
In both cases, the same units should be used in all the User Functions of the same
case.
Each type of variable available in user functions are mentioned in the above list.
Then, a unit code is following (as an integer value). The above values correspond
to the default units (set at the installation) used by the user routines of the
ProCAST solver (SI Units and degree Celsius).
One can change these units, with a text editor, using the following nomenclature
(corresponding to the standard unit codes of ProCAST, used in the d.dat file) :
temperature
1 = Kelvin
2 = Celsius
3 = Fahrenheit
length
1 = m
2 = cm
3 = mm
4 = ft
5 = in
velocity
1 = m/s
2 = cm/s
3 = mm/s
4 = ft/s
5 = in/s
6 = m/min
7 = cm/min
8 = ft/min
8 = in/min
time
1 = sec
2 = min
heatflux
1 = W/m2
2 = cal/cm2/sec
3 = cal/mm2/sec
4 = Btu/ft2/sec
5 = Btu/in2/sec
6 = cal/cm2/min
7 = Btu/ft2/min
8 = Btu/in2/min
heattransfercoefficient
1 = W/m2/K
2 = cal/cm2/C/sec
3 = cal/mm2/C/sec
4 = Btu/ft2/F/sec
5 = Btu/in2/F/sec
6 = cal/cm2/C/min
7 = Btu/ft2/F/min
8 = Btu/in2/F/min
massflowrate
1 = Kg/sec
2 = g/sec
3 = lb/sec
4 = Kg/min
5 = g/min
6 = lb/min
thermalconductivity
1 = W/m/K
2 = cal/cm/C/sec
3 = cal/mm/C/sec
4 = Btu/ft/F/sec
5 = Btu/in/F/sec
6 = cal/cm/C/min
7 = Btu/ft/F/min
8 = Btu/in/F/min
#include <stdio.h>
#include <stdlib.h>
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_convehtransfer(char*, int, real, real, real, real, real,
real, int);
#else
real func_convehtransfer(char*, int, real, real, real, real, real, real, int);
#endif
/*
* convective heat transfer coefficient (applied on external surfaces)
*/
real func_convehtransfer(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* boundary condition ID number */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
exit(1);
#include <stdio.h>
#include <stdlib.h>
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_texternal(char*, int, real, real, real, real, real, real,
int);
#else
real func_texternal(char*, int, real, real, real, real, real, real, int);
#endif
/*
* ambient temperature (applied on external surfaces)
*/
real func_texternal(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* boundary condition ID number */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
Emissivity Function
#include <stdio.h>
#include <stdlib.h>
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_emissivity(char*, int, real, real, real, real, real, real,
int);
#else
real func_emissivity(char*, int, real, real, real, real, real, real, int);
#endif
/*
* emissivity coefficient (applied on external surfaces)
*/
real func_emissivity(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* boundary condition ID number */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
#include <stdio.h>
#include <stdlib.h>
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_heatflux(char*, int, real, real, real, real, real, real,
int);
#else
real func_heatflux(char*, int, real, real, real, real, real, real, int);
#endif
/*
* heat flux coefficient (applied on external surfaces)
*/
real func_heatflux(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* boundary condition ID number */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
This function can also be used for the definition of the gap properties (if one
wants to use something else than air or vacuum). See the "Tips&Traps/Gaps in
Stress models" section for more details.
#include <stdio.h>
#include <stdlib.h>
#include "common_dll.h"
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_interhtransfer(char*, int, real, real, real, real, real,
real, int);
#else
real func_interhtransfer(char*, int, real, real, real, real, real, real, int);
#endif
/*
* interface heat transfer coefficient
*/
real func_interhtransfer(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* boundary ID number */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_masssourceflowrate(char*, int, real, real, real, real, int);
#else
real func_masssourceflowrate(char*, int, real, real, real, real, int);
#endif
/*
* mass source flow rate coefficient
*/
real func_masssourceflowrate(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numMat) /* number of the domain */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_xmasssource(char*, int, real, int);
#else
real func_xmasssource(char*, int, real, int);
#endif
/*
* masssource vector of a domain : x - component
*/
real func_xmasssource(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real time, /* current time */
int numMat) /* number of the domain */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
#include <stdio.h>
#include <stdlib.h>
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_xtranslation(char*, int, real, int);
#else
real func_xtranslation(char*, int, real, int);
#endif
/*
* translation vector of a domain : x - component
*/
real func_xtranslation(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real time, /* current time */
int numMat) /* number of the domain */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_vximposed(char*, int, real, real, real, real, real, real,
int);
#else
real func_vximposed(char*, int, real, real, real, real, real, real, int);
#endif
/*
* ambient temperature (applied on external surfaces)
*/
real func_vximposed(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* number of boundary condition */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_vxsolidtransport(char*, int, real, real, real, real, real,
real, int);
#else
/*
* ambient temperature (applied on external surfaces)
*/
real func_vxsolidtransport(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* number of boundary condition */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return 0;
}
External Function
This function is called at the start of the calculation, at the start of each timestep,
at the end of each timestep and at the end of the calculation.
Many parameters are available in this function (see the "extern" declarations).
These parameters can also be used in the other user functions (e.g.
interhtransfer.c), as shown at the end of this section.
#include <stdio.h>
#include <stdlib.h>
#ifdef WIN
#define EXPORT _declspec(dllexport)
EXPORT void func_externalcompute(char*,int,int,real);
EXPORT int tnodecontrol;
EXPORT int vnodecontrol;
#else
void func_externalcompute(char*,int,int,real);
int tnodecontrol;
int vnodecontrol;
#endif
#ifdef MP
extern real usertemp1Global(int);
extern real userfs1Global(int);
extern real uservx1Global(int);
extern real uservy1Global(int);
extern real uservz1Global(int);
extern int nodNumGlobal (real,real,real,int,real* ,real* , real* );
#else
/* These functions return numbers and not indices */
extern int userget_nbnodes_casting (); /* number of
nodes in the CASTING domains */
extern int userget_nbnodes_castmold (); /* number of
nodes in the CASTING + MOLD domains */
extern int userget_nbelems_castmold (); /* number of
elements in the CASTING + MOLD domains */
extern int user_nb_ithnode_inelem (int,int); /* Connectivity
of the elements */
/* 1st argument
: number of the node in the element */
/* 2nd argument
: element number */
extern int user_getelemtype(int); /* element type
(eg : 2 for tetrahedra, */
/* 3 for wedge
and 1 for hexahedra) */
extern int user_getmaterialnb(int); /* Domain number
*/
/* These functions take node numbers as arguments */
/* the index 1 corresponds to the beginning of the timestep */
/* the index 2 corresponds to the end of the timestep */
extern real usertemp1(int); /* Temperature */
extern real usertemp2(int);
extern real userfs1(int); /* Solid fraction */
extern real userfs2(int);
extern real uservx1(int); /* X-velocity component */
extern real uservy1(int); /* Y-velocity component */
extern real uservz1(int); /* Z-velocity component */
extern real uservx2(int);
extern real uservy2(int);
extern real uservz2(int);
extern real userx1(int); /* X-coordinate, including
deformations */
extern real usery1(int); /* Y-coordinate, including
deformations */
/*
* function called at the beginning, the end of the calculation
* and at the beginning of each timestep
* loop = 0 : function is called at the start of the computation
* loop = 1 : function is called at the start of the current timestep
* loop = 2 : function is called at the end of the current timestep
* loop = 3 : function is called at the end of the computation
*
*/
void func_externalcompute(
char prefix[], /* case name */
int loop, /* loop value : 0/1/2/3 */
int timestep, /* current timestep */
real time) /* current time */
{
if ( loop == 0 )
{
tnodecontrol = 0;
vnodecontrol = 0;
}
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
return;
}
In this example, the interface heat transfer coefficient is taken as a function of the
liquid pressure. To do so, the "extern" declaration of the liquid pressure should be
copied from the upper "externalcompute.c" function (see red line below). Then,
this value can be used in the user function (see the blue lines).
#include <stdio.h>
#include <stdlib.h>
#include "common_dll.h"
#ifdef WIN32
#define EXPORT _declspec(dllexport)
EXPORT real func_interhtransfer(char*, int, real, real, real, real, real,
real, int);
#else
real func_interhtransfer(char*, int, real, real, real, real, real, real, int);
#endif
/*
* interface heat transfer coefficient
*/
real func_interhtransfer(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* boundary ID number */
{
/* ------------- Do not change anything above this line ------------- *
* ------------- Program your function below this line ------------- */
real hcoeff;
real local_pressure;
int local_node_number;
local_node_number = 100 ;
local_pressure = userpnode1 (local_node_number) ;
return hcoeff;
}
BATCH PRE-PROCESSING
PreCAST is having a scripting language which allows to perform a complete case
set-up in batch. This method is very convenient to automatize repetitive set-up or
to apply the same set-up to different meshes.
Introduction
Most of the Pre-processing operations which are performed in PreCAST can be
defined in a script (PRS script). A script corresponding to a simple set-up is
shown hereafter :
*GENERAL
*END_GENERAL
*SEQUENTIAL
*MATERIALS_ASSIGN_MATERIAL
1 Al_AlSi7Mg03-A356
*MATERIALS_ASSIGN_TYPE
1 CASTING
*BC_ADD
Heat
*BC_ASSIGN
1 Air_cooling
*SURF_SELECT_ALL
1
*BC_STORE
1 CURRENT
*GRAVITY_VECTOR
0.000000e+000 -9.800000e+000 0.000000e+000
*INITIAL_CONDITIONS_CONSTANT
1 700.00
*RUN_PARAMS_PREDEFINED
Gravity_Thermal
*END_SEQUENTIAL
*FINAL
*END_FINAL
The principle of such script is that it follows the operations and the menu (or
button) names used in PreCAST.
All the available commands are described in the "Script language" section.
A script can be used either to perform a whole set-up, starting from a mesh file, or
to perform a partial set-up, starting from a d.dat file. It is also possible to modify a
set-up (as it is done when a d.dat file is loaded and a few items are modified - e.g.
to change only a boundary condition).
The PRS script can be created in a User interface (see the "PRS User interface"
section for more details).
The "PreCAST Script Editor (PRS)" menu is opening this PRS Editor with the
following display.
The principle is that the user is setting his case in the same way as it is done in the
"Standard PreCAST", and each time an operation is performed, the corresponding
line are added to the PRS Script in the bottom window.
For the Bourndary conditions, the operations are done in the same way as in the
"Standard PreCAST" interface. The corresponding lines in the PRS script are
shown hereafter.
The main difficulty in Batch pre-processing is the definition of the locations (sets
of elements or nodes) where Boundary conditions should be applied. The
selections like "Select All", "Deselect All", "Select Remaining" can be used very
easily. For more specific selctions, "Box" selections can be defined. To do so, the
"Parametric selection" menu (on the top right) should be used. Then the user can
select the "Box" or "Rotated_box" mode and the box can be defined interactively
and labeled with a user defined name (to be used to the selection will be called).
The different operations to specify a "Box" selection are shown in the figure
above.
• Select "Box"
• Specify a name for the selection
• Select nodes interactively on the screen (either single nodes with the left
mouse button or inside a Rubber band box with the right mouse button). You
are allowed to rotate and zoom the part to select/deselect any node.
• When the "OK" button is pressed, the dimensions (i.e. the coordinates) of the
correponding box will be written in the script.
The dimensions of the "Box" are taken as the min and max values of the
coordinates of the selected nodes along the X, Y and Z axis. This means that the
box coordinates are not depending upon the orientation of the model when the
nodes are selected.
• Select "Rotated_Box"
• Specify a name for the selection
• Select nodes interactively on the screen with at least one Rubber band box
(with the right mouse button).
• When the "OK" button is pressed, the dimensions (i.e. the coordinates) of the
correponding box will be written in the script.
The dimensions of the "Rotated_box" are defined as follow : the box will have
one face parallel to the plane of the screen (called X'-Y') at the time of the capture
of the last Rubber Band box selection. The size of the box within this plane
(X'-Y') will be the size of the last Rubber Band box selection which was made.
The "depth" of the box (i.e. Zmin' and Zmax' in the direction Z', perpendicular to
the plane (X'-Y')) is given by the selected nodes. If this depth is not appropriate,
one can select or deselect nodes before pressing the "OK" button. In order to
better visualize the extent of the box (i.e. the selected nodes), it is advised to set
the "non-hidden" mode of the display.
The following figure is showing the operations to be done in order to "Add" a new
BC (1), to select a pre-defined "Box" (2), to Store the selection (3) and to Assign
the desired BC entry of the database (4).
and specific Run Parameters can be set/modified. To do so, one should check the
corresponding check-box (1) and set the desired value (2). Then, before changing
tab, the "Store" button should be pressed (3) in order to write the corresponding
lines in the script.
Beside the interactive creation of the PRS script described above, it is also
possible to create script lines using the *GENERAL, *SEQUENTIAL or *FINAL
menus in the Script Editor Window. Then, the different keywords are available.
save
save the PRS script at any time during the pre-processing process (the script is
saved anyhow at the end of the session). Please note that the "PRS PreCAST
Script editor interface" is not creating or modifying the d.dat file. It is only
creating a PRS file.
update
when lines are written manually in the script editor, the "update" button allows to
store them in the Editor (not saved yet in the file)
indent
The indent button allows to indent the different commands for a better viewing (it
has no effect on the running of the script)
quit
Closes the script editor window. The window can be opened again with the "prs"
icon on the right of the icon bar.
Script language
Most of the pre-processing operations can be controlled by a script ("prefix.prs"
ASCII file). This section summarizes the commands which are available. The
general philosophy is to have command names that resemble the menus of
PreCAST.
The different commands are specified using keyword lines. A keyword always
start with a *. Spaces are allowed on a line before the keyword and a $ character
corresponds to commenting out of the keyword. Keyword lines are often followed
by data lines, which contain the parameters necessary to process the above
keyword. Data lines cannot be commented out but will be ignored if the
corresponding keyword line is commented out. Blank lines are also allowed in the
script file.
Warnings
Forgetting the star before a keyword name leads to ignored the corresponding
command. The script is case sensitive.
• GENERAL COMMANDS
• SEQUENTIAL COMMANDS
• FINAL COMMANDS
The PRS script can be created in a User interface (see the "PRS User interface"
section for more details).
Remark
This is a first version of the commands (ProCAST 2009.0). It is possible that some
commands may change in the near future in order to be more powerful and
efficient.
GENERAL COMMANDS
*GENERAL
This keyword should only be defined once per script file together with a
*END_GENERAL counterpart. It encloses the commands that are handled before
the actual preprocessing session starts.
*VERSION
version_number
This keyword defines version of the software used to create the PRS script (e.g.
2009.1)
*SELECTION_TOLERANCE
Tolerance
This keyword defines the tolerance for deciding whether or not a surface node is
part of a given geometrical selection.
*SELECTION_GEOMETRY
Type name parameters
This command defines a geometrical selection. The only available geometry at
this point is a box (type Box) where the parameters are the coordinates (in 3D) of
diagonal corners of the box (brick built on the coordinate axes). In this, 6
parameters are specified : x1, y1, z1 and x2, y2, z2. The “General_brick” selection
is also available and defined by 4 points (see separate schema). The
“Perpendicular_interface_brick” is computed from the normal to equivalent
interface between domain n1 and n2 (input n2 = -1 afro a free surface or a non
equivalent interface) as numbered when this selection is used for the first time.
The node closest to the first point is used as a reference. Finally, a length along
this axis must be entered. Two directions are input to describe the section of the
box. These directions are projected on the local tangent plane. The selection
“Rotated_box” is defined by giving 4 points : the origin and three ends for a set of
vectors on which a box can be built.. The “General_cylinder” selection takes 8
parameters : first are the three coordinates of the center point, then the three
components of the axis and finally the half-height and the radius of the cylinder.
*CHECK_OVERLAP
Param
This commands will activate the check of overlap in combo selections. Param can
take the following values : none (no output), console (warning messages on the
console), full (warnings, dump total overlap value to file, dump overlap.lck).
*CHECK_DEGENERATE_SELECTION
Nb
This command activates the degenerate selection (a selection that will select nodes
but no element faces, such as a thin row of nodes) checker for geometry
selections. If Nb is 0, a degenerate selection is emptied. For any other value, a
lock file is dumped.
*CHANGE_DOMAIN_NUMBER
Nb name nb_p
Sets the domain number to Nb for elements with a domain number nb_p that have
all their nodes inside the geometrical selection named "name". If nb_p is negative,
only the selection criterion will be respected (i.e. all domains are considered) .
When batch preprocessing starts from a mesh file, any domain number may be
assigned to an element (renumbering may occur in PreCast if the added domain
number is more than just greater than the maximum pre-existing domain number).
When preprocessing starts with a d.dat file, only a pre-existing domain number
may be assigned to elements (in fact, a non pre-existing number may be assigned
for a transient state but is forbidden for actually running a calculation).
*OPTIMIZE_MESH
flag
Depending upon whether flag is “on” or “off”, the mesh will be optimized upon
saving the d.dat file. If this keyword is omitted, no optimization takes place.
*SYMMETRY_ROTATION
values (6)
Nb
This keyword activates or modifies the symmetry rotation definition. The six
values are the coordinates of the two end points of the axis and are to be given
first. The last input is the number of sectors.
*SYMMETRY_MIRROR_1
values (9)
This keyword activates or modifies the first mirror symmetry. The 9 values are the
coordinates of the three points defining the mirror plane.
*SYMMETRY_MIRROR_2
values (9)
This keyword activates or modifies the second mirror symmetry. The 9 values are
the coordinates of the three points defining the mirror plane.
SEQUENTIAL COMMANDS
*SEQUENTIAL
This keyword should only be defined once per script file together with a
*END_SEQUENTIAL counterpart at the end of the block. The commands within
this block are processed in the order they appear.
*MATERIALS_ASSIGN_TYPE
Nb type
Nb is the number of the material
Type is a string describing the type : MOLD, CASTING, FILTER, FOAM,
INSULATION, EXOTHERMIC, RESERVOIR
*MATERIALS_ASSIGN_EMPTY
Nb type
Nb is the number of the material.
Type is a string describing whether the material is empty or not : FILLED,
EMPTY
*MATERIALS_ASSIGN_MATERIAL
Nb name
Nb is the number of the material in the mesh.
Name is a string containing the name of the material in the database (can be any
name in the materials db)
*MATDB_MODIFY
Name type
Values
This will modify the material db entry with name name. The only available type
of modifications at this stage is cst_density (modifies constant values of the
density) and temptbl_density (modifies the temperature table definition of the
density without affecting units,…). For modification of the exothermic properties,
the following types have been added : tbl_burntfrac (modifes the burnt fraction vs.
time table), cst_ignit (modifies ignition temperature) and cst_exotherm (modifies
*MATERIALS_ASSIGN_STRESS
Nb name
Nb is the number of the material in the mesh
Name is a string containing the name of the material in the stress database (can be
any name in the stress db).
*STRESSDB_MODIFY
Name type
Values
This will modify the stress db entry with name name. The only available type of
modifications at this stage is cst_yieldstr (modifies constant values of the yield
stress) and temptbl_yieldstr (modifies the temperature table definition of the yield
stress without affecting units,…). For Young’s modulus, cst_youngsmod and
*DELETE_MATERIAL
Nb
Deletes the material number Nb from the geometry in the d.dat file. This will not
work if enclosures are present. Warning : any prior modification of the p.dat file
by the script will be erased by this command.
*GRAVITY_VECTOR
gx gy gz
gx gy and gy are the components of the gravity vector and they should be
-2
expressed in m s . Rotation of the gravity vector cannot be defined from the
script.
*INTERFACE_ASSIGN_TYPE
Nb type
Nb is the number of the material pair
Type is a string describing the type : equivalent, coincident, non coincident.
*INTERFACE_ASSIGN_HEAT_TRANSFER
Nb type
Nb is the number of the material pair
Type is the name of an interface heat transfer condition in the db.
*INTERFDB_MODIFY
Name type
Values
This will modify the intf db entry with name name. The available types of
modification at this stage are die_combo (modifies constant values of the values
in a die combo condition : constant h, air coeff and temp, spay coeff and temp,
attached 1 or 0, mold open, mold close, spray start, spray end), cst_standard,
timetbl_standard and temptbl_standard.
The keyword die_combo becomes cst_diecombo and temptbl_diecombo was
added.
*INITIAL_CONDITIONS_CONSTANT
Nb temperature
Nb is the number of the material and temperature is the corresponding initial
temperature, which must be given in °C. The extract of boundary conditions
cannot be scripted using this command.
*BC_ADD
Type
Type is a string corresponding to the type of boundary condition to be added to
the list of BC : Symmetry, Periodic, Accordion, Temp, Heat, Vel, Pres, Inlet,
Wall, Vent, Inj, Tur, Rot, , Displacement, PtLd, SrfLd, CurrentDensity, Voltage,
MagPotential, Nucleation.
Note that if a BC is added and no selection or db entry is assigned to it, it will not
appear in the final d.dat file. The number of the newly created BC is equal to the
last BC list number +1.
*BC_DELETE
Nb
Deletes the bc with number Nb
*SURF_SELECT_ALL
Nb
Selects the entire surface after highlighting boundary condition number Nb.
*SURF_DESELECT_ALL
Nb
Deselects the entire surface after highlighting boundary condition number Nb.
*SURF_SELECT_GEOM
Nb name
Selects the surface nodes corresponding to the criterion defined in selection
geometry with name "name". This selection will be made active for bc of number
Nb in the list. If Nb is set equal to 0, the nodal selection operation is performed
only.
*SURF_RESELECT_GEOM
Nb name Nb_ext
Selects the surface nodes previously selected for bc number Nb and retains only
those inside selection named "name". The nodes that were previously selected but
are out of the selection called name are assigned to boundary condition number
Nb_ext.
*SURF_DESELECT_GEOM
Nb name
Selects the surface node corresponding to the criterion defined in selection
geometry with name name. This selection will be made active for bc of number
Nb in the list. If Nb is set equal to 0, the nodal selection operation is performed
only.
*SURF_SELECT_REMAINING
Nb
Selects the remaining surface after highlighting boundary condition number Nb.
*SELECTION_COMBO
Name
Selection keywords
This sores the result of combined selection operation and the combo selection is
named name. As many selection keywords as desired can be used and the block
should be terminated with and *END_SEL_COMBO keyword. These commands
may not be nested however. Selection combos can only be used for selection of
surface nodes.
*SURF_SELECT_COMB
Nb name
Highlits bc number Nb and selects the area defined by the selection combo named
name (CURRENT for the active combo).
*SURF_SELECT_COPY
Nb_i Nb_t
This command copies and stores the selection from bc number Nb_i to bc number
Nb_t
*BC_ASSIGN
Nb name
Nb is the number of the bc in the list and name is the name of the bc in the bc db.
If "last" is used instead of name, the last bc of the db will be assigned.
*BC_STORE
Nb selection_name
The command stores a surface selection into the bc that has number Nb in the bc
list. Selection name is either CURRENT for the active selection or the name of a
combo selection. Only CÙRRENT is available at this stage.
N.B: After saving, the list of BC will be rearranged so that the number of a given
bc may change number from one preprocessing session to another.
*BCDB_MODIFY
Name type
Values
This will modify the bc db entry with name name. Two available types of
modifications at this stage are cst_velocity (modifies constant values of the
velocity), cst_hcoeff(constant value of convective heat transfer coefficient) and
cst_temperature (modifies constant value of the temperature).
It is now possible to modify tables for heat transfer coefficients using the types
timetbl_hcoeff or temptbl_hcoeff. The number of points in the new table should
be specified as the first value. Only the values in the table and number of points
can be altered by this command. The other parameters (type of function, units,…)
must have been defined previously.
*BCDB_ADD
Name type
Values
This will create the bc db entry with name name ("default" as name will use an
automatic numbering scheme). The only available types of new bc at this stage are
cst_velocity (creates a constant velocity bc) and cst_temperature.
*BCDB_DELETE
Name
This will delete the bc db entry with name name. If "Name" does not exist, no
deletion is performed
*RUN_PARAMS_PREDEFINED
Name
This command will apply the predefined set of run parameters with name name.
*VIRTUAL_MOLD_CREATE
Nb
This command will create a virtual mold with size Nb x default size.
*PROCDB_MODIFY
Name type
Values
This will modify the proc db entry with name name. The available types of
modifications are : cst_axis (specify 6 values : 3 coordinates for the first point of
the axis then 3 for the second point; cst_revol (specify the number of revolutions
per unit time), cst_uvw : specify the uvw vector used in solid transport and
translation condition, tbl_uvw : specifiy (starting with the number of points) the
table given in solid transport or translation entry, tbl_rotation : specify the rotation
angle vs. time table, tbl_revolution : specify the revolution rate table. The other
parameters (type of function, units,…) must have been defined previously.
*PROC_ASSIGN
No name
No is the number of the volume in the list and name is the name of the proc db
entry. If “last” is used instead of name, the last bc of the db will be assigned.
FINAL COMMANDS
*FINAL
This keyword should only be defined once per script file together with a
*END_FINAL counterpart at the end of the block. The commands within this
block are processed after saving the p.dat and d.dat file obtained in the sequential
session.
*RUN_PARAM_STRING
String
String is a line of the p.dat file as it would be added by hand. If a line
corresponding to the chosen run parameter exists in the p.dat file, it will be
replaced by String. If the corresponding run parameter is not present in the p.dat
file, String will be added at the end.
SOLVER
• Run DataCAST
• Run ProCAST
DataCAST
DataCAST converts the input data stored in ASCII in the prefixd.dat file into
binary data ready for the solver. DataCAST also cleans any previously existing
result files and prepare the new ones. This means that when DataCAST is run, all
results will be erased.
DataCAST is called with the DataCAST button of the Software manager. When
DataCAST is finished, one should press Return to close the corresponding
command window (see the Module calls section of the Software Manager chapter
for more details).
ProCAST
ProCAST is running the solver. No message is displayed by ProCAST, except if
the solver is stopping for an unexpected reason.
ProCAST is called with the ProCAST button of the Software manager. When
ProCAST is finished, one should press Return to close the corresponding
command window (see the Module calls section of the Software Manager chapter
for more details).
Status
During the calculation, the status of the model can be monitored. To do so, use the
"Status" button of the Manager (see the Module calls section of the Software
Manager chapter for more details). The prefix of the case to be monitored should
be present in the "Case" field and when the Status button is pressed, a window
with the calculation status is displayed :
The status information is stored in the software installation directory (in the
dat/stat directory), under the name prefix.stat. This means that the status of any
calculation is available at any time, by any user who is using the same software
installation. As a consequence, if two calculations with the same prefix are run in
the same time, there will be a conflict and the status will be once the one of
calculation A and once of calculation B. Such a conflict however has no influence
at all on the calculations themselves.
Batch calculations
It is possible to run several calculations sequentially in batch mode. To do so, a
simple script should be written and launch in a Command window, as shown in
the example hereafter (for Windows) :
cd calculations_A
cd case_1
%ProCAST20091%\bin\procast case_1
cd ..
cd case_2
%ProCAST20091%\bin\procast case_2
cd ..
cd ..
cd calculations_B
cd case_3
%ProCAST20091%\bin\procast case_3
cd ..
cd ..
DataCAST -u option
It is possible to run a calculation until a given step (e.g. step 500) with given
settings and then to continue the calculation with different settings (e.g. if one
would like to test different heat transfer coefficients from step 500).
To do so, run the calculation as usual until step 500. Then, in order to prevent any
error, copy the results in a safe place (as if a wrong operation is performed with
DataCAST, all the results files will be erased).
Load the d.dat file in PreCAST and make the desired changes (e.g. change a heat
transfer coefficient). Change the Run parameter INILEV to 500 and set DT = 0
(so that the last stored timestep will be used as the first one of the restart) and exit
PreCAST. Then, in order to update the calculation binary files, DataCAST has to
be run. However, in order to prevent DataCAST to erase the existing result files, it
is necessary to activate the -u option.
To use the "-u" option of DataCAST, the corresponding check box should be
activated (see figure above). One could also launch it from a Command window
with the following syntax :
datacast prefix -u
Then, the solver ProCAST can be launched and the calculation will be continued
from the step 500, with the new settings. Please note that it is not necessary to run
DataCAST if only the Run Parameters are changed (i.e. only the content of the
p.dat file is changed).
TROUBLESHOOTING
During the execution of the solver, information are stored in the prefixp.out file.
In case of problem, it may be useful to check the convergence of the solver in this
file.
At each timestep, a number of information are printed in the p.out file. The most
important ones (hereafter in the case of a thermal only calculation) are the step
number (1), the timestep (2), the fraction which has already solidified (3) and the
iterations (4).
If a calculation seems not to run well, it can be useful to look at the evolution of
the timesteps, during the course of the calculation. If the timestep is significantly
reduced, it means that the convergence of the calculation is poor. In addition, the
number of iterations per timesteps (i.e. the number of line shown in (4) in the
figure below, "T" indicating the resolution of the Thermal model) is a good
indication of the quality of the convergence. The least number of iterations, the
better the convergence.
In the case of a mold filling calculation, the prefixp.out file is looking like the
following figure. One can check that the "free surface model" is well activated (1),
that the model is filling with the filling percentage (2). The iterations (3) for the
Thermal model ("T"), for the pressure solver ("P") and for the three components
of the velocity ("U", "V" and "W") are shown. When more than 5-10 iterations are
necessary, one can anticipate a convergence problem. Finally, the memory used
by the calculation (for the current timestep) is shown (4).
One of the Run Parameter is COURANT. This parameter is limiting the timestep
in order to prevent the free surface to advance more than N elements during one
timestep. When the mesh has a large number of elements and when the elements
are small, this limitation may be quite drastic for small COURANT values. When
this limit is reached, the following message appears in the prefixp.out file :
The Run Parameter LVSURF allows to switch off the filling calculation when the
casting fill fraction has reached the LVSURF value. A message is printed in the
prefixp.out file to acknowledge this switch.
The fluid flow model will still be solved until the next stored timestep and then it
will be switched off (see figure below). The calculation will then continue with a
thermal only calculation (and stress if activated).
When the calculation is not converging, the following message is printed in the
prefixp.out file and the calculation of the given timestep is restarted with a
reduced timestep. This can happen a few times during the calculation and one
should not worry about it (typically, if the inlet velocity is suddenly increased, the
timestep may be too large and it should be reduced through this mechanism).
However, if the calculation does not converge several consecutive times, it may
indicate a problem in the problem set-up or in the input data (non relevant
material properties for instance).
RESULTS VIEWING
INTRODUCTION
Once the calculation is performed, the results can be viewed with the ProCAST
post-processor, called ViewCAST. ViewCAST can be called from the Manager as
follows :
Depending upon the type of calculation which was performed, the following
entities (or variables) can be viewed (raw results) :
• Scalar fields :
• Temperature
• Fraction of solid
• Vector fields :
• Velocities
• Heat fluxes
This chapter is describing how these results can be viewed. In the Result analysis
chapter, the display of more entities, such as porosity, solidification time,
secondary arm spacings, etc... will be described. The export of the calculated
results, to be used in other softwares are described in the Results Exports chapter.
Once ViewCAST is started, if the prefix of the case is well defined, the case is
automatically loaded in the post-processor and the initial temperature field is
displayed :
The background of the screen can be switched from black to white in the
"Parameters/Reverse Video" menu :
1. Display zone
2. Scale zone
3. Information zone
4. Menus
5. Icons
6. Tape player
The post-processor is "driven" by the Menus, the icons and the Tape player :
Firstly, the case should be opened in the File menu (if it is not opened
automatically when the ViewCAST is launched from the Manager). Then, the
field (scalar) to be displayed (e.g. Temperature) is selected in the Contour menu.
In the case of vectors, the desired field is selected in the Vector menu - see the
Field selection section for more details.
The times (or steps) to be displayed are selected in the Steps menu, whereas all
the parameters to configure the display (e.g. symmetry, cut-off values, type of
display, ...) are defined in the Parameters menu - see the Display parameters
section for more details. Then, the Picture menu allows to select the type of
display (i.e. snapshot, slice, scan slices or cut-off view) - see the Display types
section for more details.
Once these selections are made, the display is activated with the Tape player,
either as individual pictures or frame, or as animations (see the Tape Player
section for more details).
The X-Y Plots menu allows to draw curves (time evolutions) of the different
fields (see the Curves section).
The Actions menu allows to calculate some parameters from the raw results (such
as SDAS, isochrons, Niyama, etc...) - see the Results analysis chapter for more
details.
Please note that the Actions and X-Y Plots menus are replacing the former
module PostCAST (version 400 and before), as well as the File/Export menu.
To File/Exit menu allows to quit the post-processing session. Upon Exit, the user
is asked whether he would like to keep the information about the current
orientation of the model as well as the definition of the material display (i.e.
which materials are active and in which mode) for the next launch of the
post-processor.
GIF/AVI capture
All the pictures which are displayed in ViewCAST can be captured as GIF/AVI
The file name can be specified and one can browse in the desired directory (by
default it is stored in the working directory and the file name corresponds to the
case prefix) - the suffix (.gif or .mpg) is optional. One has to select between a Still
picture or an Animated picture and to select between the GIF or the AVI format. It
is also possible to activate the Reverse Video (with the "Reverse Video" check
box) in order to have a white background.
Then the picture is stored when the "Apply" button is pressed. For an animation, it
will start to store the animation when the Forward button of the Tape player is
activated. The "Cancel" button allows to exit without any action.
One has the choice between "Append" (i.e. to append or add the picture to the
previous one(s)), to "Overwrite" (i.e. to replace the existing file by this one) or
"Cancel" to abort the capture operation.
The AVI files are compressed using a "Motion jpeg" compressor. It has been
observed that with old versions of Powerpoint, such "Motion jpeg" files were not
recognized, unless they are renamed as prefix.mpg (instead of prefix.avi).
When an animation is stored in the GIF format, it can be interrupted at any time
(and the GIF animation will be stored until the interruption). However, it is not
possible to interrupt an animation when an AVI file is created. If the animation is
interrupted before the end, the AVI file will not be created. Thus, one has to
specify the starting and ending steps before the creation of an AVI animation.
FIELD SELECTION
Contour
The Contour menu is divided according to the different models which can be
activated in the ProCAST solvers.
The different "Contour" fields which are available are shown hereafter
One should note that the "Solidification Time" corresponds to elapsed time
between the liquidus and the solidus temperatures.
The Shrinkage Porosity is only calculated in the Run Parameter POROS is set to a
non-zero value.
As the stresses (i.e. the stress tensor) are calculated at Gauss points in the solver, it
is necessary to compute in the post-processing the different stress entities that can
be viewed (e.g. Effective stress, Principal Stress, Sigma X, ...). As these
calculations can be rather long, it is possible to compute all the desired one in one
shot, with the "Prepare Stress Results" option. In this case, the following panel
appears :
The stress result files which are present and up-to-date are shown in Green. The
ones which are not present are shown in Red. If there are files which are present
but not up-to-date (i.e. these files are older than the latest computed stress results
by the solver), they appear in Orange. In this panel, it is possible to select all the
desired stress results to be computed and to press "Apply" to start the
computation. The results which are in green will not be recomputed (unless the
"Overwrite Existing Files" is checked).
Please note that the fields which are shown in the menus correspond only to the
available result files in the working directory. For instance, if results are computed
in the "Action" menu - such as Niyama (see the Result analysis chapter for more
details), the corresponding item will be added in the menu.
Additional Contours
Since version 2007.0, it is possible to have new customized contours in the
Additional sub-menu.
These contours corresponds to prefix.usf files. The Fluid Front Tracking results
(see the "Results analysis/Fluid Front Tracking" section for more details) are
stored in such file and are thus accessible in this "Additional" sub-menu.
Vector
Vector results, such as Heat Flux and Fluid velocity can be selected in the Vector
menu.
It is possible to view both vectors and contours, if desired. In this case, the vectors
will be displayed in white (respectively black).
If vectors only are shown, the color of the vector can be set in the Parameters
menu (see the Display Parameters section for more details) to the vector
magnitude, to the temperature or the pressure. In order to shown the vectors only,
the Contour/None option should be selected.
DISPLAY TYPES
The type of display can be selected either with the following icons
Slice - 2D cuts (shown within the 3D model). One or several slices can be
selected, either along the X, Y and/or Z planes or along any plane. This mode can
be accessed directly with the F6 key.
Scan - Scan of 2D slices along either the X, Y or Z axis. This mode can be
accessed directly with the F7 key.
CutOff - X-Ray view of the inside of a model. The model is made partially
transparent (according to a criterion such as above a given temperature) in order to
see features which are inside the model, such as pockets of liquid, surrounded by a
solid shell (which is made transparent). This mode can be accessed directly with
the F8 key.
The selected mode can be viewed by a red square around the corresponding icon.
Examples of Snapshots (during filling and after), of slice and CutOff (X-Ray) view
Each Display type, except Snapshot, has ad-hoc settings (available in the
Parameters menu), as described hereafter (Slice Data, Scan Data and Scan Data).
Slice definition
For Slice Data definition (see figure below) :
5. Press "Apply", the corresponding selection will appear in the list (6) and the
slice will be displayed.
It is possible to use the Tape Player at this stage to see this slice over the
timesteps, either as single pictures or as animations.
As shown in the figure below, several slices can be selected. Moreover, the slice
characteristics can be stored in a file (file named "prefix.clip" in the local
directory) with the "Store" button and can be retrieved in a later session with the
"Read" button. The "Delete" button allows to delete the selected slice. The Show
"Yes"/"No" toggle allows to activate or deactivate slices, without deleting them.
In the following case, two slices are activated, as shown in the frame above.
When one slice only is activated, one can draw this slice with or without the
Background (using the "Display Background" button.
With "Display Background : None", only the slice is shown (upper right). With
"Display Background -", the slice is shown together with the background (upper
left). With the "Display Background +", the foreground is shown, up to the slice
(lower left). When the model is rotated (for instance around a vertical axis in the
above example), the background becomes the foreground and vice-versa. This
option is not available with more than one slice.
In this case, the following window opens. One should define the slice plane by the
coordinates of three points. To select interactively the points on the mesh, the "Get
Co-ord" button can be used (to select the point which is highlighted in red).
Once the three points are selected, it is possible to move this plane along its
normal, with a given Offset.
It is also possible to rotate this plane with the "Rotate" button. In this case, the
following window allows to define the rotation axis, as well as the angle of
rotation.
Finally, the "Copy Plane" is used to copy a plane which was already defined in
order to modify its characteristics. This is especially useful when one would like
to define several parallel planes which are not in the X, Y or Z directions. One
will define one plane with the standard Anyplane definition procedure and then,
one will copy this plane and offset it to the appropriate value. This can be repeated
for the other parallel planes.
Scan definition
The definition of the Scan parameters is done in the Parameters/Scan Data
menu, which opens the following window :
The Scan direction is selected, as well as the number of planes (or slices) which
will be shown. The largest dimension of the model in the selected direction is
divided by the number of planes.
Then, the display and the scan through the planes is activated with the Tape
Player.
Beside this slice mode, it is possible at any time to slice interactively the model,
using the "Scan" slider (see figure below).
Cut-off definition
The Cut-off (or X-Ray) parameters are defined in the Parameters/CutOff Data
menu, which opens the following window :
The Cut-off parameters are defined by two values (e.g. two temperatures). Then
the user has to select among five different display possibilities :
• Above Red : only the zones which are above the upper value (in red) are
shown
• Below Blue : only the zones which are below the lower value (in blue) are
shown
• Inbetween Bounds : only the zones which are in between the upper and the
lower values are shown
• Outside Bounds : only the zones which are below the lower value and above
the upper value are shown
• Isosurfaces : the isosurfaces corresponding to the two values are shown
first time, automatically, the "cut-off" mode is activated, with the definition
"above 0.01" (which corresponds to a porosity value higher than 1%). Of course,
it is always possible to go back to the normal snapshot mode, by clicking on the
snapshot icon .
DISPLAY PARAMETERS
Firstly, the timesteps to be viewed should be defined in the "Steps" menu, which
opens the following window :
The user can choose between Steps and Time increments. In blue, the available
minimum and maximum values are shown. The time values are always indicated
in seconds. If the selected Step increment is smaller than the available stored one,
it will take the first available step for the display.
One can also set "ViewCAST" such that the update is made in a "continuous"
mode during the calculation (this is useful when ViewCAST is launched while the
ProCAST calculation is still running). To do so, the "Setup" button (of the above
screen) should be pressed and the Update interval can be specified (please note
that if it is intended to leave the ViewCAST session opened for a few hours, it is
advisable to set an Update interval not too small, as this will load the CPU) :
Reverse Video
By default, the background of the screen is black. It can be reversed to white with
the Reverse Video option.
Free Surface
To visualize a mold filling four options are available to view the free surface (i.e.
the surface between the liquid metal and the air) :
The option "Free Surface (Foreground)" is used to view the free surface even
when it is hidden by the liquid or the rest of the casting. One can see in the upper
figure hereafter (Back face) the free surface in the ingate. If we rotate the casting
(lower figures), the free surface is behind. If the Free Surface is ON, we do not
see it. If the Free Surface is set as "Foreground", one can see it although it is
behind. Of course, one can also see ONLY the free surface.
The "Foreground" option can be quite useful to view entrapped gas pockets
(which are fully surrounded by liquid).
Enclosure
In the case of a radiation problem with an enclosure (this is valid only for surface
enclosures as solid ones are considered as materials) the display mode of the
enclosure can be selected among six choices : Invisible, Wireframe, Hidden,
Solid, Translucent or Shade :
Feature Angle
When the geometry is displayed in Wireframe, one can tune the "density" of lines.
In the following example, a Feature Angle of 10 and 60 degrees has been used.
Vector settings
When vectors are selected, it is possible to color them according to the vector
magnitude or according to Temperature or Pressure. By default, the color is
White. When the "Parameters/Vector Settings" menu is called, the following
panels opens.
One can choose between an automatic vector length or a manual vector length. It
is also possible to define a Unique vector length (i.e. all the vectors will be
displayed with the same length, regardless of the vector magnitude (except if it is
zero).
When vectors are shown together with a Contour Plot (i.e. a scalar variable such
as Temperature), it is not possible to color the vectors.
The vectors can be plotted with or without the arrow at the tip of the vector.
Displacement Mag.
In the case of Stress calculations, the display of the deformations can be
artificially increased in order to better view them. The real displacements are
multiplied by the "Displacement Magnitude" value.
Titles
Titles (text), arrows and cirlces can be added to the graphics, using the
Parameters/Titles menu, which opens the following window :
Then, one can "Add ->" either a String (text), an Arrow or a Circle. When a String
is selected, the text has to be typed in the corresponding field and then the text
should be placed with the mouse. When an Arrow is selected, it should be located
with the mouse. The first click should be at the starting point of the line and the
second click should correspond to the end of the arrow. For circle, the two
opposite corners of the square enclosing the desired circle should be defined. To
delete an item, click on the desired one (in the left grey zone), which will
highlight it in red and then press the Delete button.
All the information of this panel can be Stored in an ASCII file and retrieved
(with the Read button) for later use.
Symmetry
For cases which have been calculated with symmetry, it is possible to
"reconstruct" the full geometry, using the symmetry menu :
The figure below illustrate a mirror symmetry. For more details about the
definition of the symmetries, please refer to the Radiation section of the
Pre-processing chapter.
Display Foam
In the case of a lost foam calculation, one can display or not the foam with this
toggle menu.
Display Pipe
In case of porosity, if piping or macroshrinkage occurs, the corresponding
volumes will be shown as empty. However, in the visualization of the "Shrinkage
porosity", it may be useful to view together the porosity and the pipe shrinkage.
To do so, the "Display Pipe" OFF mode will allow to see the empty regions as
"Shrinkage porosity" with a value of 1 (i.e. 100% void). This mode is only
applicable for the Contour "Shrinkage porosity" and will not affect the viewing of
the other fields, such as the Temperature, Fraction of solid, ...
Particle tracing
The particle tracing allows to follow streamlines during the filling. The particle
traces are calculated during the filling calculation (as it is needed to have access to
all the timesteps, which is not the case during the post-processing). To do so,
"virtual particles" are launched at each inlet node (defined by a velocity BC, an
To view the Particle traces, activate "Particle Tracking" in the "Parameters" menu.
The traces can be viewed either alone or together with any contour or vectors. One
should note that the traces are always displayed in the foreground of a contour
plot (no hidden view of the traces).
In order to allow the viewing of successive batches of particles, the "tail" of the
traces are erased, as the traces are progressing, leaving the place to new traces.
Depending upon the geometry and the mesh size, it is possible that one or two
particles are going out of the model. Please note that this does not harm the other
results.
Tilt
In case of a tilt pouring calculation, it is possible to disable the viewing of the
model rotation. By default, the display of the rotation is active.
Display Undeformed
Scale
The scale can be modified by a direct click on the scale itself (1). Then, the Scale
settings panel opens (2).
For the temperature scale, if there is a material where the solidification range is
defined (i.e. a casting material), there is also the possibility to select automatically
the Solidus-Liquidus interval (Tsol-Tliq).
In the same line, when the Temperature is displayed, the Solidus and Liquidus
temperatures are indicated on the color scale :
When a scale was defined other than "Automatically", the minimum and
maximum values are stored for each field (the storage is done upon exit of the
model if the "Store the last view and Exit" option is selected. When the model is
loaded the next time, these settings will be automatically retrieved. The values are
indicated in the "Stored" field (the values can not be modified interactively, use
the Min-Max fields above). The values are stored in a local file called prefix.scale
(this file can be copied in an other case if one wants to use the same values for this
new model).
Finally, the color scale can be change, by clicking on the color map itself. The
following panel opens and the colors can be changed. It is also possible to store
and load a different color scale.
One should note that the scale and the color map is stored in memory for each
field (e.g. temperature, fraction of solid, velocity magnitude, ...). Thus, if the scale
is changed for the display of temperature and then fraction of solid is shown,
when the temperature field is displayed again, the modified scale is kept.
TAPE PLAYER
To activate the display, the Tape Player must be used. This allows to display
either individual pictures or animations.
As for a Tape recorder, the Tape Player has the following functions.
The definition of the first and last times or time-steps, as well as the time or
time-step increment between two frames is made in the Steps menu (see the
Display parameters section for more details).
CURVES
Curves can be viewed in ViewCAST, using the X-Y Plots menu. This allows to
visualize time evolutions of the calculated fields, such as Temperature, pressure,
velocity, etc...
Please note that these capabilities are replacing the former PostCAST module.
Interval
When Interval is selected, the user has to specify the Nodal interval. This means
that curves will be selected every N nodes (1, 501, 1001, 1501, ...). Only the nodes
of the materials which are active (as Solid) will be selected.
In addition, the user can select the Units which will be used for the display of the
curves. When the Apply button is pressed, the curves are read in the files and
displayed.
Nodes
When "Nodes" is selected, the following node selection window appears :
In addition, the user can select the Units which will be used for the display of the
curves. When the Apply button is pressed, the curves are read in the files and
displayed.
The node numbers (as well as the defined scale) can be stored (Export) and
retrieved (Import) in a file. To do so, the "File" button should be pressed and the
following window is opening :
In the case of an Export, the file name should be specified and it will be stored in
the working directory. For Import, one can "Browse" the desired file.
234
The first line contains the scale. As the scale is optional, the line should start by a
"-1" and should be followed by xmin, xmax, ymin, ymax. Then, there is a "0" or a
"1" to specify whether the Y-scale is defined automatically of not. A value of "0"
means that the specified ymin and ymax are taken into account, whereas a value
of "1" means that the Y-scale is set automatically.
Then, the list of the nodes should be specified.
Finally, a file named "prefix.tt" is created when the curves are displayed, with the
x and y coordinates of the displayed curves. These data can be then either viewed
together with other results (see the External section below) or exported to
spreadsheets (e.g. Excel).
External
When curves are selected with the Nodes mode (see above), the displayed curves
are stored in the same time in a text file named "prefix.tt". The fist line of the file
(see below) contains the number of time-steps (132), the number of stored curved
(3) and the corresponding node numbers (567, 9587 and 23129). Then, each line
contains the time and the field values for each curve.
Any file which as the above format can be loaded and displayed in ViewCAST
with this External option :
The desired files should be selected with the Browse. Please note that if the format
should be as described above, the name of the file is free.
Once this selection is accepted with the Apply button, the selected curves will be
displayed in the same time as the ones selected with the Nodes option.
GEOMETRY MANIPULATION
Interactive zoom (zoom up when the mouse is move towards the bottom
of the screen)
Rotation
When the Rotate icons is activated, the following Rotate window is opened :
With the X, Y and Z sliders, the rotation can be prepared. Exact angles can be
manually entered in the fields on the right. The rotation will be applied only when
the "Rotate" button is pressed. If the "Rotate" button is pressed again, a new
rotation (with the same values) is applied again. The "Reset" button resets all the
sliders to zero. By default, rotation are with positive angles. If one clicks on the
"+" sign, it becomes negative (for negative rotation).
Finally, the X, Y and Z keys can be used to rotate the model by 10° around the x,
y and z axis respectively. When the <ctrl> key is pressed in the same time, the
rotation direction is reversed. When the <shift> key is pressed in the same time,
the rotation is by 30° instead of 10°.
Interactive Rotation
At any time, the model can be rotated interactively with the mouse. The left
mouse key should be pressed and the model is rotated when the mouse is moved.
The center of rotation corresponds to the center of the screen (green arrow).
If the "Shift" key is pressed during the rotation, the model will rotate around an
axis which is perpendicular to the screen and located in the center of the screen
(green cross). The rotation is activated upon the horizontal motion of the mouse
(while the left mouse key is pressed).
Zoom
Beside the zoom icon (which allows to increase or decrease the size of the
model), the F2 and F3 keys can be used to increase, respectively decrease the size
of the model (zoom in and zoom out).
Stored Views
It is possible to store 6 different views in the "Parameters/Store View" menu. The
views are containing the orientation of the model as well as the material selection.
When a small arrow is following the View-X, it means that a view is stored
already. To store a view in a View-X which is available, just click on the
Parameters/StoreView/View-X. To retrieve (or Restore) a given view, use the
"Restore" option (one can also use the Ctrl-... key where ... is the number of the
view (1 to 6)). With the "Replace" option, the current view will replace the one
which was previously stored.
The "Reset view" (or Ctrl+0) will automatically reset the orientation of the model
in the isometric view with the gravity pointing downwards.
These views are stored in a file named prefix.lv (only if the option "Store the last
View and Exit" is selected upon exit of ViewCAST). Please note that in View-1, it
is always the last view which is active on the screen upon exit which is stored
(which will replace any previously stored one).
Please note that this *.lv file is not compatible with versions before 2006.0.
Material selection
The Material selection icon opens the Material selection window, which
allows to define which material are display as well as the type of display.
The list of the all the available materials is displayed. On the right, the colored
columns allow to specify the display attribute of each material, according to the
following nomenclature :
For each material, one should click in the desired column and a "X" will be
displayed. To define all the materials with the same selection, one could click on
the label (e.g. click on "IN" will make all the materials invisible).
Finally, when the apply button is selected, the display will be updated with the
corresponding selections.
Explode materials
In the Materials selection window, there is the "Explode Materials" button, which
opens the following window.
For each material, it is possible to define an offset for its display. If some
materials have no interface between them (EQUIV), they will be considered in the
same group and they can not be exploded separately.
When the Apply button is pressed, the corresponding offsets are applied, leading
to a picture as follows :
One can enter a node or an element number. By pressing the "Node" button, all
the elements which are connected to the specified node are displayed :
If the "Element" button is pressed, the nodes which belong to this element are
listed :
Pick option
The icon allows to pick a location on the geometry and display the node
number, the node coordinates as well as the field value (e.g. Temperature, velocity
magnitude, ...).
One should first click on the icon (which becomes outlined in red) and then click
on any point on the surface of the model. The information are displayed in a
yellow frame.
In the case of Stress models (i.e. deformed models), the displayed coordinates will
depend upon the "Displacement Magnitude" which is set. In order to have the real
deformed coordinate, one should set a Displacement Magnitude to 1 (with a value
of 0, it will correspond to the initial location of the mesh at the beginning of the
calculation).
ViewCAST exit
Upon Exit of ViewCAST, a message is asking to the user whether he would like
to keep the Last View. If Yes is answered, a file named "prefix.lv" will be stored
in the working directory. This file is containing the model orientation, zoom and
position, as well as the material selections.
RESULTS ANALYSIS
CRITERION FUNCTIONS
ViewCAST allows to process the results which are calculated in the solver in
order to built criterion functions or metallurgical results (Remark : this
corresponds to the functionalities which were present in PostCAST in previous
releases).
R G L calculation
RGL is a generic menu which allows to calculate the solidification rate "R", the
cooling rate "L" and the gradients "G", as well as any combination of those.
The solidification rate corresponds to the velocity of a given isotherm (e.g. the
liquidus isotherm). Thus, the user has to specify at which temperature he would
like to calculate the solidification rate (see "R,G Temp" in the figure below). One
can notice that it is allowed to calculate the velocity of any isotherm for other
purposes than a solidification rate.
Firstly, the user has to choose between two methods to calculate the solidification
rate. In Method 1, when each node reaches the specified temperature, a point is
located along the temperature gradient some distance away and the time that it
takes for the isotherm to reach that point is determined. R is then calculated as that
distance divided by the difference in time.
In Method 2, R is calculated as the cooling rate divided by the temperature
gradient. Method 1 takes longer to compute, but it does not depend on the cooling
rate. The results obtained by Method 2 are affected by the temperature levels used
to calculate L.
Four options are available for the calculation of the thermal gradient. One can
either compute the "Total" gradient (i.e. all components of the gradient) or in
either the X, Y or Z directions. Please note that the gradient is calculated at each
node, when the "R,G Temp" is reached. This means that the gradient corresponds
to a different time at each node.
Then, the user has the possibility to combine the R, G and L variables together to
obtain a criterion function (called Mapping factor). To do so, the coefficients a, b,
c and d should be specified
Finally, the units (length and temperature) used in the RGL calculation, as well as
the output format (*m.unf, Patran or I-DEAS) should be defined.
When the *m.unf output format is used, the Mapping Factors (M), the cooling
rates (L), the temperature gradient (G) and the Isotherm velocity (R) can be
selected for visualization in the Contour Menu.
Feeding length
The Feeding length provides the capability to calculate the distance between the
solidus and some user defined critical temperature which represents some fraction
solid beyond which feeding is impaired. This distance is then compared with a
"critical feeding length," which is a simple linear function of the hydrostatic
pressure. If the feeding distance exceeds the critical length, then porosity would be
likely. The Feeding length was designed for the cases of directional solidification
(such as DS or SX casting). Thus, it is not guaranteed that the Feeding length will
work well in other cases.
Please note that as the hydrostatic pressure is used in the Feeding length
calculation, the gravity vector should be defined in PreCAST (but it is not
necessary to run fluid flow). If a thermal only calculation was run and the gravity
was not defined in PreCAST before the run of the calculation, it is still possible to
load the case again in PreCAST, to set the gravity vector, to save the case and to
exit. Then, "DataCAST -u" should be used (see the "Solver" chapter for more
details on DataCAST -u, as well as the special care which should be taken in the
use of this capability in order not to scratch the results).
Then, the units used in the calculation, as well as the output format have to be
selected. Once the Apply button is pressed, the Feeding length is calculated. It can
then be viewed in the Contour Menu.
Isochrons
An Isochron is corresponding to the elapsed time from the beginning of the
calculation until a specific temperature is reached. The Isochrons can be
calculated in the Actions / Isochron menu.
In order to select the desired temperatures, two possibilities are available : "Semi
Auto" and "Specify Temps".
In Semi Auto, the user can select 15 temperatures (to calculate 15 isochrons plots)
by specifying the starting temperature, as well as a "delta" increment between the
temperatures (then the units and the output format should also be specified).
With the "Specify Temps" mode, the user can select any temperature (up to 15
isochrons). Again, the output format and the units should be selected.
Once the Apply button is pressed, the Isochrons are calculated. They can then be
viewed in the Contour Menu.
Alpha case
"Alpha case" is formed as the result of reaction between liquid Ti metal and the
mold wall, leading to the formation of TiO2.
Alpha case corresponds to the thickness of the TiO2 layer at the surface of the Ti
alloy casting in contact with the ceramic shell mold, in investment casting. This
thickness depends upon the cooling rate, the casting geometry and the amount of
oxygen at the mold/casting interface.
The alpha case thickness (tα) is calculated with the following equation :
where :
The above parameters, as well as the output format, should be specified in the
following window. Please note that the Bulk Concentration (c∞) must be smaller
than the Limit Concentration (c*).
Once the Apply button is pressed, the Alpha case is calculated. It can then be
viewed in the Contour Menu.
The secondary dendrite arm spacing (also called SDAS or λ2 ) can be calculated
according to the following equations, where tf is the local solidification time and
M a constant which depends upon the alloy properties.
Tstart and Tend are the temperatures which are used to compute the local
solidification time. Usually Tstart corresponds to the liquidus temperature,
whereas Tend should be the temperature just above (e.g. 1 degree above) the first
eutectic transformation. The Exponent corresponds to the power 1/3 in the above
equation (it is thus recommended to set the Exponent to 0.33333) and M is the
coarsening constant according to the above equation (which is alloy dependant).
3
The units for M should be [microns /sec].
Once the Apply button is pressed, the SDAS is calculated. It can then be viewed
in the Contour Menu.
POROSITY
Porosity
When POROS > 0 is used, one can visualize the Contour called "Shrinkage
porosity". Values corresponding to a level below 0.01 shall be considered as
microporosity (to be viewed best with slices) and values above 0.01 are
considered as macroporosity (to be viewed best with cut-off mode).
See the "Porosity models" section for more details.
Niyama
See the Criterion Functions section for more details.
In the case of die casting, when a stress calculation is run, ProCAST calculates
automatically the fatigue life in the dies.
The model is based upon a "strain-driven" approach and a power law relationship,
which corresponds to low cycle fatigue.
During the calculation, the accumulated plastic strain (if any) is recorded at each
location of the die. Then, the life of the die is calculated as a power law of this
plastic strain. If this life is shorter than the previously calculated life at this given
location, this latter value is stored. If there is no plastic strain, then, the elastic
strain is recorded and the life of the die is calculated with an other power law
relationship. Again, if the computed life is shorter as the previously calculated (at
this same location), this latter value will be used.
The parameters of the model correspond to measured values for a typical die alloy
(H13 steel), published by the Society of Automotive Engineers. One should note
that the values are measured at room temperature and thus, the effect of
temperature is only taken into account through the temperature-dependant stress
properties (i.e. Youngs modulus, Yield stress, Hardening, ...).
Thus, one should consider that ProCAST calculates a "Fatigue life INDICATOR"
and that it does not correspond to absolute values of number of cycles.
No data are required for this model and it can be used as an indicator for different
steel grades generally used for molds. One should notice that it should be mainly
used to compare different designs with the same mold material (e.g. to compare
the effect of cooling channels or of cycling time). One should not compare the
influence of different steels on the "Fatigue life indicator".
The hot tearing indicator allows to model cracks which are forming during the
solidification (i.e. it corresponds to openings in between dendrites which are not
yet totally solidified and which are opening because of a tensile stress). Hot
tearing are formed only above the solidus temperature.
The hot tearing indicator is a "strain-driven" model based upon the total strain
which occurs during the solidification. The model is computing the elastic and the
plastic strain at a given node when the fraction of solid is between CRITFS
(usually 50%) and 99%. There is no parameter for this model (except CRITFS
which is a general stress Run Parameter).
As the amount of plastic strain will strongly depend upon the stress properties in
the mushy zone, the hot tearing indicator should be used to compare different
designs with the same alloy. One should not compare hot tearing indicator levels
for different alloys.
CRACKING INDICATOR
The cracking indicator model couples the stress calculation with the porosity
calculation. It is valid with any porosity model. The cracking indicator is
corresponding to a void fraction indicator (i.e. it calculates the amount of voids
created by cracks, which have nucleated and grown under the influence of
stresses).
Plastic strain (and plastic strain only) will allow cracks to nucleate and grow (thus
to create "crack void fraction"). The presence of porosity will increase the amount
of crack nucleation and growth. All the plastic strain is taken into account
(including the one forming in the mushy zone).
There is no inputs needed for that model. One should note that the constants
which are in the modified Gurson model, are not temperature dependant (as there
is no indication in the literature about how these values should change with
temperature). However, the temperature dependency is taken into account in one
way by the stress material properties and thus the plastic strain (e.g. there will be
less plastic strain at low temperature).
This model is an "Indicator" which can be used for any material. The value of the
scale of the indicator will strongly depend upon the material and the stress data.
Thus, one should not compare two different materials, but one should use it to
compare the same material in different situation (i.e. different cooling conditions,
different stress state, different porosity level).
This model corresponds to a "damage" model, which couples for the first time
stress calculation with defects in the casting (i.e porosity). This model is very new
and should be used with care as there is very little experience in this field at this
stage. However, it could be very interesting to see the effect of design decisions
on the level of damage (e.g. a reduced cooling will change the amount of porosity
and/or change the amount of stress and thus, it will change the amount of cracks
which may appear).
The cracking model is activated with the Run Parameter CRACK = 1 or 3. The
default value is 0. See the "Stress Run Parameters" for more details.
FRECKLES INDICATOR
To activate this model, the casting material properties should have been created in
PreCAST with the Thermodynamic databases (see the "Thermodynamic
Databases" section for more details), using the Ni database. When the properties
are calculated, a file named "prefixls.dat" is created. This file (which contains the
liquid solute concentrations) should be present in the working directory in order to
be able to calculate the Freckles indicator.
Moreover, the Run Parameter FRECK with a value of 1 should be added manually
in the p.dat file.
Freckles will occur only if the liquid density inside the mushy zone is decreasing
with decreasing temperatures. In order to predict whether an alloy will have
freckles, one can look at the liquid density in the prefix_rho.dat file (created by
PreCAST when the material properties are created with Computherm). The liquid
density is the last column of the prefix_rho.dat file. If the liquid density in the
mushy zone is not decreasing (for decreasing temperatures), it is useless to run a
Freckle calculation.
These values are calculated in the solver and are then transported by subsequent
fluid movement. As these values are obtained by the resolution of an additional
transport equation, it was observed that the additional CPU time can be significant
(up to 20%). Thus, a simplified option is also proposed (where the additional CPU
time is significantly reduced, but with of course less precise results). The
simplified option can be activated with JUNCTION values of 11, 12, 13 and 14
respectively.
To activate the calculation of the Fluid Front Tracking, the Run Parameter
"JUNCTION" should be added manually to the p.dat file, with a value of 1, 2, 3,
4, 11, 12, 13 or 14 respectively (see the "Run Parameters/Flow" section for more
details).
The JUNCTION results can be viewed in VisualCAST in the FLUID section and
in ViewCAST in the "Contour/Additional/Fluid Front Tracking" sub-menu (see
the "Results viewing/Field Selection" section for more details).
• At each point of the free surface, the displacement of the free surface during one
timestep, multiplied by the free surface area is calculated. This value is
cumulated with the value of the previous timestep. In addition, this value is
transported with the free surface and with the fluid flow.
• When two free surfaces are meeting, both values are added.
• When there is no more free surface at a given location, the value of the Front
tracking indicator can still be transported by the fluid movement (during the
filling - and after if FLOW = 1).
This indicator allows to identify the locations where oxides and impurities trapped
at the free surface are most likely to end-up as well as to study the flow junctions.
The following figures are showing the Front Tracking indicator "Free surface
cumulated volume" for the same geometry cast with 4 different gravity
orientations (and thus 4 different filling patterns). In each case, the gravity is
pointing downwards.
The top figure shows the values during the filling, whereas the bottom figure
shows the results near the end of the filling.
Remarks
• This Fluid Front Tracking is a qualitative INDICATOR only (not a quantitative
value).
• For the Lost Foam, this value corresponds somehow to the amount of foam
which was burnt in contact with this free surface. It is thus an indication of the
potential defects which occur during lost foam burning.
• The numerical diffusion is slightly "diluting" the values of the indicator behind
the free surface, instead of remaining a "sharp" value at the free surface.
• This algorithm can not be used for centrifugal casting (as a Relative rotating
coordinate system is used).
• At each point of the free surface, the free surface area multiplied by the time.
This value is cumulated with the value of the previous timestep. In addition, this
value is transported with the free surface and with the fluid flow.
• When two free surfaces are meeting, both values are added.
• When there is no more free surface at a given location, the value of the Front
tracking indicator can still be transported by the fluid movement (during the
filling - and after if FLOW = 1).
This indicator allows to identify the amount of oxides formed at the free surface
and where they are most likely to end-up as well as to study the flow junctions.
The following figures are showing the Front Tracking indicator "Free surface time
exposure" for the same geometry cast with 4 different gravity orientations (and
thus 4 different filling patterns). In each case, the gravity is pointing downwards.
The top figure shows the values during the filling, whereas the bottom figure
shows the results near the end of the filling.
Remarks
• This Fluid Front Tracking is a qualitative INDICATOR only (not a quantitative
value).
• For alloys which are oxidizing (like Al), it corresponds somehow to the time
exposure during which oxidation can occur.
• The numerical diffusion is slightly "diluting" the values of the indicator behind
the free surface, instead of remaining a "sharp" value at the free surface.
• This algorithm can not be used for centrifugal casting (as a Relative rotating
coordinate system is used).
The following figures are showing the "Material Age" for the same geometry cast
with 4 different gravity orientations (and thus 4 different filling patterns). In each
case, the gravity is pointing downwards.
The top figure shows the values during the filling, whereas the bottom figure
shows the results near the end of the filling.
The following figures are showing the "Material Age" for the same geometry cast
with 4 different gravity orientations (and thus 4 different filling patterns). In each
case, the gravity is pointing downwards.
The top figure shows the values during the filling, whereas the bottom figure
shows the results near the end of the filling.
RESULTS EXPORTS
INTRODUCTION
The results can be exported either in Patran or I-DEAS format, except the Stress
results which can be exported in an ASCII format.
GEOMETRY
Firstly, the format of the output file should be selected (1). Then, as Patran or
I-DEAS files do not contain the unit information, it is necessary to specify in
which units it is desired to export the geometry information (2).
In the case of stress calculations, as the geometry is deforming, one can choose to
export the deformed geometry at a given timestep (3). It is also possible to export
the geometry with the "reversed displacements" (4). This latter option is useful if
one would like to run a stress model with a "negative" distortion in order to see
whether the final shape will correspond to one of the original drawing. In this
case, one should export the model in the d.dat format.
RADIATION FACES
For a radiation model, the radiation faces can be exported in the Patran or I-DEAS
formats. One should also specify the units which will be used in the exported file.
TEMPERATURE
For the export of Temperatures, three options are available for the selection of the
time-steps to be exported, as shown in the menu below.
Interval
With the Interval option, one can export temperatures according to the steps
which are defined in the Steps menu (only the Steps selection will be used and not
the Time selection) :
Then the format of the exported file (Patran or I-DEAS) and the units should be
selected.
Specify Steps
Any step to be exported can be selected with the "Specify Steps" option. The
selected time-steps should be entered in the following list :
The above window allows also to select the format of the export (Patran or
I-DEAS), as well as the units. The data are exported when the Apply button is
pressed.
Select Steps
The "Select Steps" option allows to choose the desired steps in the list of the
available timesteps :
The selected steps are highlighted in red. The above window allows also to select
the format of the export (Patran or I-DEAS), as well as the units. The data are
exported when the Apply button is pressed. The clear button allows to clear all the
selections.
Once the data are exported, a log file with the exported steps is created. This file is
named prefixt.log :
HEAT FLUX
The heat flux can be exported in exactly the same way as the temperature - see the
"Results Exports / Temperature" section for more details.
DISPLACEMENTS
The displacements can be exported in exactly the same way as the temperature -
see the "Results Exports / Temperature" section for more details.
In order to Export the geometry with "reversed" displacements, please refer to the
"Results Exports / Geometry" section for more details.
STRESS
1. Simple Export
The Advanced export allows to export stress results directly in Sysweld or Abaqus
format. A generic ascii format allows also to export to any other Mechanical
software.
Simple Export
The stress results can be exported in exactly the same way as the temperature - see
the "Results Exports / Temperature" section for more details.
The only difference is that the stress results are exported in a "Neutral" ASCII file
format in order to be imported in any stress software.
where x, y and z are the node coordinates and s1 to s6 are the six components of
the stress tensor at the node.
As ProCAST is calculating the stress data at the Gauss points, these values are
interpolated at the nodes.
Advanced Export
The Advanced Export is achieved in PreCAST (although the calculation is already
done), in two stages :
a) The stress results should be first "Extracted" (see the "Results extraction"
section for more details).
Stress Extract
The stress results should be "Extracted" first (in the "initial conditions/Extract"
menu). It is mandatory to extract ALL the domains. If you would like to extract
only a limited number of domains, you should first "Delete" the undesired
domains (with the "Geometry/Delete-Add Materials" menu) and then proceed
with the Extract.
If only one stress calculation was performed (i.e. no chaining with Extracts), one
should make an extract of the own results of the case. This means that in the
"Browse", one should select the case itself (which is currently loaded in
PreCAST). It is of course also possible to extract results from a different case.
Stress Export
Once the Extract is performed, the Export can be done (in the "File/Export"
menu).
The available formats are "Systus/Sysweld" (*.ASC for the mesh and *.SSF for
the stress results), Abaqus (*.inp for the mesh and *.asf for the stress results) and
a Generic ASCII format (*.gmf for the mesh and *.gsf for the stress resutls).
You should select the desired format and press the Store button.
Generic Format
Abaqus Format
Sysweld Format
Abaqus Import
Once the Mesh and stress data have been exported from ProCAST, the two files
prefix.inp and prefix.asf are created. These two files can then be complemented
and modified in order to contain the full Abaqus set-up.
The exported mesh file (prefix.inp) has the following format. It can be used
directly in the Abaqus set-up.
The exported mechanical data file (prefx.asf ) has the following format (to be used
directly in the Abaqus set-up - see hereafter)
The Abaqus set-up file (prefix.inp) shown hereafter was "built" direclty from the
exported files from PreCAST. One should of course make sure that the material
properties and models used in Abaqus are well the same (or consistent) as the
ones used in ProCAST.
Sysweld Import
The PreCAST export is creating the Sysweld format files *.ASC for the mesh and
*.SSF for the stress results.
These files can be used in the Sysweld set-up with the following procedure :
• Prepare your Sysweld case, using the standard preprocessing methodology (use
the same - or consistent - mechanical and thermophysical properties as in
ProCAST).
• Run the project to generate one state card in transitory files.
• Import the thermo-mechanical state using the ad-hoc SIL script (see next
figures). This state is automatically stored in card N° 1 of the transitory files.
INTRODUCTION
The goal of Parallel processing is to accelerate the computing time of a
calculation, by distributing it on several processors.
In the SMP technique (often called Shared Memory Processing), the calculation is
split in different CPU's which are sharing the same memory (see figure below).
This technology is usually limited to a maximum of 32 CPU's.
When each computer (node) has more than one CPU, sharing the same memory,
one can combine the SMP and DMP principles to make a SMP/DMP combined
architecture. Such a configuration (see picture below) requires a very fast network
between the CPU's.
The Parallel version of ProCAST is based upon the DMP technology, using MPI.
Depending upon the machine/cluster which is used, one can use either a DMP or a
combined SMP/DMP architecture. One should note that it is also possible to use a
single computer with several CPU's sharing the same memory.
As the MPI technology is based upon the "Message Passing Interface", it means
that the CPU's have to communicate between them. This emphasizes the
importance of the network linking the different nodes. One can use either a
network with a 1 Gigabit/s Ethernet switch to link the nodes. It is possible to have
faster communications with for instance a Myrinet network (from Myricom). This
latter solution is more expensive, but the gain in performances can be quite
significant.
Then, each sub-domain is allocated to one processor, as shown in the figure below
(each color represents a sub-domain). The communication between the
sub-domains is done automatically, through the fast network.
The following pictures illustrate the partitioning of different cases. In the lower
geometry, the partitioning for 2 and 4 CPU's is shown on the left and right,
respectively.
If such a partitioning is very efficient for thermal, radiation and stress calculations
(as well as for flow calculations with an already filled cavity), it is not the case for
filling calculations. As illustrated in the figure below (where the filling starts from
the biscuit in the bottom of the picture), the first stage of the filling will occur only
in the red sub-domain (CPU 1), whereas the final stage of the filling will take
place in the purple sub-domain (CPU 2). This means that the CPU 2 will not do
anything during a while at the beginning of the calculation, whereas CPU 1 will
"sleep" near the end of the calculation. This will lead to a very poor balancing
between the processors and thus a very bad performance.
However, as one does not know how the cavity will be filled (this is why a
calculation is performed !), a "dynamic partitioning" algorithm has been
developed. At a given step, the flow front is at a given location (in blue in the
figure below). The software is creating a layer ahead of the the flow front (the
width of this layer corresponds to a number of elements, defined by the Run
Parameter NFFWID). Then, both the liquid region and this layer are partitioned in
sub-domains. Once the liquid is reaching the end of this layer (i.e. when the
distance between the flow front and the end of the layer is less than NFFSAF
elements), a new partitioning is performed. Thus, the frequency of the partitioning
is directly influenced by the value of NFFWID. The smaller the value, the larger
the numbers of partitioning. As the partitioning is taking some CPU time, one
should specify a not too small value for NFFWID).
One should note that in the case of a filling problem, in the presence of a mold,
two distinct partitioning are made. One for the thermal and radiation problem
(which is done only once at the beginning of the calculation) and one for the
casting domain only, which is dynamically re-partitioned.
HP-MPI
A Portable Implementation of MPI (Message Passing Interface). Depending upon
the platforms, MPI can be called with different names.
Parmetis
A Parallel graphic partitioner
PetSc
The Portable, Extensible Toolkit for Scientific computation
Thus, the Pre- and Post-Processing should be made in the Scalar version, as usual.
The two "Parallel" Run Parameters (NFFWID and NFFSAF) should be added
manually in the p.dat file. If they are not specified, the default values of NFFWID
= 10 and NFFSAF = 2 are used. Recommended values are NFFWID = 10 to 20
and NFFSAF = 2 to 5 (see the "How does ProCAST Parallel works" section for
more details about the meaning of these Run Parameters).
Options (1) and (2): can be used to define additional default options to the
mpirun command.
MPI_ROOT: indicates where MPI run times and libraries are located. The
different MPI used are delivered in the ProCAST installation and default locations
are provided.
Host file
The "host file" list allows to select which CPU have to be used for a given
calculation. If you have only one machine (with several CPU's), the "host file" list
can remain blank. If you have more than one machine, you should fill the "host
file" list (use the "Edit" button) with all the available machines followed by the
number of CPU's of each one.
Then, when you want to launch a DMP calculation on specific machines, you
should select them in the list (highlight the desired ones in blue - hostname1 and
hostname3 in the above example) and then launch the calculation on 4 CPU's (as
hostname1 has 2 CPU's and hostname3 has also 2 CPU's). The hostname selection
(as shown above) is valid for all the subsequent calculations which will be
launched (even in the Manager is closed in the mean time) until the host file list is
modified. If a DMP calculation is launched with less CPU's than the number of
CPU's selected in the "host file" list, then only the first CPU's in the list will be
used (i.e. it is not necessary to deselect the unused CPU's).
One can see in the above execution windows that a "Parallel" section is appearing.
Please note that the format of the window may vary from one platform to an other.
For the ProCAST solver, one still have in both cases the possibility to "Execute
DataCAST first".
On Linux, the execution in batch mode can also be used (in this case, the Batch
log file can be defined - if nothing is specified, a file named DMP_runlog.txt will
automatically be created
Then, for the Parallel solver, some parameters and options should be defined,
before the Run can be launched.
First, the number of processors should be defined. If the radio button "Number of
Processors" is selected.
With the current version, the "Options(1)" and "Options(2)" fields may remain
blank (no specific options are necessary by default).
The Parallel solver can also be launched from the "Run List" of the Manager (see
the "Software Manager/Run list" section for more details).
Linux (HP-MPI) :
cd .../working_directory/
$ProCAST20091/bin/procastDMP_run prefix CPU selection -mpiroot
$ProCAST20091/hpmpi
or
-np N selection of N processors
-hostlist "hostname list" - See remark at the end of this section
Example :
cd .../working_directory/
$ProCAST20091/bin/procastDMP_run Test -np 4 -hostfile
hostfile.dat -mpiroot $ProCAST20091/hpmpi
cd .../working_directory/
%ProCAST20091%\HPMPI\bin\mpirun CPU_selection
%ProCAST20091%\bin\procastDMP.exe prefix
Remark 1 :
If DMP calculations have to be launched on CPU's which are physically on
different machines, it is highly recommended to use Linux, rather than Windows.
As the communications between the machines are managed by the Operating
System, it has been observed that in some cases, the communication between
Windows machines was not optimum, thus slowing down the overall calculation,
leading to a poor scalability. This was not observed on Linux, where good
scalability on multi-machines was always obtained.
Remark 2 :
Depending upon the machine configuration, it may be necessary to add the
following option (after the CPU_selection) :
-e LD_LIBRARY_PATH="%MPI_ROOT%\bin"
Example :
cd .../working_directory/
%ProCAST20091%\HPMPI\bin\mpirun -np 4
%ProCAST20091%\bin\procastDMP.exe Test
HOSTFILE and HOSTLIST definition for HP-MPI (for both Linux and Windows)
On HP-MPI, there is two ways to specify which CPU's have to be used for a given
calculation : the "hostfile" and the "hostlist". In both cases, the hostfile or the
hostlist should be specified in the launch command (or in a script which will
launch the calculation).
In order to explain how it is working, let's assume that the cluster has 4 machines
with 2 dual cores each (for a total of 16 CPU's). The machines are named
machine1, machine2, machine3 and machine 4.
Hostfile :
For instance, if a calculation with 8 CPU's has to be run on the machine3 and
machine4, the hostfile.dat should be the following (located in the local working
directory or in any directory which will be specified in the command line) :
machine3 4
machine4 4
and the calculation should be launched in the following way (let's assume that the
calculation is located in the /Data/Test working directory) :
cd .../working_directory/
$ProCAST20091/bin/procastDMP_run prefix -np 8 -hostfile
.../working_directory/hostfile.dat -mpiroot
$ProCAST20091/hpmpi
Hostlist :
REPEATABILITY
The goal of a parallel solver is to obtain the same results as the scalar version and
regardless of the number of processors. This is called "Repeatability".
One should note that Parallel processing (using an implicit solver like in
ProCAST), involves the resolution at once of a linear system on distributed
processors, with the appropriate communications between the processors. As a
first consequence, the algorithms which are used in the parallel version (especially
for the pre-conditioner) can not be exactly the same as the ones used in the scalar
version. This may lead to small differences in convergence (and thus in timesteps)
and in round-off. Moreover, as iterative solvers are used, the solution is never
exact and it may slightly depend upon the way it is solved (which can be slightly
different on 1, 2, 4 or 8 processors).
Thus, one should expect to have very similar results between the scalar and the
parallel version and between runs made with different numbers of processors,
however, most of the time, one will not have exactly the same result. This is
especially true in filling calculations where very small differences may have an
effect on the filling pattern (as different meshes or different COURANT values
may have too).
LIMITATIONS
In ProCAST 2009.1, not all the modules and features of ProCAST have been
parallelized yet. This section is describing the limitations of the current version, as
well as the precautions in the use of the parallel solver. As the use of a
non-parallelized feature may degrade completely the performances of the solver,
these features are made non accessible and the solver will stop in such case. The
tests are mainly done through the values of the corresponding Run Parameters.
However please note that when a Parallel calculation is terminated, many MPI
print-out may appear. This means that the ProCAST error message may be
"embedded" into MPI messages.
The following features are not parallelized (the list may not be exhaustive) and
thus should not be used with the parallel solver :
THERMAL module :
• POROS = 4 and 8 (only POROS = 1 is available)
• Periodic boundary conditions
• MiLE algorithm (for non-steady continuous casting)
• Solid Transport (for steady state continuous casting)
FLOW module :
• RESERVOIR option
STRESS module :
• User functions for the stress properties as a function of microstructure
Remark:
Please note that the path of the working directory should not contain any blank
space (or special characters).
MACHINE CONFIGURATION
In order to have an efficient parallel processing, it is important to set-up the
machines in the right way. This section explains the basics principles of the
machines installation and configuration in order to run the ProCAST Parallel
solver. Please note that the following is not necessary for the Windows DMP
version.
Cluster of processors
Concerning the operating system, it is important to make such that the installation
of each node is the same in order to optimize the performances.
SSH
During the Parallel calculation, the software is communicating with the different
nodes (computers of the cluster) through "Secured shells" (ssh). Thus, the
machines should be configured in order to allow "ssh" communication between
the nodes. If a "paraphrase" (password) is set to activate the ssh, one should
perform the two following commands when a new Command Window is used :
ssh-agent $SHELL
ssh-add
then, the "paraphrase" should be given. If this operation is not done, the
paraphrase will be asked for each node when the ProCAST Parallel solver will be
launched. If ssh is configured without paraphrase, then, this is not necessary.
If the Parallel solver is launched from the Manager, it is necessary to perform the
above operation in the Command window before the Manager is launched (in this
same command window).
The disk on which the software and libraries are installed, as well as the disk(s) on
which the data will be located (it can be different disks) should be mounted in
such a way that they can be accessed in the same way (i.e. with the same path)
from any machine (node).
HARDWARE AND OS
The pre-requisites and requirements for the hardware and for the operating
systems (OS) are the following (at this stage only Linux and Windows are
available) :
Windows - HP-MPI
Windows XP and Windows CCS (Compute Cluster Server) only.
In order to get the level of the kernel or of glibc, the following commands should
be run on the machine :
• rpm -q kernel
• rpm -q glibc
In order to get the level of the kernel or of glibc, the following commands should
be run on the machine :
• rpm -q kernel
• rpm -q glibc
INTRODUCTION
The porosity module of ProCAST is based upon the model developed by Pequet,
Gremaud and Rappaz. For further details, please refer to :
It is developed jointly by Calcom ESI (ESI Group) and the Swiss Federal Institure
of Technology, Lausanne, Switzerland.
The model can be summarized as follows (abstract of the above mentioned paper)
:
b) The temperature and fraction of solid results are then processed in the
Advanced Porosity Module in order to calculate the macro- and micro-porosity, as
well as the pipe shrinkage.
d) The calculation of the nucleation and growth of the pores is made according to
the gas model. The APM that was initially designed only for binary Al-based
alloys in which a single gas – hydrogen - was responsible for porosity (see paper
above), has been extended to multi-gas systems. In this module the effect of
several alloying elements on gas solubilities, and of course on final porosity
fraction, is now taken into account. For details on the modeling of porosity
formation in a multi-gas and multi-component configuration, see the following
papers (can be found in the software installation, under
"dat/manuals/Pdf/APM_papers.pdf") and also in the "APM Examples" section
(see in the "Alloying elements effects" and "Several gases" paragraphs) :
• Couturier G., Rappaz M. (2006). "Modeling of porosity formation
in multicomponent alloys in the presence of several dissolved gases
and volatile solute elements", TMS, Symposium on Simulation of
Aluminium Shape Casting Processing.
• Couturier G., Desbiolles J-L., Rappaz M. (2006). "A porosity
model for multi-gas systems in multicomponent alloys", TMS,
Modeling of Casting Welding and Advanced Solidification
Processes, pp. 619-626.
• Couturier G., Rappaz M. (2006). "Effect of volatile elements on
porosity formation in solidifying alloys", Modelling and Simulation
in Materials Science and Engineering, 14, pp. 253-271.
The flowchart of this porosity module is described in the following figure (for a
more detailed description, please refer to the "APM Appendix/APM Flow Chart"
section) :
Please note that there are no restriction to activate other models during the thermal
calculation, such as fluid flow or stress. However, it is important to remember that
the Advanced Porosity model does not handle remelting. If this occurs, it is not
guaranteed that the solution will converge.
The calculation condition file (prefix_poro.d (ASCII format)) contains all the
necessary information to perform a calculation. It is divided in blocks of data
containing different type of information :
Material properties
alloy.param:
specific.mass.param:
viscosity.param:
coarsening.param:
Process information
pressure.param:
feeding.param:
gravity.param:
Calculation settings
grid.param:
calculation.param:
Each block starts by the "block.name" and ends by "end". Inside a block,
parameters are inserted. Each parameter is always followed by one argument
(always one argument). The order of the blocks or of the parameters within a
block is free.
Please note that only SI units must be used for the porosity calculation definition
(however the thermal calculation can be set-up as usual with any units).
Many of the following parameters are optional: for them a default value is defined
in the APM source code. The user has the possibility to change this hard-coded
value: he can define it in the "prefix_poro.d" file, but if he wants to customize the
APM he can also define it in a "default_poro.d" file located at
"dat/pref/default_poro.d". The definition of a parameter in the "prefix_poro.d" file
takes over the definition in the "default_poro.d" file. The definition of a parameter
in the "default_poro.d" file takes over the hard-coded value. Below, for each
block description, a table mentions if each parameter is needed, optional or not
required (i.e. can not be defined at this location) in the "prefix_poro.d" and the
"default_poro.d" files. The hard-coded value is also mentioned.
Material Properties
When the "Advanced Porosity Model" is used in conjunction with a ProCAST
calculation, all the necessary thermo-physical properties (alloy composition,
fraction of solid, density and viscosity curves) are automatically extracted from
the ProCAST input data file ("prefixg.unf") if they have been defined in PreCAST
and stored in "prefixg.unf" by DataCAST. Thus, nothing has to be specified and
no special care has to be taken. But such parameters can be defined also in
"prefix_poro.d" file as described below. The definition in "prefixg.unf" takes over
the definition in "prefix_poro.d" file.
alloy.param: (optional)
The effects of several solute elements on gas solubility products can be taken into
account by the APM. This is valid if the gas.db database contains the "interaction
coefficients". See the Appendix 3 for the description of Gas Thermodynamic
Database.
For this, the APM reaches the concentration of each solute element in the liquid
phase: the solidification path is previously calculated by PreCAST. Three
segregation models can be used in PreCAST: lever rule, Scheil or back diffusion.
The resulting file, "prefix.phs", contains the concentration of each solute element
in the liquid phase as a function of solidification. The APM extracts directly these
concentrations in "prefix.phs". If this file does not exist, the solute element
concentrations are assumed constant: they are read in "prefixg.unf" file or in
"prefix_poro.d" if not found in "prefixg.unf":
The figure above shows, at which stage in PreCAST, in the material database, the
"prefix.phs" file is created. By selecting and applying the Scheil, Lever or Back
diffusion model this file is written out. The alloying elements which appears in the
list above are also used by the APM module: the APM detects the alloying
elements, and then reaches their respective concentration evolution in "prefix.phs"
file. If the "prefix.phs" file does not exit, the solute element concentrations are
assumed constant by the APM (values entered by the user in PreCAST: see figure
above). Please note that the APM module reads the "prefixg.unf" file and not the
"prefixd.dat" file. Therefore datacast has first to be used to generate the
"prefixg.unf" out of the "prefixd.dat" file to make this information available for
the APM module.
Important remark :
Please note that the file "prefix.phs" must be present in the working directory of
all APM calculations (whereas this file is created only during the material
properties calculation in PreCAST). This means that if the same alloy is used for
several calculations, one should not forget to copy this file in the local working
directory.
specific.mass.param: (optional)
Of course, in order to allow porosity calculations, one has to have an alloy density
which is defined as a function of temperature. This density curve is defined in
PreCAST. If this density curve has not been defined in PreCAST, the user can
define it in "prefix_poro.d" with the following parameters:
Note that this version of the "Advanced Porosity Model" does not allow to have
"expanding" materials (i.e. volume increasing for decreasing temperature).
viscosity.param: (optional)
coarsening.param: (optional)
Another important parameter for the calculation of the liquid pressure drop is the
permeability (K) of the mushy zone, which is described by the Kozeny-Carman
relationship:
Beside the fraction of solid (gs), the permeability is a function of the secondary
dendrite arm spacing (λ2), which can be calculated according to the following
coarsening law :
where tL is the time at which the liquidus temperature was reached and M is the
coarsening constant.
with :
The following frames are showing both possibilities available for the user:
nucleation.growth.param:
The nucleation and growth parameters of pores (i.e. bubbles) are defined in this
above block. All these parameters are optional.
When a pore will nucleate, it will appear with the radius defined under
"nucleation.radius" (optional). Please note that this initial pore radius value should
not be defined with smaller values than 5 microns. Otherwise, the pressure inside
the bubble will be so high that it will never nucleate and the calculation will not
run well.
Finally "transition.solid.fraction" is the critical solid fraction below which the pore
is spherical, and upper which the pore is constrained by the dendrite network.
gas.param:
The gases responsible for porosity in the considered alloy are detected by the
porosity module. The thermodynamic data of the gases responsible for porosity in
the alloy are read in the gas database "dat/db/gas.db" by the porosity solver. See
the Appendix 3 for the description of Gas Thermodynamic Database.
The user can change the location of the gas database with an environment
variable: with $ProCAST20091_DB the path becomes
"$ProCAST20091_DB/db/gas.db".
It is first recalled that the gases responsible for porosity in the considered alloy are
now detected by the porosity module.
The alloy is found in ProCAST run-files ("prefixg.unf" file). If not found, the
alloy category has to be defined in "prefix_poro.d" file (see alloy.param: section).
The user has to define in atomic fraction the nominal concentration of each
gaseous element in solution in the liquid phase. Let's assume a copper base alloy.
The gases responsible for porosity in this alloy are: H2, SO2 and H2O. The user
has to define the nominal concentration of H, S and O in solution in the liquid
alloy in the "prefix_poro.d" file:
If the user does not remember the gaseous elements responsible for porosity, he
can run the APM module in order to read these gaseous elements in the console.
For H2 in Al-base alloys, the APM also specifies the typical concentration levels:
- low: 2.3e-6 at, this value corresponds to 0.1 ccSTP/100g Al.
- medium: 4.6e-6 at, this value corresponds to 0.2 ccSTP /100g Al.
- high: 6.9e-6 at, this value corresponds to 0.3 ccSTP/100g Al.
1/ in PreCAST, the user mentions the alloy base (ex: Ni) in the composition panel
of the casting material:
with
gas.formation.coefficient = 2.24e+10
The above procedure has already been applied to Ni, Ti, Mg base alloys in
"dat/db/gas.db" ASCII file.
With this procedure in term of porosity, the alloy will have a behaviour similar to
Al-base alloys in which dihydrogen is the single gas responsible for porosity.
One should be careful that the initial gas element concentrations do not induce
formation of bubbles in the liquid metal before the pouring of the metal. In order
to prevent such case, a warning will be printed at the beginning of the calculation
(see figure below) if the following situation arises. Then it is advised to reduce the
initial gas element concentrations.
With APM, it is also possible to run a porosity calculation with no gas element
(concentrations equal to zero). In this case the uncoupled resolution algorithm will
be forced and microporosity will form as soon as the liquid pressure will reach
minus Laplace's overpressure (see "coupled.solving" parameter description).
Process Information
pressure.param
The type of process, as well as the pressure definition are specified in this block
(casting.type).
Firstly, one should specify the process ("casting.type") which is modelled, among
three choices : gravity, injection or continuous.
gravity :
For all types of gravity casting (sand, permanent mould, investment, ...), the
"gravity" type should be used. In this case, the software is automatically setting
the pressure to the highest surface of the casting. This means that the user does not
need to specify anything, except the value of the pressure ("p.imposed"). The
highest surface is determined with the help of the gravity vector.
With this mode, the metal at the top of the casting will be free to pipe, according
to the value of "mobility.limit" (see "feeding.param" block).
injection :
The type "injection" has to be used for hpdc, lpdc and squeeze casting, as well as
all processes where the liquid metal is injected. In this case, the injection point has
to be specified with the "x,y,z.pressure.coordinate" (see below) - it is advised to
select the last point to solidify as the feeding will be stopped when this point will
solidify.
As the metal is injected, no piping will occur, as there will be a continuous
feeding of metal (however, depending upon the geometry and the thermal
conditions, macroshrinkage may occur). In this case, the "mobility.limit" does not
need to be specified and it will be automatically set to 1.
continuous :
For all continuous casting processes, the "continuous" type should be used. As for
"injection", one will have a continuous feeding of metal and thus, the
"mobility.limit" does not need to be defined. In this case, it will automatically be
set to 1. Moreover, the feeding point (usually the last point to solidify) should be
specified with the "x,y,z.pressure.coordinate" (see below - please note that the
coordinates have to be set in [m]). In addition, in the case of a vertical continuous
casting, one should specify the liquid level (with "p.level"). In this case, the value
of the applied pressure will be automatically adapted to take into account the
appropriate metallostatic pressure.
The applied pressure is specified with "p.imposed". This pressure is either applied
on the highest surface (in the case of "gravity"), or at the
"x,y,z.pressure.coordinate" location in the case of "injection" or "continuous". The
value of p.imposed can be defined either by a constant or by a function of time
(with a value block F(TIME)).
In order to define the pressure as a time table, one should replace the pressure
value by a pointer (i.e. a name starting by a "*" - see below).
Then, an additional value block should be specified with the time function. The
name of the value block should be the same as the one of the pointer (without the
"*"). Then, the following syntax should be used :
If the level of the liquid is not corresponding to the level of the coordinate where
the pressure is applied, this can be accounted for with the keyword "p.level". This
has to be used in continuous casting (as the "effective" level of the liquid is
changing during the process) or if a part of the casting is not modeled. When both
"p.level" and "x,y,z.pressure.coordinate" are defined, the pressure applied at the
p.level altitude (in absolute axis coordinated. in the direction of the gravity)
corresponding to the applied pressure (defined by "p.imposed"). Below p.level,
the metallostatic contribution is added to p.imposed. This can be described with
the following piece of code (considering that the gravity is in the -Z direction) :
if (z > p.level) then
pcell = p.imposed
else
pcell = p.imposed + rho * g * (p.level - z)
endif
As the "injection points" can now be also defined "inside" a volume (and not only
on the surface), this allows to mimic the effect of a third phase to still apply the
pressure in front of a gate while it is totally frozen.
The two above examples (in the blue tables) are totally equivalent.
The following example is showing how these multiple injection points can be
used. In case A, the full gating system is modeled and one injection point at the
biscuit is used. In example B, the gates are not modeled and an injection point in
front of each gate is defined (please note that the injection point is inside the gate
and not inside the casting). In case C, the gating system is modeled (like in A), but
the injection points are set as in B (in front of each gate). One can see in the
following graphs that the results are identical in the three cases. The green arrows
are showing the injection points.
The main advantage of using the middle approach (case B) is that the CPU time of
the APM calculation is significantly reduced, as the volume of the gating system
and the biscuit (which can be sometimes rather large) is not taken into account in
the APM calculation. In the above example, the CPU time of case B was reduced
by a factor 8 with respect to case A (or case C) - the number of APM cell was
reduced in this case by a factor 7.
In the next example, the left injection point was removed in the right figure. One
can see directly the effect on the porosity results.
feeding.param
The "mobility.limit" corresponds to the critical solid fraction from which the
liquid at the top free surface can not move anymore locally (for piping). If the
"mobility.limit" is very low (0.01), it means that as soon as the solid fraction is
equal to 1%, the liquid can not move down anymore to feed the liquid pockets
below. Thus one will have almost no piping, but macroshrinkage in the center of
the casting. On the other hand, if a larger value of "mobility.limit" is used (0.9),
the liquid can move down up to a fraction of solid of 90%. This will induce a
significant piping and thus reduce the amount of macroshrinkage which may
occur in the center of the casting. The value of "mobility.limit" is dependent upon
the type of microstructure. In the case of columnar structure, the dendrites are
linked to the solid shell and thus can not move too much, leading to a small value
of the "mobility.limit". On the other hand, in the case of equiaxed grains, they can
move down more freely and thus, the "mobility.limit" is higher.
If one would like to prevent piping (such as in high pressure die casting, squeeze
casting or continuous casting), the "mobility.limit" has to be set to 1.0. In this
case, the liquid will continuously feed the casting.
when the solid fraction (gs) is becoming one, K is equal to zero. As in the Darcy's
equation 1/K is used, it is going to infinity. In order to prevent that, the
permeability is taken as a constant for fractions of solid above the value of
"permeability.solid.fraction.cutoff". Values around 0.95 to 0.98 are recommended.
If values higher than 0.98 are used, a value of 0.98 is automatically set. If nothing
is specified, a value of 0.98 is used.
In the presence of eutectic, the permeability drop at the dendrite root is limited
because the dendrite network does not coarse anymore as soon as gs>1-ge (ge is
the final eutectic fraction). The "permeability.solid.fraction.cutoff" parameter can
be used to model this effect.
At very high solid fractions, the interdendritic flow becomes very small and the
pressure drop calculation becomes very difficult (on a mathematical point of
view). In order to improve the convergence of the model, the feeding calculation
is stopped at a fraction of solid above "feeding.solid.fraction.cutoff". Values
around 0.95 to 0.98 are recommended. If values higher than 0.99 are used,
automatically a value of 0.99 is set. If nothing is specified, a value of 0.98 is used.
This value has to be decreased in the case of bad convergence of the model.
Cells with fs < "activation.fraction" minus 5% are seen as liquid by the APM
solver. The consequences are:
- calculation time drastically reduced (even for low values of
"activation.fraction"),
- microporosity more and more concentrated for increasing value of
"activation.fraction",
- microporosity replaced by macroporosity for increasing values of
"activation.fraction",
- for intermediate values of "activation.fraction" (~30%), APM macroporosity
looks like what is obtained by ProCAST standard porosity module (when
POROS=1 and PIPEFS=30%).
It is recommended to set a value lower than 30%, otherwise macroporosity is
overestimated. For consistency reasons, and for the conservation of the global
volume of porosity (micro-macro), viscosity and density curves are also truncated
in APM solver.
gravity.param
The user has to define the gravity vector in the "prefix_poro.d" file if not defined
in PreCAST (The three components of the gravity must be defined):
Calculation Settings
grid.param
The "grid.spacing" corresponds to the cell size of the FVM mesh. One should
select a size which is in relationship with the size of the mushy zone (in order to
have several cells through the mushy zone thickness).
In cases, such as hpdc, sometimes, the gates between the casting and the
overflows are very thin. This may lead to a lack of cells at the gate and thus, the
overflow may become "isolated" form the casting (i.e. no cells in between the
casting and the overflow). In such case, the porosity calculation will not be
possible, as there is no way to "transmit" the pressure from the casting to the
overflow. This could also happen if very thin sections, such as flashes are
modelled. In this case, these "orphan" cells should be removed in order to allow
the calculation to run. As this procedure of "orphan search" is CPU intensive, it
can be disabled in cases where it is clear that this will not occur (for bulky parts).
Otherwise, it is highly recommended to set the "grid.search.orphan" value to ON.
calculation.param:
The "region.number" (or domain number) in which the porosity calculation should
be done must be specified at this time. Please note that it is possible to run a
porosity calculation in more than one domain. In this case, the domain list should
be specified in a value block (see below). One should however make sure that all
the domains are connected.
In the "Advanced Porosity Post-processor" format, the porosity and the liquid
pressure are stored at each FVM cell. This allows a very fine resolution of the
results that can be displayed with the "Advanced Porosity Post-processor" (see the
"Advanced Porosity Post-processing" section for more details).
In the "ProCAST" format, the porosity results, calculated at the FVM cells are
"interpolated" to the FEM nodes. This allows to view the "Advanced Porosity"
results in ViewCAST. Nevertheless, it is recommended to visualize the results
with the "Advanced Porosity Post-Processor" because a loss of small
macroporosities and small microporosity regions is induced by this
"interpolation".
The "store.type" allows to select which result files format should be stored.
In order to define a list of regions (to perform the porosity calculation in more
than one region), the region.number should be specified by a pointer (i.e. a name
starting by a "*" - see below).
Then, an additional value block should be specified with the list of regions. The
name of the value block should be the same as the one of the pointer (without the
"*"). Then, the following syntax should be used :
coupled.solving
In copper- and iron-base alloys, as soon as the nucleation criterion is reached, the
pore growth rate is larger than the solidification shrinkage rate, whereas this is
never the case in aluminium-based alloys. As a consequence, whatever the initial
gas concentration is, liquid will always be rejected from the mushy zone in
copper- and iron-base alloys. This will induce a fast liquid pressure increase in the
mushy zone. A very small time step is thus required to capture this increase. For
cases with a great number of active cells, the calculation time will be too long.
Two solving modes are thus possible:
calculation.param:
coupled.solving 1
end
calculation.param:
coupled.solving 0
end
-> the alloy mass conservation equation is not coupled to the gas conservation
equations, and solidification shrinkage is assumed to be entirely compensated by
pore growth (no liquid flux) as soon as nucleation criterion is reached (when
liquid pressure will reach minus Laplace's overpressure).
coupled.solving parameter is optional. Its default value is equal to 1 for all alloys,
excepted for Fe and Cu-base alloys for which it is equal to 0.
crit.pore.fraction
Example of the type of porosity calculation that can be carried out for a complex shape
A383 casting. The geometry is shown in figure (a), while the calculated porosity
fraction is displayed on selected cuts when the applied pressure is 100 bars (b) and 1
bar (c) and (d). The effect of solute elements on hydrogen solubility has not been taken
into account in (b) and (c) and has been taken into account in (d). cH =
0.2ccSTP/100g, λ2 = 50µm. Here the cellular mesh was composed of 400'000 cells.
Warnings
• One should note that the "Advanced Porosity" module is based upon very
complex physical algorithms and thus its mathematical handling may be quite
delicate in some cases. It is thus possible that the solution is not converging (or
is giving error with the pre-conditioner). In such case, it is advised to either :
store more thermal timesteps, reduce the cell size and the finite element size.
• One should also take a special care to make sure that there is no remelting.
• Finally, when the mushy zone is quite small (in case of large castings with low
conductivity materials), the finite element size and the cell size should be
adapted (i.e. they should be much smaller than the mushy zone thickness).
Let's look now at the strictly needed parameters in "prefix_poro.d" for gravity
casting, injection and continuous casting.
Gravity
For gravity casting, in the "prefix_poro.d" file, the list of strictly needed
parameters is (the $$$ are indicating where a value should be specified by the
user) :
#--------- Gas Information ----------
gas.param:
H $$$ # typical value for Al: 2.3e-6 to 6.9e-6 [molar fraction]
end
pressure.param:
casting.type gravity # Gravity setting
p.imposed $$$ # [Pa.s] 101325.0 [Pa.s] = 1 bar
end
grid.param:
grid.spacing $$$ # [m]
end
calculation.param:
region.number *regions
store.frequency $$$
end
value regions
data PARAMETERS
$$$ # Porosity Domain 1
$$$ # ...
$$$ # Porosity domain N
end value
Injection
Simulated final porosity fraction (slices) in a complex shape High Pressure Die Casting
(100 bar).
For injection, in the "prefix_poro.d" file, the list of strictly needed parameters is :
#--------- Gas Information ----------
gas.param:
H $$$ # typical value for Al: 2.3e-6 to 6.9e-6 [molar fraction]
end
pressure.param:
casting.type injection # Injection setting
p.imposed $$$ # [Pa.s]. 101325.0 [Pa.s] = 1 bar
x.pressure.coordinate $$$ # [m]
y.pressure.coordinate $$$ # [m]
z.pressure.coordinate $$$ # [m]
end
grid.param:
calculation.param:
region.number *regions
store.frequency $$$
end
value regions
data PARAMETERS
$$$ # Porosity Domain 1
$$$ # ...
$$$ # Porosity domain N
end value
Continuous casting
For continuous casting, in the "prefix_poro.d" file, the list of strictly needed
parameters is :
#--------- Gas Information ----------
gas.param:
H $$$ # typical value for Al: 2.3e-6 to 6.9e-6 [molar fraction]
end
pressure.param:
casting.type continuous # Continuous setting
p.imposed $$$ # [Pa.s]. 101325.0 [Pa.s] = 1 bar
x.pressure.coordinate $$$ # [m]
y.pressure.coordinate $$$ # [m]
z.pressure.coordinate $$$ # [m]
p.level *plevel
end
value plevel
data F(TIME)
$$$ $$$ # Time [s] level [m]
$$$ $$$ # Time [s] level [m]
$$$ $$$ # Time [s] level [m]
end value
grid.param:
grid.spacing $$$ # [m]
end
calculation.param:
region.number *regions
store.frequency $$$
end
value regions
data PARAMETERS
$$$ # Porosity Domain 1
$$$ # ...
$$$ # Porosity domain N
end value
APM Examples
This section of the manual is illustrating the effect of different parameters (gas
concentration, secondary dendrite arm spacing, viscosity, pressure) on the
porosity results.
Casting geometry, and cooling conditions. Here the solid fraction is represented at the
same time for "good cooling" conditions (a) and for"bad cooling" conditions (b).
The geometry used in the examples below is given in the figure above. Two kinds
of cooling conditions are used: in the first case an adiabatic condition is applied at
the top of the casting and on the vertical walls connected to the top (a), while in
the second case the adiabatic condition (dashed line) is only applied on the top
side (b). Elsewhere the cooling is governed by a convective heat transfer
coefficient. Cooling conditions of figure (a) will be called "good cooling
conditions", and those of figure (b) will be called "bad cooling conditions". Such a
simple geometry could be employed by a user to calibrate the APM for his alloy
(see paragraph "Configuration of the porosity model" concerning the generic
"default_poro.d" file).
Gas concentration
The effect of the nominal gas concentration is illustrated in the figure below in the
case of good cooling conditions: as expected, the final porosity amount increases
with the nominal gas concentration.
ccSTP/100g, (b) 0.1 ccSTP/100g, (c) 0.2 ccSTP/100g, (d) 0.3 ccSTP/100g. Good
As shown in the figure below, this effect is still observed in the case of bad
cooling conditions. One has to pay attention to the fact that this effect can be
totally hidden if the cooling conditions are so bad that the pure shrinkage porosity
fraction becomes predominant over the gas-shrinkage porosity fraction.
ccSTP/100g, (b) 0.1 ccSTP/100g, (c) 0.2 ccSTP/100g, (d) 0.3 ccSTP/100g. Bad
To study the effect of λ2, it is assumed here that the value of this parameter is not
linked to the cooling rate (wrong in reality). As the liquid pressure drop in the
mushy zone is very high for very low λ2 (see Kozeny-Carman relation) the
porosity amount is expected to be higher for very low λ2 values. This effect is
slightly compensated by the curvature contribution since the maximum pore
radius varies also with λ2 (see appendix 2 in the "APM Appendix" section). The
effect of the secondary dendrite arm spacing (λ2) is illustrated in both following
figures in the case of good and bad cooling conditions. As observed, the expected
trend is found by the APM.
Final porosity fraction for a secondary dendrite arm spacing equal to ((a) λ2 = 10µm,
Final porosity fraction for a secondary dendrite arm spacing equal to (a) λ2 = 10µm,
It is observed in the figure below that the liquid pressure drop is negligible for a
high λ2 value ((d) 100µm). But as the available spacing is high, the porosity
growth is facilitated. That is the reason why the porosity fraction is globally
higher in case (d) than in case (c) (see figures above).
Ultimate liquid pressure for a secondary dendrite arm spacing equal to (a) λ2 = 10µm,
Liquid viscosity
The effect of the liquid viscosity is illustrated in both figures below. As expected
the APM shows that the final porosity fraction increases with the liquid viscosity
(see Darcy's law). This effect is accentuated when the cooling conditions are bad.
Final porosity fraction for a liquid viscosity equal to (a) µ = 1e-4Pa.s, (b) µ = 1e-3Pa.s,
Final porosity fraction for a liquid viscosity equal to (a) µ = 1e-4Pa.s, (b) µ = 1e-3Pa.s,
Imposed pressure
The effect of the imposed pressure is illustrated in both figures below. As
expected the APM shows that the final gas-shrinkage porosity fraction can be
significantly reduced if the imposed pressure is increased. Nevertheless the APM
will never predict the reduction or pure shrinkage porosity for an increasing
imposed pressure because actually not mechanical effect is taken into account in
the porosity model.
Final porosity fraction for an imposed pressure equal to (a) 0.1bar, (b) 1bar, (c) 5bars,
Final porosity fraction for an imposed pressure equal to (a) 0.1bar, (b) 1bar, (c) 5bars,
Final porosity fraction in an Al-7wt%Si-2wt%Mg part when (a) the effect of Si and Mg
on H solubility are not taken into account, (b) only the effect of Si is taken into
account, (c) only the effect of Mg is taken into account, (d) the effects of Si and Mg on
Several gases
In brasses, the formation of several gases and a volatile solute element are
responsible for porosity: H2O, H2, SO2 and Zn. Zinc is a volatile element in
melted copper base alloys, and thus can contribute to the formation of
microporosity. This Zn contribution is explained in the following reference (can
be consulted in "dat/manuals/Pdf/APM_papers.pdf"):
The porosity module is applied here to a brass alloy: Cu-10wt%Zn alloy. In the
results presented in the figure below, the alloy was assumed free of sulfur: only
the combined effect of hydrogen, oxygen and zinc in solution was taken into
account in the porosity calculation.
As observed in the figure below, the final gas-shrinkage porosity fraction strongly
depends on the nominal concentrations of H and O in solution in the liquid alloy.
In iron base alloys, H2, N2, CO gases are responsible for porosity. The porosity
module is applied here to a Fe-1wt%C alloy. The evolution of the carbon
concentration in the liquid phase is stored in the file containing the solidification
path: "prefix.phs" ASCII file. As a consequence the concentration of carbon,
which is one element of a gas responsible for porosity (CO), has not to be defined
in the "gas.param:" block of "prefix_poro.d" file: its concentration will be directly
read by the APM in the "prefix.phs" file. Only the concentration of N, H and O
has to be defined:
gas.param:
As observed in the figure below, the final gas-shrinkage porosity fraction strongly
depends on the nominal concentrations of N, H and O in solution in the liquid
alloy.
The "Run" button should be hit (in the "Advanced Porosity Solver" section) to
start the calculation. If the ProCAST Thermal calculation is not yet run, it is
possible to click on the "Execute DataCAST/ProCAST first" check box before the
"Run" button.
If the above window does not show the "Advanced Porosity Solver" section, one
can activate it in the "Installation settings" window (by clicking in the "Advanced
Porosity module" check box, under "Modules display") :
casting: volume.m3
all.domains 5.204001e-006
domain 1 2.601000e-006
domain 2 2.000000e-009
domain 3 2.601000e-006
end
macroporosity.pipe: affected.volume.m3
all.domains 3.710001e-007
domains 1 1.620000e-007
domains 2 0.000000e+000
domains 3 2.090000e-007
end
The first block (general:) contains the number of domains in which the APM
calculation is run (number.of.domains), the time (in seconds) for which the
solidification has been calculated in the Thermal ProCAST calculation
(casting.process.time.sec) and the CPU time for the APM calculation in seconds
(CPU.time.sec).
The second block (casting:) contains the total volume (all.domains), as well as the
3
volumes of each APM domain (in m ).
In the third block (modeling:), the cell size (which is the same for each domain),
as well as the number of cells in all the domains and in each domain are indicated.
The microporosity in all the domains and in each domain are given in the fourth
block (microporosity:). The "affected.volume.m3" column corresponds to the
3
volume in which the porosity is larger then zero (in m ). The
"porosity.fraction.in.affected.volume" corresponds to the average microporosity in
the "affected.volume.m3" (the value is in fraction). The next column (labeled by
the "xxx" above due to the very long keyword) corresponds to the
"volume.upper.crit.fraction.m3". This is the volume in which the microporosity is
above the value defined in the "prefix_poro.d" by the keyword "crit.pore.fraction"
3
(in m ). Finally, the last column (labeled by the "yyy" above due to the very long
keyword) corresponds to "porosity.fraction.in.volume.upper.crit.fraction". This is
the average of the microporosity in the "volume.upper.crit.fraction.m3" (the value
is in fraction).
Finally, the last block (macroporosity.pipe) is giving the volumes where there is
3
pipe shrinkage and macroporosity (in m ).
In the "2009.1" format, the porosity results, calculated at the FVM cells are
“interpolated“ to the FEM nodes. This allows to view the "Advanced Porosity"
results in ViewCAST. Nevertheless, it is recommended to visualize the results
with the “Advanced Porosity Post-Processor” because a loss of small
macroporosities and small microporosity regions is induced by this
“interpolation“.
The "store.type" allows to select which result files format should be stored (see
the "Calculation Settings" section for more details).
To access the "Advanced Porosity Post-processor", one has to make a right mouse
click on the "VisualCAST" button of the manager. This will open a sub-menu and
one should select "Advanced Porosity results" (see below) :
When a cut-off value of 1% is used, the following result is obtained (which is very
similar to the ViewCAST result above) :
One should note that the "Advanced Porosity Post-processor" results are more
precise than the "ProCAST" format results. This is due to the fact that the
Advanced porosity calculation is performed at cells (which are all shown in the
"Advanced Porosity Post-processor" format, whereas these cells results are
"interpolated" to the FEM nodes for the "ProCAST" format results. If there is only
a few cells which are empty (i.e. a small macroshrinkage), this may be "diluted" in
the surrounding microporosity during the extrapolation to the FEM nodes. Thus,
this small macroshrinkage may not be visible in the ViewCAST.
In "ProCAST" format, this distinction can not be made and both the microporosity
and the macroporosity is shown (as positive values).
APM APPENDIX
The Finite Volume porosity solver is a "post treatment" of the thermal results
obtained with the Finite Element ProCAST solver:
• The temperatures at nodes (“prefixt.unf” file) are read by the APM and
interpolated at the center of cells (1),
• The solid fraction at cells (2) is deduced from these temperatures and from the
gs(T) curve defined in the pre-processor of ProCAST (PreCAST) (read in
“prefixg.unf” file).
• The physical properties (ex: density curve and viscosity of the alloy) and some
casting process parameters (ex: gravity) are directly defined in PreCAST (3)
and read in the “prefixg.unf” file by the APM because these properties are also
required for the thermal-flow calculation.
• APM is a complete approach that takes into account the effect of solute
elements segregation on gas solubility if the solidification path has been
calculated by PreCAST (“prefix.phs” file). The alloy characteristics (alloy
base and composition) defined in PreCAST and the solidification path, are
directly read by the APM (4) in the “prefixg.unf” file and the “prefix.phs” file
respectively. That information allows the porosity solver to detect
automatically the gases responsible for porosity in the considered alloy and to
load the corresponding gas thermodynamic properties and solute elements
interaction coefficients (5). This data is loaded from the gas thermodynamic
database (“dat/db/gas.db” file).
The list of the strictly needed parameters in the “prefix_poro.d” file is limited (6).
A warning message is written by the APM each time a badly spelled keyword was
defined in the “prefix_poro.d” file. Other parameters are optional: for them a
value is defined in the APM source code. The user has the possibility to change
this hard-coded value: he can define it in the “prefix_poro.d” file, but if he wants
to customize the APM he can also define it in a “default_poro.d” file located at
“dat/pref/default_poro.d” (7). The definition of a parameter in the “prefix_poro.d”
file takes over the definition in the “default_poro.d” file. The definition of a
parameter in the “default_poro.d” file takes over the hard-coded value.
details see references [2-3]). As the interface energy, γgl, is on the order of
1 Jm-2 and the pore curvature radius, r, is equal to a few tens of micrometers, the
curvature contribution (Laplace's overpressure) cannot be neglected and strongly
influences the pore fraction, gp. While the relationship between gp and r is
straightforward for spherical pores, a simple model for the curvature of a pore
constrained to grow in a well developed dendritic network (i.e, gas-shrinkage
porosity) is derived in this section.
Regular stacking of dendrite arms showing the space available for pores (rmax):
hexagonal arms without impingement (left), and cylindrical arms with impingement
(right).
The left scheme of the above figure is an illustration of a too simple geometrical
solution that does not take into account secondary dendrite arms impingement.
The present model considers a simplified 3-dimensional network of cylindrical
secondary dendrite arms and takes into account their impingement (right scheme
in the above figure). Assuming that pores can grow in between the cylindrical
arms, assumed to be infinite in length, the maximum radius of the pore is simply
given by:
As the liquid phase is assumed to completely wet the solid, the contact angle at
the triple point (pore-solid-liquid) is zero. Equations above are shown in the
0.5
following figure, together with another relation, rmax = 0.5 λ2(1 - gs ), that
corresponds to the solution of the left geometry in the above figure. It appears that
the new relations above seem more adapted to the modelling of gas-shrinkage
0.5
porosity. Indeed, with rmax = 0.5 λ2(1 - gs ), rmax/λ2 does not exceed 0.025
for gs >0.9 . Taking λ2 = 40 µm, rmax will be smaller than 1 µm, and ∆pr will
-2
be greater than 1.8 MPa (for γgl = 0.9Jm ). Therefore, pores will have almost no
chance to grow if they do not nucleate before gs = 0.9. Doing the same
calculation with the above equation (rmax/λ2 < 0.15 for gs > 0.9), the curvature
contribution is on the order of 300 kPa during the last stage solidification, thus
allowing gas-shrinkage porosity formation.
The above equations, i.e. (rmax/λ2)(gs), have been hardcoded in the APM. The
user will also have the possibility to define his own (rmax/λ2)(gs) relationship in
the "prefix_poro.d" file or in the "default_poro.d" file. As the reading function of
"prefix_poro.d" can not read any table function of gs, (rmax/λ2)(gs) table is
assumed to be a function of the temperature (although the values of the abcissa are
indeed solid fractions) :
nucleation.growth.param:
available.half.spacing *r(gs)
end
value r(gs)
data f(temp)
0 0.7
0.2 0.6
0.8 0.5
1. 0
end value
First, the equations governing the formation of gases are given (for details see ref
[2]).
• formation of gas α,
• formation of gas λ,
with
Second, the gas database structure is given in the following table. It is organized
in three levels represented by three colours. The first level (red) specifies the alloy
category. In each category, this database has to distinguish properties of gases and
properties of chemical elements that compose these gases because an element can
be contained in several gases and a gas can be composed of several elements. As a
consequence the second level (blue) contains some separate blocks of information
that concern each gas and each element. For example in the category Cu, "gas:"
blocks called water, dihydrogen, sulphur dioxide are defined. The existence of
these gases implies the definition of a "gas.element:" block for each of the
following chemical elements that appear at least one time in these gases: H, O and
S. In each "gas:" (or "gas.element:") block, characteristics and thermodynamic
properties of the gas (or element) are specified (third level in green). The reading
functions of gas.db file allow the definition of these properties in any order.
Representation of the structure of the gas database stored in file dat/db/gas.db (dummy
values are shown in this table)
Below is the correspondence between the gas parameters of the above equations
and parameters represented in green in the above table.
• The composition of the gas ("composition:") and the gas formation coefficient
("gas.formation.coefficient:") are needed parameters. The last parameter
the liquid phase (kA in the following equation) is also a needed parameter, as
well as the activity coefficient of A ("activity.coefficient:") in the liquid phase
l
in absence of solute element ('f A in the above equation).
The definition of a parameter in the "prefix_poro.d" file takes over the definition
in the "default_poro.d" file. The definition of a parameter in the "default_poro.d"
file takes over the hard-coded value.
APM References
The following references can be found in the software installation
(dat/manuals/Pdf/APM_papers.pdf)
[1] Péquet Ch., Gremaud M., and Rappaz M. (2002), “Modeling of Microporosity,
Macroporosity and Pipe Shrinkage Formation during the Solidification of
Alloys using a Mushy-Zone Refinement Method”, Met. Mater. Trans., 33A,
2095.
[2] Couturier G., Desbiolles J-L., Rappaz M. (2006). "A porosity model for
multi-gas systems in multicomponent alloys", TMS, Modeling of Casting
Welding and Advanced Solidification Processes, pp. 619-626.
MICROSTRUCTURES
INTRODUCTION
The only parameters that the user may need to specify are the nucleation
parameters (see below). This is due to the fact that this is not an intrinsic property
and that it may depend upon the metal treatment. Moreover, it may be necessary
to define the growth kinetics of the eutectic phase (see below)
The model which is used is based upon the "Gaussian distribution" model
proposed in 1987 by Rappaz et al.(Acta Metall., 35, (1987), 1487 and 2929). This
model defines the relationship between the number of nuclei and the
undercooling. The distribution of the nuclei with undercooling has the form of a
Gaussian distribution and thus, the integral of this curve is an "S-shape" function
(see graphs below).
Thus, the nucleation behavior of the primary dendritic phase is fully defined by
the three above parameters.
Thus, the nucleation behavior of the eutectic phase is fully defined by the two
above parameters.
Thus, the above constant fully defines the growth characteristics of the eutectic
growth.
Default values
The following table is showing the default values which are used in the
microstructure module for the models described above and for the different alloys.
Of course, the nucleation data may change from one alloy to the other, due to
different metal treatment. These default values corresponds to values proposed in
the literature.
Set-up
The set-up of a microstructure calculation is very simple as the model is based
upon the chemical composition.
For the thermal properties, only the thermal conductivity and the density should
be defined in addition to the chemical composition. Then, the enthalpy needs to be
calculated in PreCAST, with the Thermodynamic database (based upon the
specified chemical composition), using the Lever model for Fe alloys and the
Scheil model for the other systems. Thus, at the end, the thermal conductivity, the
density, the enthalpy, the fraction of solid, the liquidus and solidus temperatures
are defined.
Finally, the Run Parameter MICRO should be set to 1 (for the users of previous
versions, one should not anymore use values different from 1, as the selection of
the micro model is now automatic). A value of 0 will disable the microstructure
calculation.
If the nucleation or growth parameters have to be changed from the default values
(see Table below), the corresponding Run parameters can be modified (see the
Microstructure Run Parameters section for more details).
Coupled/Uncoupled algorithm
Since version 2006.1, the micro module can be used in a "coupled" or
"uncoupled" mode (see the Microstructure Run Parameters section for more
details about how to activate these modes).
The difference between the two modes can be described by the following flow
chart.
In the "coupled" mode, the microstructure calculation is fully coupled with the
thermal calculation. The solid fraction is calculated inside the thermal calculation
loop by the micro solver. The source term (i.e. the latent heat release) for the
energy conservation calculation is given by the micro calculation. The micro
calculation is called at every iteration until convergence is obtained. This mode
will lead to very accurate results, but at the expense of rather small timesteps and
larger CPU time (2-5 times larger than with the "uncoupled" mode).
In the "uncoupled" mode, the microstructure calculation is not coupled with the
thermal calculation inside the thermal calculation loop. The micro calculation is
called only after the convergence of the thermal calculation and thus it is called
only once per timestep. The source term (i.e. the latent heat release) for the energy
conservation calculation is based upon the solid fraction change between the last
timestep (t-dt) and the step before the last one (t-2dt). As a consequence, the
contribution of the latent heat is "delayed" for one timestep. For this reason, the
timestep should not be too large in order to have relatively accurate results. This
algorithm is fast, however, the results are less accurate and the timestep should be
controlled.
The default mode is "uncoupled" (before version 2006.1, all the micro
calculations were performed in "uncoupled" mode).
Results
Depending upon the alloy composition, the type of microstructure will be
different. As a consequence, the type of results which are computed will also be
different. All the microstructure results can be visualized in the post-processing, in
the "Contour/Micro" menu.
The "Primary Dendrite Radius" corresponds to the primary grain size. The
"Secondary Dendrite Arm Spacing" (also called SDAS) is the distance between
the secondary dendrite arms of the primary phase.
The "Eutectic Grain Radius" corresponds to the radius of equiaxed eutectic grains
which are nucleating in between the dendrites of the primary phase. Finally, the
"Eutectic Inter-lamellar Spacing" is the characteristic distance of the eutectic
structure.
In the case of grey iron (lamellar eutectic), in addition to the quantities described
above, other quantities can be calculated :
If Mg is present in the chemical composition of the base alloy, please note that if
the amount is larger than 0.004%, the software will consider that it is
corresponding to a Nodular cast iron (SGI). If it is well a grey iron, please
decrease the amount of Mg in the chemical composition to less than 0.004%.
In the case of Nodular cast iron, nodules of graphite, surrounded by austenite are
formed, instead of eutectic grains. Thus, the "Nodule count" (which corresponds
to the density of graphite nodules), as well as the "Austenite Radius" and
"Graphite Radius" are calculated. The solid state transformations, as well as the
mechanical properties are calculated, as described above. In addition, the
"Elongation" is calculated from the microstructure results.
The Nodular cast iron (SGI) model is activated as soon as there is an amount of
Mg in the chemical composition larger than 0.004%.
The Nodular cast iron (SGI) model is activated as soon as there is an amount of
Mg in the chemical composition larger than 0.004%.
In the case of a steel with the composition of carbon equivalent less than 0.53 , the
following properties are obtained.
The fraction of peritectic is the solid fraction formed from the reaction of liquid
and the existing primary solid phase. The fraction of proeutectoid refers to the
fraction of proeutectoid ferrite or cementite formed from the austenite phase as a
function of time during the solid phase transformation. The carbon equivalent
value controls which type of the proeutectoid phase will form (the "carbon
equivalent" corresponds to %C + (%Si + %P)/3 ).
References
For cast iron, the mechanical properties calculations are based upon the
microstructure, according to two following papers :
Stefanescu et al, Proceedings of the 4th Decennial International Conference on
Solidification Processing, Sheffield, (July 1997), 609.
Goettsch et al, Metallurgical and Materials Transactions, 25A:5, (1994), 1063
EXAMPLES
To do so, a very simple geometry was used, as shown below. The casting is
cooled from the right with a chill, whereas the rest is in a sand mold.
This set-up produces a full range of cooling rates, as shown in the cooling curves
hereafter.
The following figure is showing the two alloys which were used, with the
corresponding chemical compositions. The only difference between the two
calculations is the amount of Carbon (from 3.2% to 3.5%).
The above figure is showing the different kind of results which are automatically
computed for the two alloys. One can see that for the Alloy B, there is no
"Primary phase", as the alloy is lying on the eutectic composition.
The following figures are showing the comparison of the different results. On the
left, the Alloy 1 is shown (3.2% C - Hypo-eutectic) and the Alloy 2 is shown on
the right (3.5% C - Eutectic).
Nodule counts
Austenite Radius
Ferrite fraction
Pearlite fraction
Brinell Hardness
Elongation
Tensile strength
Yield strength
One can see that the SDAS is only available for Alloy 1, as there is no dendrites
(i.e. no primary phase) in Alloy 2.
In order to run iron micro modeling, the following set-ups are needed in PreCAST :
a. Run Parameters/Micro:
i. MICRO=1
ii. Select Default Values (Gray Iron or Ductile Iron)
iii. Based on the inoculation level change the values accordingly
1. EUNUCL (For ductile iron, default value is 2000 )
2. EUPOWER (For ductile iron, default value is 2.5 )
b. Run Parameters/Thermal
i. MACROFS
ii. PIPEFS
iii. FEEDLEN
iv. MOLDRIG (0~1, change it according to the hardness of mold)
c. Run Parameters/Micro
i. GRAPHITE (0~1, default=1.0)
ii. FADING (0~1, default=1.0) (0 for large castings, 1 for small ones)
iii. MGTREAT (time between the Mg treatment and the beginning of the
calculation)
It should first be pointed out that the normal equilibrium diagram really represents
the metastable equilibrium between iron and iron carbide (cementite). Cementite
is metastable, and the true equilibrium should be between iron and graphite.
Graphite occurs extensively in cast irons (2-4 wt % C), but it is usually difficult to
obtain this equilibrium phase in steels (0.03-1.5 wt %C). Therefore, the metastable
equilibrium between iron and iron carbide should be considered for steel, because
it is relevant to the behavior of most steels in practice.
It is convenient to combine the effect of the silicon with that of the carbon into a
single factor which is called the carbon equivalent (CE):
CE=%C+%Si/3
Steel
When the weight percent sum of all elements other than Fe is more than 5% (such
as some alloy steel or stainless steel), only equiaxed dendrite model will be
activated.
When the melt alloy cools from an initial temperature higher than the liquidus
temperature down to a temperature slightly below it, the primary delta dendrite
phase begins to nucleate in the liquid until a recalescence occurs due to the heat
released from the growing nuclei. During the recalescence, the nucleation ceases
and the nuclei grow rapidly into dendritic grains and soon impinge on each other
at the end of recalescence. Then the growth of delta dendritic grains is replaced by
the coarsening of delta dendritic arms. When the temperature reaches to the
peritectic temperature, the peritectic transformation starts if there is still liquid
available. Otherwise, the austenite phase precipitates from the delta phase until all
of the delta phase is transformed into the austenite phase. Lastly, when the
temperature of the casting is cooled down to the alpha phase transformation
temperature, the alpha phase precipitates from the austenite phase to the eutectoid
growth of alpha and Fe3C phases leads to the formation of pearlite phase.
When the melt alloy cools from an initial temperature higher than the liquidus
temperature down to a temperature slightly below it, the primary austenite
dendrite phase begins to nucleate in the liquid. The austenite phase will grow until
the end of solidification. There is no peritectic reaction here. When the
temperature of the casting is cooled down to the alpha phase(C<0.8%) or
cementite phase(C>0.8%) transformation temperature, the alpha phase or
cementite phase precipitates from the austenite phase to the eutectoid temperature.
Below the eutectoid temperature, the coupled growth of alpha and Fe3C phases
leads to the formation of pearlite phase.
Steels with less than about 0.8% carbon are thus hypo-eutectoid alloys with ferrite
and pearlite as the prime constituents, the relative volume fractions being
determined by the lever rule which states that as the carbon content is increased,
the volume percentage of pearlite increases, until it is 100% at the eutectoid
composition. Above 0.8% C, cementite becomes the hyper-eutectoid phase, and a
similar variation in volume fraction of cementite and pearlite occurs on this side
of the eutectoid composition.
Cast Iron
The term cast iron, like the term steel, identifies a large family of ferrous alloys.
Cast irons are multicomponent ferrous alloys. They contain major (iron, carbon,
silicon), minor (<0.01%), and often alloying (>0.01%) elements.
Cast iron has higher carbon and silicon contents than steel. Because of the higher
carbon content, the structure of cast iron, as opposed to that of steel, exhibits a
rich carbon phase. Depending primarily on composition, cooling rate and melt
treatment, cast iron can solidify according to the thermodynamically metastable
When the metastable path is followed, the rich carbon phase in the eutectic is the
iron carbide; when the stable solidification path is followed, the rich carbon phase
is graphite. Referring only to the binary Fe-Fe3C or Fe-C system, cast iron can be
defined as an iron-carbon alloy with more than 2% C. Important notice is that
silicon and other alloying elements may considerably change the maximum
solubility of carbon in austenite. The formation of stable or metastable eutectic is
a function of many factors including the nucleation potential of the liquid,
chemical composition, and cooling rate. The first two factors determine the
graphitization potential of the iron. A high graphitization potential will result in
irons with graphite as the rich carbon phase, while a low graphitization potential
will result in irons with iron carbide. The metastable phase amount has both
direct and indirect effects on the properties of ductile iron castings. Increasing the
volume percent of hard, brittle carbide increases the yield strength, but reduces the
tensile strength and elongation, of ductile iron castings. Because there is no
graphite expansion for the metastable phase, the formation of carbide increases the
likelihood of internal casting porosity.
The two basic types of eutectics - the stable austenite-graphite or the metastable
Classification
Historically, the first classification of cast iron was based on its fracture. Two
types of iron were initially recognized:
(Fe3C eutectic)
• Gray iron: Exhibits a gray fracture surface because fracture occurs along the
graphite plates (flakes); it is the result of stable solidification (Graphite
eutectic).
With the advent of metallography, and as the body of knowledge pertinent to cast
iron increased, other classifications based on microstructural features became
possible:
• Graphite shape: Lamellar (flake) graphite (FG) as shown in Fig. 4.1 and 4.2,
spheroidal (nodular) graphite (SG) as shown in Fig. 5.1 and 5.2, compacted
(vermicular) graphite (CG), and temper graphite (TG); temper graphite results
from a solid-state reaction (malleabilization.)
• Matrix: Ferritic, pearlitic, austenitic, martensitic, bainitic (austempered).
Graphite is a hexagonal-close pack form of carbon that can grow in both the liquid
and solid forms of iron. In theory, in irons above the eutectic composition of
carbon, the graphite first nucleates in the liquid, and then continues to grow in the
solid. In irons below the eutectic composition, the carbon does not start to grow
until the iron reaches eutectic temperature. As seen in a micro, the larger nodules
are from growth initiated in the liquid, and the smaller nodules are from growth
that does not start until solidification temperatures are reached. The graphite
nodules continue to grow as the iron cools, so the amount of growth that occurs in
the liquid is smaller than what would be assumed by the micro.
The expansion from the graphite that grows in the liquid pushes liquid back into
the riser, and does not offset shrinkage. So in order to minimize shrinkage, it is
necessary to maximize the late formation of graphite without having to reduce the
actual amount of graphite. Understanding what happens in a non-steady state
solidification of Ductile Iron suggests a way that this can be done.
It can be seen from the diagram on the previous page, which the maximum
amount of carbon that can be formed in late graphite is determined by the eutectic
composition, and as long as the iron is at eutectic or above, the amount of late
graphite will be the same.
Eutectic Model
This model can be applicable to both regular and irregular eutectics. In the case of
regular eutectics, growth of both phases of the eutectic structure is non-faceted in
nature. For irregular eutectic, the growth process of one of the phases is faceted.
Growth of the faceting phase requires considerably higher entropy of fusion.
Examples of faceted growth are graphite growth in stable austenite/graphite
eutectic and Silicon in Al-Si eutectic. The metastable austenite/cementite eutectic
is an example of non-faceted/non-faceted type eutectic growth. Growth of both the
stable and metastable eutectic are addressed here. Growth of the stable eutectic
usually proceeds at a higher temperature. A higher cooling rate results in the
formation of a metastable eutectic. This model assumes bulk heterogeneous
nucleation at foreign sites which are already present within melt or intentionally
added to the melt by inoculation.
The nucleation of the eutectic grains is based upon the model proposed in 1966 by
Oldfield (ASM Transaction, 598, (1966), 945). The number of nuclei is a power
law (Oldfield proposed a quadratic law) of the undercooling. The growth of the
grains is controlled by thermal undercooling at the solid/liquid interface. Solutal
undercooling is neglected here since solute diffusion during eutectic solidification
Growth of Ferrite:
Even though ferrite can form either from the breakdown of pearlite or from the
direct decomposition of austenite, it is assumed here that ferrite results only from
the latter source. The following assumptions are made for modeling the growth of
ferrite:
Graphite precipitation
One more micro run parameter for iron alloys solidification simulation is added
called graphite precipitation (GRAPHITE Run Parameter -see the "Micro Run
Parameters" section for more details) which tells the degree of graphite
precipitation during solidification. It varies from 0 to 1. 1 means that the graphite
expansion potential is completely considered in the simulation so the casting will
have a relatively low tendency of shrinkage. 0 means that the graphite expansion
does not occur hence there is no compensation for the shrinkage of the liquid
during solidification by graphite expansion. During the micro calculation, the
computed expansion part of the density (as a function of the phases present) is
multiplied by GRAPHITE. If GRAPHITE = 0, there will be no expansion,
whereas with a value of 1, the full expansion contribution will be taken into
account in the density. This density is only used if a porosity calculation (POROS
= 1) is made during the microstructure calculation (see the "Cast Iron Porosity
model" section for more details). The value of GRAPHITE will not affect at all
the computed microstructure.
The default value is set to 1.0. This parameter is used to adjust the porosity
formation to the real foundry condition.
Nodular cast iron (SGI) is obtained by a Magnesium treatment of the melt. The
graphite formation tends to diminish as the time interval between the treatment
and the solidification increases (fading effect). As a consequence, the expansion
effect will be reduced as graphite formation is decreasing. This will of course have
an effect on the amount of porosity.
The run parameter FADING will allow to adjust the inoculants coarsening speed
as :
Normally the value of FADING can be any number between 0 and 1. The default
value is 1.
Small FADING value corresponds to slow coarsening, which is the case for heavy
section ductile iron castings. Thus, for large castings, it is advised to set FADING
to 0.
The value of FADING will affect microstructure results, thus the densities and
thus the porosity results as illustrated in the following figure.
MGTREAT :
As the calculation is starting usually at the beginning of the filling or when the
casting is full, one should take into account the time between the Mg treatment
and the beginning of the calculation. To do so, one can define the Run parameter
MGTREAT (see the "Micro Run Parameters" section for more details).
Case studies
A simple geometry casting is used to illustrate the application of the
microstructure module on the solidification of different alloys. The casting is
cooled from the right with a chill (higher heat transfer coefficient). The two bigger
surfaces (back and front) are symmetry. All the other faces are adiabatic.
1. Al 4.9wt%Si
For this alloy, as the temperature cools down, the primary phase forms first. The
possibility to have and the amount of eutectic phase depends on the cooling rate.
With faster cooling rate, there is less amount of primary phase but more eutectic
phase. Fig. 6 shows the comparison of current calculation with some experiment
and other modeling results for the solidification of this alloy.
4) Fraction of eutectic
1) Nodule counts
2) Austenite radius
3) Graphite radius
4) Pearlite Spacing
5) Fraction of Ferrite
6) Fraction of Pearlite
7) Fraction of Eutectic
9) Tensile Strength
11) Elongation
12) Hardness
We can see that it is not 100% stable phase (Eutectic) every where for this casting
after solidification. On the higher cooling area (right side), the metastable phase
(ledeburite) formed. The fraction of stable phase plus the fraction of metastable
phase is 1. The hard brittle metastable phase can form at higher cooling area
when the inoculation is not good enough. The metastable phase can increase the
yield strength but reduce the tensile strength of the ductile iron castings.
2) Fraction of Eutectic
6) Fraction of Ferrite
7) Fraction of Pearlite
8) Pearlite Spacing
9) Tensile Strength
4) Fraction of peritectic
5) Fraction of eutectoid
6) Fraction of proeutectoid
TTT/CCT MODELS
Introduction
The formation of microstructures associated with solid state phase transformations
during cooling or during heat treatment can now be simulated with ProCAST
using models based upon time-temperature-transformation (TTT) or Continuous
Cooling-transformation (CCT) diagrams. TTT/CCT diagrams can be found in
handbooks or scientific publications for a variety of alloys. They provide
information about the starting and ending times of a phase transformation for
isothermal (see illustration hereafter) or continuous cooling conditions.
where fmax is a maximum for the fraction transformed (e.g. equilibrium amount
of ferrite), n is an exponent and τ is a characteristic transformation time. These
parameters are determined from the information contained in the TTT/CCT
diagram. They depend on the local temperature and cooling rate.
A differential form of this expression is integrated over the cooling path in order
to account for the local thermal conditions. It permits to calculate microstructure
maps based on the thermal history at every node of the finite element mesh.
Pre-processing
Coded TTT/CCT diagrams are grouped in a database which contains an entry for
each of them. In short, the operations required to activate the TTT/CCT module
simply consist of assigning in PreCAST an entry of the TTT/CCT database to the
casting domains and setting to 1 the TTTMIC Run Parameter. In more details the
operations are as follows:
On the right of the window, two frames are shown. The top one shows the list of
the casting materials. Mold materials are not visible in this list as the TTT/CCT
microstructure module is not applicable to these materials. The lower frame
contains the list of all available TTT/CCT diagrams in the database.
To assign a TTT/CCT diagram, one should select the desired TTT/CCT diagram
in the database list and click on the Assign button. The selected diagram will be
assigned to all the casting materials as the model is currently limited to one
diagram only. After assignment, the corresponding entry number in the TTT/CCT
database is displayed in the right column of the upper frame.
Database
Currently the TTT/CCT database contains 8 steels which are coming from the
Sysweld database.
These alloys can also be found in the standard database (for the thermophysical
properties).
Post-processing
The results of TTT/CCT microstructure calculations can be visualized in the
VisualCAST post-processor.
Results
As an illustration the TTT/CCT microstructure module has been applied to the
heat treatment of a carbon steel gear component. The figure hereafter illustrates
the proportion of microstructure constituents (pearlite, bainite and martensite)
obtained after heat treatment for two types of cooling conditions: water quenching
and air cooling.
MICROSTRUCTURES - STRESS
COUPLING
Two different type of coupling between Microstructures and Stress are available
in ProCAST :
In some cases, the mechanical properties during the process can depend upon the
microstructure. This is especially true during heat treatments. Moreover,
mechanical properties can also be affected locally by defects, such a porosity.
As such dependencies may be very different from one alloy to an other one, the
definition of these mechanical properties are done with the help of User
Functions. When these User Functions are activated, at each timestep and each
node (or Gauss point), the mechanical properties are evaluated in the
corresponding User Function. This allows to well see the effect of the
microstructure on the mechanical behavior of the part, during the process.
The mechanical properties which can be defined in this way are the following :
Elasticity
• Youngs modulus
• Poisson ratio
• Thermal expansion
Plasticity
• Yield stress
• Hardening parameters
• Plastic modulus
• Ultimate stress
• Hardening exponent
• Kinematic hardening 1
• Kinematic hardening 2
Viscoplasticity
• Viscoplastic parameters
• Viscous param
• Power
• Strain power
• func_youngsmodulus()
• func_poissonratio()
• func_thermalexpansion()
• func_yieldstress()
• func_hardeningplasticmodulus()
• func_hardeningultimatestress()
• func_hardeningexponent()
• func_hardeningkinematic1()
• func_hardeningkinematic2()
• func_viscoplasticviscousparam()
• func_viscoplasticstresspower()
• func_viscoplasticstrainpower()
In order to activate the corresponding User Function, the "Table" radio button
should be selected. Then, the "User Function" check box should be selected. Even
the "Table" radio button is selected, it is possible to leave the table empty (if
everything is defined in the User function). It is however possible to fill the table
and the corresponding values can be retrieved in the User function itself.
The unit of the Stress property, used in the User Function, is the one specified in
the Table window (MPa in the exemple below) :
Please note however, that the Temperature unit used in the User Function is the
one specified in the default unit file, "UserFunctions_units.dat", located in the
"dat/pref" directory, or the one given in "prefix_units.dat" file, located in the
execution directory (see the "User Functions" section for more details).
Once the "User Function" is selected in the Stress properties definition, the user
should copy the corresponding User Functions from the Template files (using the
"Copy" function of the Manager - the templates are in the
...\templates\MechanicalProperties\ directory of the software installation).
The figure hereafter is showing an example of this C function. The yield stress is
assumed to be the average value of the yield stresses of pearlite, bainite and
martensite weighted by the local fractions of these phases.
In the User Function, the user has access to the following entities :
Phase_fraction_quantities [ getFraction() ]
Defect_quantities [ getDefect() ]
(at this stage only one quantity is available)
Defect.porosity_POROS1 (porosity amount calculated by the POROS = 1 model)
Warning
When Phase fractions are used, one should of course make sure that the
corresponding quantities are well calculated by the Microstructure module for
this particular alloy system. If one of the variables is not activated, the calculation
is stopped with an explicit error message. In order to know which variable are
available, one could run a calculation with constant stress properties and see in
the post-processor which microstructure fields are displayed.
In the argument of each function, there is the Material ID (or Domain ID). This
means that if such function should be used in different materials, it is possible to
set different properties for each domain (respectively material). The variable is
"matID". The other available variable is the temperature ("temp").
ProCAST has the possibility to compute the final stress properties as a function of
the final microstructure.
To do that, five User Functions are provided to the user in order to define up to
five mechanical properties.
The structure of these User Functions is identical as the ones described in the
"Stress Properties depending upon the Microstructure" section. The Phase
fractions, Microstructure parameters and Defects quantities that can be used in the
Function are the same. The only difference is that as we are calculating the final
mechanical properties at room temperature, the temperature is not available as an
argument of the function. However, the Material ID (matID) is available to the
user.
The following figure is showing an example of the final Yield Stress at room
temperature as a function of the phase fractions. Please note that the name of the
property, as well as the units are defined in the function, and will be displayed in
the post-processing.
Notes
The following figure is showing the final Yield stress property in a simple part,
after two different heat treatments (quenching and air cooling). As the
microstructure is different in both cases, the final properties are thus also different.
"2-D"
Since version 2008.0, two possibilities are available for 2D modelling :
In order to use the real 2D version, nothing special has to be done, except to load
in PreCAST a 2D mesh instead of a 3D one (if a "planar" surface mesh *.sm is to
be used, just rename it as *.mesh before loading it into PreCAST). At this stage,
no specific documentation is available (besides the standard ProCAST
documentation) for the real 2D solver.
INTRODUCTION
Pseudo 2D modelling
The principle is that 2-D calculations are replaced by 3-D calculations on one
layer of elements. To do so, the 2-D mesh generator was modified in order to
automatically create this layer of element, in order to reproduce either a 2-D
cartesian situation or a 2-D axi-symmetric case.
On the left (see figure below), the original 2-D mesh is transformed in a "one
layer" 3-D mesh (right) for cartesian calculation (only a small thickness is
modeled).
On the left (see figure below), the original 2-D mesh is transformed in a "one
layer" 3-D mesh (right) for axisymmetric calculation (only a small sector is
modeled).
MESHING
To call the 2-D mesh generator, one should do a RIGHT mouse click on the
"MeshCAST" menu in the manager and the following sub-menu will appear
One can then select between 2-D or 3-D meshing. If MeshCAST-2D is selected, a
new windows opens with the 2-D meshing interface. This interface was adapted
from the 2-D mesh generator of CALCOSOFT. It allows to mix triangles and
quadrangles in the same mesh and to control quite well the size and the shape of
the mesh.
Firstly, the MeshCAST-2D window opens and the user can select between a new
project (i.e. a new mesh) or to load a previous project.
Then, the drawing area should be defined. The units should be selected (1), and
the size of the drawing area must be defined (2). A grid is superimposed to the
drawing area (to help for the drawing). Its characteristics can be defined (3).
Finally, the type of problem (cartesian or axisymmetric) has to be selected (4).
The "Continue" button should be used to go to the next window (5).
The drawing can start at this stage. The type of entity to draw should be selected
with the corresponding icon (1). Then, this entity can be drawn in the graphics
area (2). One can change between the select and deselect mode with the
corresponding button (3). The check box in-between is to activate the select or
deselect all mode.
One should notice that the intersections should be defined by the user, with the
left icon in (4). One should select the two lines which are intersecting and then
press this icon.
Once the drawing is finished, the domains definition can be done in the next
window, after pressing the "Continue" button (5).
Before meshing, the different domains should be defined. One should first "Add"
a new domain (1) - the corresponding domain number will appear in the list (2).
Then, the contour of the domain should be selected in the graphics area (3). This
operation should be repeated for each domain. Then, one can go in the "Meshing"
window with the "Continue" button (4).
The different domains are listed in (1). This mesh generator allows to define both
structured meshes (quadrangles) or free triangular meshes (2). Firstly, one should
select the domain to be meshed (1). Please note that structured meshes should be
meshed first. Then the type of mesh should be selected (2). If a structured mesh is
selected, one should define the four faces defining this domain (3). Then, the
number of elements along each of the four faces should be defined (3). Please note
that it is possible to have more than one line to define a face. Finally, the mesh
button can be pressed (4) to generate the mesh.
In the case of "unstructured" mesh (i.e. triangles), the operation is more simple as
one should just select the domain to be meshed (1), select the "triangle" mode (2),
define the average mesh size (3) and mesh (4).
Once all the meshes are generated, one can "Continue" to the next window.
The last operation which can be done is the grouping of different domains
("combine"). This operation is optional. One should highlight all the domains
which should be grouped into one (1) and then the "Combine" button should be
pressed (2).
Once this is done, it is possible to save the mesh (3). At this stage, automatically,
the 2D mesh is extruded into a 3D one.
In the case of axisymmetric cases, it is important to make sure that the axis of
symmetry is well at y = 0. Due to round-off errors, one may have slightly negative
coordinated of the nodes which are on the axis. In this case, the "Tools/Check
Axisymmetric" menu should be activated.
CASE SETTING
If the "cartesian" option was selected, the mesh will look like that :
If the "axisymmetric" option was selected, the mesh will look like that :
Such cases should be set-up exactly in the same way as 3-D cases. One should just
take care of the following points :
A symmetry boundary condition should be set on the two large faces of the model,
as shown hereafter.
CAFE-3D
Pre-processing
The set-up and the use of the CAFE-3D module is fully described in the
CALCOSOFT-3D manual (this manual is available in PDF format in the
ProCAST installation, in the dat/manuals/pdf directory). This section is
summarizing the specific aspects of the CAFE model related with ProCAST.
• "post-processing" mode
• weak coupling mode
• full coupling mode
In all cases, the ProCAST thermal case should be set as usual in PreCAST. In
addition to the standard set-up, one should specify in PreCAST the "nucleation"
boundary conditions. To do so, in the "Boundary Conditions/Assign surface"
menu, one should "Add" nucleation BC's in the list (one entry per type of
nucleation BC).
Then, the different surfaces of the casting should be selected and stored in the
corresponding nucleation entry (as for any other BC). Please note that no data
should be allocated to these nucleation entries, as this will be done in the separate
CAFE pre-processor.
If the "full" or "weak" coupling mode is used (for the CAFE calculation), the two
following run parameters should be set as :
THERMAL = 2
MICRO = 1024
Before starting the CAFE set-up (in the CALCOSOFT-3D CAFE pre-processor),
one should run first DataCAST. This will create the unformatted mesh file
(prefixg.unf), which is necessary for the loading into the CAFE pre-processor.
Then, the CAFE-3D pre-processor can be started. To launch it, one should make a
right click in the Manager and select the "CAFE pre-processor".
Please refer to the CAFE-3D manual for details about the CAFE-3D set-up.
The CAFE-3D pre-processor will create all the necessary inputs for the
microstructure calculation. All these inputs are stored in an ASCII file called
"prefix_mica.d". This file can also be edited by hand.
Solver
To run the CAFE calculation, one should press the "ProCAST" button of the
Manager which will open the following window :
To use the Full or weak coupling mode, the standard ProCAST executable must
be launched (one can in this case either use the Standard Run button or the one
under "CAFE Solver (full-weak coupling mode)". The CAFE will be activated as
the Run parameter MICRO=1024 is defined. The distinction between weak and
full coupling is defined by the parameter "nheat" (card 70) in the prefix_mica.d
input file (weak coupling : nheat = 0 / full coupling : nheat = 1).
Post-processing
To view the CAFE results, the CALCOSOFT-3D post-processor must be used. It
can be accessed directly from the ProCAST Manager with a right click on the
"ViewCAST" button.
INVERSE MODELING
INTRODUCTION
The Inverse Modeling module of Procast allows to determine, from simple and
well controlled temperature measurements, thermophysical data or boundary
conditions.
The principle of direct and inverse modeling is presented in the figure hereafter.
Direct calculation
In a standard Procast calculation (i.e. a "direct" calculation), one has the following
inputs and outputs :
Inputs :
• geometry
• thermophysical properties
• initial conditions
• boundary conditions
Outputs :
• thermal history
Inverse calculation
The principle of inverse modeling is to provide the thermal history as an input (i.e.
measured cooling curves at several locations) and to have one of the four inputs
described above as an output (i.e. unknown).
a) Thermophysical calculation
Inputs :
• geometry
• initial conditions
• boundary conditions
• thermal history
Outputs :
• thermophysical properties
Inputs :
• geometry
• thermophysical properties
• initial conditions
• thermal history
Outputs :
• boundary conditions
In this manual, the unknowns will be called by the generic name of "properties" or
"beta value". A "property" (or a "beta value") can thus be a specific heat, a
thermal conductivity, a latent heat, an interface heat transfer coefficient, a film
coefficient, a heat flux or an emissivity.
MODEL SET-UP
Before running any inverse calculation, first, the "direct" problem has to be set-up.
For the setting of a direct ProCAST calculation, please refer to the Pre-processing
chapter.
Once the direct problem is set-up, the file prefixd.dat contains all the information
about :
• geometry (mesh)
• thermophysical properties of materials
• interface heat transfers
• boundary conditions
• initial conditions
• run parameters
The inverse modeling set-up intend to specify on which data the inverse
calculation will be done and to configure the run parameters for the calculation.
PreCAST allows to set-up the inverse problem in an interactive manner (with the
"inverse" menu), in order to create the prefixid.dat file.
One have first the choice to select among a "Material Properties" problem or a
"Boundary Conditions" problem.
Material Properties
If one wants to run a "Material Properties" inverse calculation, the following panel
should be activated (with the "Activate" check box) and configured.
For each material present in the model, the user has to decide whether he would
like to run an inverse calculation on the Specific heat (Cp), the thermal
conductivity (K) and/or the latent heat (L).
By default, all the properties are deactivated ("N" stands for No). By clicking on
the "green N", it will toggle to "Y" (for Yes). In the example below, it is intended
to run an inverse calculation on the Aluminium alloy specific heat and latent heat
(no inverse calculation on the thermal conductivity).
Please note that one should run an inverse calculation on one material domain
only.
If the property (like the specific heat) is defined in the material property file as a
constant, the inverse calculation will be done on this constant value. If the value is
defined as a temperature function, the inverse calculation will be performed for
each temperature of this function.
Important
For an inverse calculation involving "Specific heat" determination, the RUN
PARAMETER "THERMAL" should be set to 2 ( THERMAL 2 ). This is
mandatory, otherwise, the solution will not converge. In any other configuration
not involving specific heat, any thermal model can be used (THERMAL = 1 or 2).
Boundary conditions
On the left, all the available interfaces are listed, whereas on the right, all the
"Heat" boundary conditions are present.
For Boundary conditions, one has the choice to perform an inverse calculation on
either the heat transfer coefficient (H), the heat flux (Q) and/or the emissivity (E).
If the property (like the interface heat transfer coefficient) is defined in the
database as a constant, the inverse calculation will be done on this constant value.
If the value is defined as a temperature (or time) function, the inverse calculation
will be performed for each temperature (time) of this function.
Inverse settings
Finally, some setting should be defined.
If the measurements are in different units than °C, one should specify it with the
"Measurement Units" setting.
multiplied by the beta value. Values of varb between 0.05 - 0.2 are convenient and
correspond to a variation of 5 to 20% of the beta values
Default value : 0.1
itermax : maximum number of iterations before the stop of the
calculation
It is useful to define the maximum number of iteration at which the calculation is
stopped. In some cases, if the tolerance is too small, the variation of the beta
values will not be within the tolerance, although the calculation would have
converged. A value between 15 and 30 iterations is reasonable.
Default value : 15
For each beta value (i.e. for each unknown property), the following parameters are
defined (default values are automatically proposed by precast, see below):
ibindex :
material number, interface number, boundary number (this corresponds to the
number which appears before the material, interface or boundary condition item in
the corresponding window)
itunit :
unit code for the temperature or time "tbeta" used for temperature-dependant or
time-dependant properties
ibunit :
unit code for the beta value
tbeta :
temperature or time corresponding to the beta value (zero when a CONSTANT
value is used)
beta :
beta value (unknown property), initial guess the initial guess corresponds to the
value initially defined in the HEAT boundary condition of the direct calculation.
ibeta :
flag to indicate what is the type of the beta value (constant, time- or
temperature-dependant) :
0 : The corresponding beta value is not used in the Inverse calculation (i.e. a
constant value equal to "beta" will be used in the calculation and nothing will be
done at the level of the inverse calculation). This is very convenient to "freeze" a
given beta value which does not need anymore to be calculated.
1 : The type of the beta value is CONSTANT
2 : The type of the beta value is TEMPERATURE dependant
3 : The type of the beta value is TIME dependant
weight :
weighting coefficient for the beta value
betamin :
minimum accepted value for beta
betamax :
PreCAST proposes default values that may be changed by the user in the
prefixid.dat file (by editing the prefixid.dat file).
For physical reasons, the beta values should not be smaller than a specified value,
betamin (i.e. a heat transfer coefficient should not be negative !). This limit is set
in order to speed up the convergence.
betamax : maximum accepted value for beta
For physical reasons, the beta values should not be larger than a specified value,
betamax (i.e.. a thermal conductivity of steel can not exceed a value of pure
copper or pure silver !). This limit is set in order to speed up the convergence.
INVERSE RUN
Then, the calculation can be started. To do so, a right click on the "ProCAST"
button on the manager is opening a sub-menu, which allows to start the inverse
solver.
The status of the inverse calculation can be viewed with the "invstat prefix"
command (from a second Command Window).
Output Files :
During the execution of the calculation, several files are created, beside the usual
ProCAST files :
prefixir.dat
This file contains the evolutions of the beta values along the iterations, as well as
the evolution of the residual.
prefix.stat
This file contains information of the status of the calculation (see
below)prefixic.dat
This file contains the calculated temperature evolutions (of the last calculated
iteration) corresponding to the locations of the measured curves.
prefix.tmp
This file is a temporary file created during the calculation. This file is needed only
during the calculation and is useless for the user. This file is automatically deleted
at the end of the calculation However, if the calculation is interrupted by the user
before completion, it should be deleted manually by the user.
prefix.list
When the calculation is run in the background mode, the printout which appears
normally on the screen is listed in this file. This file is not written in a continuous
mode, but only when the buffer of the machine is unloaded (thus, this may take a
few minutes after the beginning of the calculation before having something in this
file).
As inverse calculation could be quite long and are often run in "background
mode", there are three ways to monitor the status of the calculation. The
monitoring of the calculation is a good way to check whether the calculation is
stable and whether it is converging in good conditions.
prints the status of the calculation (it is equivalent to "prostat" in Procast). The
number of iterations already calculated is printed as well as the beta values of the
last iteration. This can also be accessed from the manager with a right click on the
"Status" button.
The residual of the current iteration is also printed. The residual is the average
temperature difference between the measured and the calculated curves (average
on all the steps and all the curves, at the given iteration). The residual will
decrease towards zero along the iterations. The monitoring of the residual is a
good way to see how the convergence is reached.
This command should be called in the current directory where the prefix* files are
present (this is not like "prostat" which can be called from any directory).
The beta values obtained at each iteration are stored in the file prefixir.dat Thus,
during the calculation, it is possible to visualize the evolution of each beta value
(from the first to the current iteration) and see whether the calculation is
converging or not. The format of the file prefixir.dat is the same than the one of
th
the measurement. Curve "i" of the file prefixir.dat correspond the i beta value.
In addition, the residual is stored as the last curve of this file. Thus, it is possible
to graph the evolution of the residual as function of the iterations.
The goal of inverse modeling is to find the best set of beta values with which the
calculated temperature curves will match the measured ones.
At the end of each iteration, the calculated curves corresponding to the location of
the measurement points are stored in a file labeled prefixic.dat This file is updated
at each iteration. Thus, it is possible to visualize during the calculation (with
ViewCAST) the superimposition of the measured curves and the calculated ones.
Of course, during the first iteration, the match will be bad, however, at the end of
the calculation, the match should be quite good.
Final result
At the end of the calculation, the final beta values correspond to the values of the
last iteration in the file prefixir.dat.
iterations (for a further run) in order to reduce the computing time, without
changing the quality of the result.
In order to check the quality of the determined beta values, one can superimpose
the measured and the calculated curves (see section "Monitoring the calculation"
above). The quality of the match will be an indication of the quality of the final
beta values, as given by the residual.
Remarks
Limits
FILE FORMATS
Definitions :
ntcunit :
unit code for the temperature measurements
ntccode :
code which specifies how the thermocouple locations are defined :
ntccode = 0 : coordinates of the thermocouple
("0" is not implemented in the current version, use only ntcode=1)
ntccode = 1 : node number corresponding to the thermocouple location
node :
node number in the mesh used for the calculation corresponding to the
measurement points (ntccode=1) (the order of the node numbers should
correspond to the order of the cooling curves in the file prefixim.dat)
imat :
material number where the thermocouple is located (ntccode=0)
ixyzunit :
unit code for the coordinates when ntccode=0
x,y,z :
x, y and z coordinates of the thermocouple
tau :
time constant for the filtering of the measurements [s]
sigma :
weighting coefficient for temperature [K]
conv :
convergence tolerance [-]
varb :
variation of each beta value during an iteration [-]
itermax :
maximum number of iterations before the stop of the calculation
ibindex(i) :
material number, interface number, boundary number
itunit(i) :
unit code for the temperature tbeta(i)
ibunit(i) :unit code for the beta value, beta(i)
tbeta(i) :
temperature corresponding to the beta value, beta(i)
beta(i) :
beta value (unknown property), initial guess
ibeta(i) :
flag to indicate if the beta value will be taken into account in the inverse
calculation
1: the beta value is an unknown of the problem and will be taken into account in
the calculation (it is a constant value)
2: the beta value is an unknown of the problem and will be taken into account in
the calculation (it is a temperature-dependant value)
3: the beta value is an unknown of the problem and will be taken into account in
the calculation (it is a time-dependant value)
weight(i) :
weighting coefficient for the beta value
betamin(i) :
minimum accepted value for beta
betamax(i) :
maximum accepted value for beta
relax(i) :
maximum variation of the beta value between two iterations
All the parameters and variables defined above are defined in groups (or families),
which are labeled by keywords. The keywords are :
MEASUREMENTS
PARAMETERS
THERMAL_CONDUCTIVITY
SPECIFIC_HEAT
LATENT_HEAT
INTERFACE_HEAT_TRANSFER
EXTERNAL_FILM_COEFFICIENT
EXTERNAL_HEAT_FLUX
EXTERNAL_EMISSIVITY
Format :
MEASUREMENTS
ntcunit [integer]
Loop on the thermocouples (i)
ntccode(i) [integer]
if ntccode(i)=0
imat(i) [integer]
ixyzunit(i) [integer]
x(i) [real]
y(i) [real]
z(i) [real]
else if ntccode(i)=1
node [integer]
endif
End Loop on the thermocouples
PARAMETERS
tau [real]
sigma [real]
conv [real]
varb [real]
itermax [integer]
For all the following keywords, the structure of data is the same. Only the
necessary keywords should be specified, which correspond to the desired beta
values.
THERMAL_CONDUCTIVITY
SPECIFIC_HEAT
LATENT_HEAT
INTERFACE_HEAT_TRANSFER
EXTERNAL_FILM_COEFFICIENT
EXTERNAL_HEAT_FLUX
EXTERNAL_EMISSIVITY
Loop on the beta values (i)
ibindex(i) [integer]
itunit(i) [integer]
ibunit(i) [integer]
tbeta(i) [real]
beta(i) [real]
ibeta(i) [integer]
weight(i) [real]
betamin(i) [real]
betamax(i) [real]
relax(i) [real]
End Loop on the beta values
As the data are read by a "Free Format Reader", data can be placed in any fashion
(lines, columns,...), as long as the sequence of the data corresponds to the
specified one. However, each line should not exceed 256 characters. Moreover,
comments can be included in the data file after a "#" sign (all the characters
between the "#" sign and the end of the line will be ignored and can be used as
comments).
Description :
nx :
number of values on the X-axis (this corresponds to the number of time-steps if
the X-axis is time)
ny :
number of curves
node#i :
node number corresponding to the curves. These number can be any integer
numbers (as they are not used by the inverse calculation). The only requirement is
that ny integer values are present. The reason of these values is that it corresponds
to the format which is compatible with PostCAST.
x(j) :
value "j" on the X-axis
y(i,j) :
value "j" on the Y-axis for curve "i"
INVERSE APPENDIX
Inverse model
Thus, if the problem to solve has N unknowns and M iterations are needed to
reach convergence, it will involve (N+1) x M calculations. For instance if one has
This shows that in order to keep reasonable computing times, it is necessary that
each direct calculation is small and quick. Indeed, if one direct calculation is
lasting 1 minute, the inverse problem example shown above will take 165 minutes
to solve (about 3 hours). If instead, one direct calculation is lasting 2 days of CPU
time, the inverse calculation will take 1 year !!!
There are no generic rules about the design of inverse experiments as it depends
upon the goals that are set. The main and most important rule is "Make it simple".
This means that one should stick to 2D models and that if possible only one type
of unknown should be determine at a time.
In principle, there are no limitation in the inverse modeling towards the maximum
number of unknowns. However, it is strongly advised to determine one thing at a
time. For instance, design one experiment for the determination of the heat
transfer coefficient between the water and the mould in the case of a cooling
channel and design an other experiment for the determination of the
casting-mould interface heat transfer coefficient. Although it is possible in the
principle to determine both together, it is advisable to simplify the problem by
decomposing it.
As a reminder, the 4 inputs which are necessary for the inverse modeling are :
a) geometry
b) initial conditions
c) temperature measurements
d1) material properties (in case of boundary condition determination)
d2) boundary conditions (in case of material properties determination)
It is thus of primary importance to make sure that the quality of these 4 inputs are
as good as possible.
a) geometry
The best way to have an accurate geometry description in 2D is to consider 2D
axi-symmetric problems. Thus, the modeling of cylinders or tubes is
recommended. One should be careful that the running system and the feeding
system must also be axi-symmetrical. In some cases, the use of 2D cartesian can
also be applied with success for the modeling of plates or the determination of
boundary conditions on flat surfaces.
b) initial conditions
Usually, within the direct solver, it is only possible to set easily constant initial
temperatures. Thus, it is important to perform an experiment where the initial
temperatures of the different domains are as constant and well known as possible.
If the filling stage is very important (as one can loose a significant amount of
superheat during the filling), one should consider whether the design is
appropriate or if a lower average constant initial temperature can be considered.
c) temperature measurements
The quality of the temperature measurements are of primary importance in order
to have a good result. Thus, one should be careful to use appropriate
thermocouples (not too thin or not too big). The time response of the
thermocouples is also important if one wants to capture fast transient phenomena.
When measuring in a solid material, one should ensure that there is an intimate
contact between the tip of the thermocouple and the material.
Concerning the placement (location) of the thermocouples, please refer to the next
section.
The idea is to use similar materials where the materials properties are better
known, as the heat transfer coefficient will not depend upon slight differences in
material properties. For instance, in the example shown above, it is not important
to take a steel which is resistant at high temperatures (like it is needed for an hpdc
die), as the inverse experiments will be run only a few times (unlike the hpdc die
which will undergo many cycles, under high loads).
However, the user is able to check the quality of its design by himself. It is
strongly recommended to model first the experimental set-up before even building
it. Of course, as the properties or boundary conditions, such a modeling has to be
done with guessed values (one should notice that it is important to have at least an
idea about the order of magnitude of the unknown value before starting to explore
inverse modeling techniques).
Based upon this first modeling, it is possible to run a few additional calculations
by modifying the guessed values. This corresponds to some kind of "sensitivity
analysis". Based upon the results of these calculations, it will be possible to see
whether the calculated temperature curves at different locations are sensitive or
not to the variation of the unknown (or guessed) values. This provides a useful
mean of determining where are the best locations to place the thermocouple.
Indeed, one would like to have the largest temperature variations when the
unknown values are changing slightly (for instance, it is useless to place a
thermocouple at the first floor of a building to measure the influence of the
opening of a window at the 15th floor !).
Of course, it is advised to place thermocouples as close as possible from the
location where the property has to be determine. For instance, if one wants to
determine an interface heat transfer coefficient, it is advisable to place
thermocouples quite close from that interface, on either sides. However, one
should be careful not to place thermocouples too close from the surface as it is
very difficult to measure accurately very close from surfaces. Thus, it is better to
have a better quality measurement, a little bit further away from the surface, than a
low quality measurement, very close from the surface.
Concerning the number of thermocouples, one shall first say that the computing
time does not depend at all upon the number of thermocouples (it only depends
upon the number of unknowns). However, experience shows that it is advisable to
use a small number of thermocouples (less then 10) to obtain good results. Thus,
the user has to concentrate its attention about where are the optimum locations of
the thermocouples rather than "flooding" the experiments with many
thermocouples. Moreover, the inverse solver is based upon a "least square fit" on
all the thermocouples, in a global fashion. Thus, if some thermocouples are
"useless" and provide not very accurate measurements, they may degrade the final
answer.
Finally, the advantage of modeling first the experiments, before running them, is
also to show how much CPU time will be needed to run one direct calculation.
Thus, it will show whether the set-up is appropriate or not for fast inverse
calculations.
If the calculated curves are quite close from the measured ones, one can run the
inverse calculation. If on the opposite, the calculated curves are very far away
from the measured ones, it is good to re-run once more the direct calculation with
new guesses which seems more appropriate. This "manual" inverse modeling may
speed-up considerably the inverse calculation as convergence will be reached
much faster.
One expects that the thermal conductivity and the specific heat will be both
function of the temperature. However, in a first stage, it is good to run an inverse
calculation with constant properties. This means that the inverse modeling is
looking for one value of specific heat, one value of thermal conductivity and one
value of latent heat. Figure A1.1 shows the result of such a calculation with the
superimposition of the measured and calculated curves. The corresponding
constant material properties are also shown. One can see that the correspondence
between the measurements and the calculation is only fair, which is not a surprise
as constant properties were considered. This first calculation provides a way to
"calibrate" the unknowns. Then, it is possible to refine those unknowns in a
second inverse calculation. Figure A1.2 shows the superimposition of the
calculated and measured curves, when thermal conductivity is described by a
temperature function using 3 unknowns and the specific heat by a temperature
function using 4 unknowns (the latent heat is still described by one value). The
initial guesses used in this second inverse calculation corresponded to the result of
the first inverse calculation. One can see on figure A1.2 that the correspondence
between the calculation and the measurements are much better in this second case.
Then a third inverse calculation was performed, using a 6 points specific heat
curve and a 5 points thermal conductivity curve (an one latent heat). Again the
initial guesses used in this latter calculation were taken as interpolation of the
results of the previous calculation. Figure A1.3 shows a very good correlation
between measurements and calculation.
One shall notice that in the above example, it would certainly been possible to
find directly the final result (of Figure A1.3). However, in complex problems, it is
not always possible to start with a reasonable initial guess and thus, convergence
can be hard to find (or even impossible in some cases).
In such a case, the user has to decide the type of dependency that should be used
in the inverse calculation. To illustrate that, one can use the example of the
determination of the equivalent specific heat of an alloy (as shown in Figure
A1.4). One can see that the curve is almost constant above and below liquidus,
whereas it exhibits very strong variations between the liquidus and solidus
In such a case, the user has to define himself the temperature points at which the
unknowns will be calculated. As in the above mentioned example, these points
should be selected in between the solidus and liquidus temperatures (with a
reasonable number of points in order to describe the curve).
Above the liquidus or below the solidus, one can consider constant values or
variable values as desired.
Concerning the number of points to be selected, one should select enough points,
but not too many. Indeed, if too many points are selected, one can obtain
oscillations of the final beta values.
• Are the thermocouples well defined ? Is the order of the list of the
thermocouple corresponding to the order of the measured curves in the
measurement file ?
• In case of time- or temperature-dependent beta values, are the tabulation points
appropriate (for instance, if one is looking for a variation between 500 and 700
and the tabulation points are chosen between 600 and 800, it might not work).
• Are they any (or enough) measurements in the range of the tabulation points?
(for instance if one looks for a temperature dependant heat transfer coefficient
between 100 and 800°C and that the measurements are only between 400 and
700°C, the values below 400 and above 700°C will never converge).
• Are the initial temperatures of the model corresponding to the ones of the
measurements (or are the initial measured temperatures constant ?).
• Is the geometry of the model and the known boundary conditions well defined
?
• Are the material properties of the different materials used in the measurements
well known ?
CONTINUOUS CASTING
This chapter presents the methodology to model continuous casting with
ProCAST.
Two different approaches, steady state and non steady state, will be presented.
PRINCIPLES
The modeling of Continuous casting can be addressed in two different ways :
Note :
"DC casting" can be addressed as well and it should be understood that it is
included in the generic name of "continuous casting".
For steady state computations, a fixed domain is modeled and the solid is
transported at the casting speed through this domain.
For non-steady calculations, the domain is extending with time, as the continuous
casting process proceeds. A new algorithm has been developed and implemented
in order to allow this evolving domain. This algorithm is called MiLE (for Mixed
Lagrangian-Eulerian).
In the same line, it is not possible to model the start-up phase of continuous
casting with the steady state method (again the MiLE algorithm allows to do that).
At the beginning, the casting is divided into two domains (1 and 2 above). Then,
as the continuous casting process starts, the bottom domain (2) is moving down
(while the domain 1 is staying at the same position). In order to have a continuity
between domains 1 and 2, new elements have to be introduced between domains 1
and 2 (creating the domain 3 above). During the process, layers of new elements
are introduced in domain 3 at the interface between domains 1 and 3.
The following figure is showing the principle of the MiLE algorithm (Mixed
Lagrangian-Eulerian), with the region where the calculation is Eulerian and the
ones where it is Lagrangian.
thickness are created in between domains 1 and 2. Then, during the casting
process, these layers are "unfolded" as an accordion.
The number of pre-defined layers as well as the thickness of the layers are
specified in the pre-processor (see the pre-processing section for more details).
During the calculation process, the domain 2 is translated at the casting speed and
the accordion will unfold itself in between domains 1 and 2.
The following figure is showing how the accordion is unfolding during the
continuous casting process.
As the mesh in the accordion and in the domain below the accordion (labeled
domains 2 and 3 in previous figures) are moving, a transport term of the mesh in
accounting for this motion.
However, the MiLE allows to couple stress calculations with the heat and fluid
computations. The only restriction is that the solid should not cross the "accordion
interface" (i.e. the interface between domains 1 and 3). Thus, the accordion
interface should be located in the appropriate position in order to avoid such
situation (it is possible to have a non-planar accordion interface to resolve this
question - see the pre-processing section for more details).
PRE-PROCESSING
To handle the fact that we have a continuous process, a transport term of the solid
is added to the equations. As soon as the metal is solid (dark red in the figure
above), it should be transported at the casting speed (i.e. the continuous casting
velocity - see the blue arrow in the figure above).
If fluid flow is activated, the flow will automatically account for the transport of
the heat in the liquid zone (light yellow in the figure above). In the mushy zone,
the model accounts for a "mix" of the transport term of the solid and the transport
of fluid of the liquid.
If the fluid flow model is not activated, the transport term is not only applied to
the solid, but also to the liquid and to the mushy zone.
3. At the interface between the mold and the casting, an interface heat transfer
coefficient is set as usual.
To define the "Solid transport" velocity, the following panel should be filled :
One can see in the figure below how the temperature are changing in the casting
during the calculation. Only the last result corresponding to the steady state is of
interest. All the other intermediary results should be discarded.
A Wall BC (zero velocity) should be applied on the external faces of the casting
domain and on the interfaces between the casting and the mold if a non-coincident
interface is used (please note that in the present example, the large faces are set as
symmetry planes).
A Pressure BC should be applied (with a 1 bar pressure) on one node of the inlet
(on top of an inlet velocity BC).
Finally, the FLOW = 1 Run Parameter should be set. In addition, the value of
LVSURF should be set to a higher value than 1 (e.g. 1.10). As mentioned above,
the value of PREF should be set to the same value as the one used in the Pressure
BC (1 bar in this example). It is also very important to set FREESF = 0.
The mesh should contain a small extra volume which has the same area than the
real inlet (see the picture below).
This allows to specify exactly the inlet velocity corresponding to the casting speed
(by making the ratio of the surfaces).
Then, a Wall BC should be applied to the rest of the top surface (see below).
Finally, the Pressure BC should be applied on one node of the inlet surface.
Such non constant solid transport velocity should be defined in the ad-hoc User
functions.
Firstly, the velocities at the inlet, outlet and along the faces of the casting should
be assigned to a given Velocity BC (please note that in the example of the
following figure, the large faces are defined as symmetry planes).
In the same way, the solid transport term should be set with User functions. Firstly
the Solid transport entry should be defined as follows (with Function for the X
and Y components) :
This will allow to have the velocities defined as shown in the following picture
(only the imposed velocities are shown in this case) :
When the calculation is run, the following temperature and velocities are obtained,
showing well that the velocities in the solid and in the mushy zone are transported
at the casting speed (which is turning with the casting).
Finally, as for any flow calculation, a Pressure BC should be set on the top of the
casting inlet.
Firstly a thermal only calculation can be run until the steady state is reached. Then
a new calculation is set where both the thermal and the flow solvers will be
activated. However, the initial temperatures of this new calculation will be the one
of the first thermal only calculation. As the steady state was already reached in the
first thermal only calculation, the steady state of the second (thermal+flow)
calculation will be reached much quicker (as the influence of the flow on the
thermal solution will reach steady state very quickly). For instance, in the
previous example of curved continuous casting, the steady state with the
thermal+flow calculation was reached after about 50 seconds (of physical time),
corresponding to 170 steps, whereas the steady state of the thermal only
calculation was reached already after 60 steps - corresponding also to 50 seconds.
(in addition, please note that the CPU time for a thermal only calculation is much
shorter as a calculation for a thermal-flow calculation).
When applying this method, one has to be careful when a nozzle is present (or
when the inlet of metal is not applied on the whole top surface). In this case, it is
necessary, in the thermal only calculation, to set an "imposed temperature" on the
whole inlet surface (and to remove it afterwards for the thermal+flow calculation).
Remark
Please note that the POROS 1 model is not suitable for continuous casting (both
steady state or MiLE) and thus POROS should be set to 0.
The following set-up corresponds to a THERMAL ONLY case. For the activation
of the fluid flow or of the stress, please check the MiLE Examples.
Meshing
The first requirement is at the level of the mesh. The initial casting domain should
be meshed in two distinct domains, as shown in the figure hereafter (the molds
and bottom blocks should be defined as usual).
This will allow to define the "Accordion" location, which will be at the interface
between these two domains. The Accordion should be located high enough in
order to make sure that there will be no solid which will "cross" the accordion (i.e.
the Accordion interface should remain fully liquid).
In this chapter, the domain "above" the accordion (shown in green in the previous
figure) will be called the "upstream domain" (considering that we are casting in
the vertical downwards direction). The domain "below" the accordion (shown in
red) will be called the "downstream domain".
If the nozzle is deep and the solidification along the mold wall is expected to
reach a higher level than the bottom of the nozzle, it is allowed to have a
non-planar accordion interface (see below).
Geometry menu
The definition of the parameters in the "Geometry menu" should be done as usual.
The MiLE algorithm allows to use symmetries, however, as it is not possible to
activate the radiation with View Factors (ON) with the MiLE algorithm, it is not
necessary to define symmetries in this menu (but only in the Boundary Conditions
menu).
One should note that it is not possible to use a Virtual mold with the MiLE
algorithm.
The Volumes which are indicated in this menu correspond to the initial volume of
the mesh (and not the volume when the Accordion is unfolded).
Materials definition
The materials should be defined as usual. Useless to say that the material
properties on both side of the accordion should be identical and it should be
labeled as "Casting".
Interfaces
All the interfaces between mold domains should be defined as usual.
The interface(s) between the upstream domain and the mold(s) can be set as
coincident (COINC).
The interface(s) between the downstream domain (including the bottom block if
any) and the mold(s) should be set as non-coincident (NCOINC). This is because
the downstream domain (including the accordion which will belong to this
domain) and the bottom block will move with respect to the mold during the
casting process. Thus, a non-coincident interface will handle the heat transfer
between the moving casting and the stationary mold.
Boundary conditions
In order to set the Accordion link, two steps should be done in the set-up.
The "Link" button will open the following panel. The "Apply" button should be
pressed without any change in the content of the default panel (i.e. zero translation
vector and it is important to have one non-zero value in the Rotational Axis
definition).
The number of layers of the Accordion, as well as the element thickness (once the
element is unfolded) should be defined. As all the layers are created at the
beginning of the calculation (in DataCAST), it is advised to define a number of
layers which is not too large, corresponding to the final desired length of the
casting (in order to keep a reasonable size of the model).
Once the Accordion is set, it is necessary to define the cooling conditions on the
side of the continuous casting slab, once it is going out of the mold (if there is a
mold).
As shown in the figure hereafter, the lateral cooling of the casting (below the
accordion interface) is rather complex. One has first an interface heat transfer
between the casting (and bottom block if any) and the mold. Then, as the casting
is going out of the mold, there is a convective cooling (i.e. a Heat BC). As the
domain is moving, one should be able to handle the transition between an
interface heat transfer and a Heat BC transfer.
In order to achieve that, the Heat BC must be defined by a User function (see the
"User Functions" section of the manual for all the details). To do so, firstly, the
Heat BC entry should be defined as follows :
The Film coefficient should be defined by a "Function". This user function will
allow to define a film coefficient as function of the position of the surface
element, so that it applies only when the corresponding surface of the casting is
outside the mold. If we suppose that we have a vertical continuous casting
(downwards), the heat transfer coefficient will be set to zero if the position of the
surface is higher than the exit altitude of the mold (e.g. 0.02 m in the Y-direction)
and it will be set to a given value (e.g. 3000 W/m2K) if the position is lower than
the exit altitude of the mold.
Such a User function (which can be of course much more sophisticated) will allow
that there is no convective heat transfer inside the mold and that the right
convective cooling will be applied for any surface going out of the mold.
Process conditions
A Translation velocity (corresponding to the casting speed) should be applied to
the "downstream domain", as well as to the bottom block (if any).
The "Translation" panel allows to define the translation as x(t), v(t) or v(x) - see
below.
Initial conditions
The initial conditions will be defined as usual.
Run Parameters
In addition to the usual Run parameters, one Run parameter should be added
manually in the p.dat file.
Remark
Please note that the POROS 1 model is not suitable for continuous casting (both
steady state or MiLE) and thus POROS should be set to 0.
MILE EXAMPLES
Thermal + Flow
In addition to the Thermal set-up (see the "Non Steady State cases (MiLE)
section"), the following definitions should be done to activate the flow.
Boundary conditions
An inlet velocity BC should be set on the top surface. In the following example,
the inlet velocity is applied on the whole top surface (which is equal to the exit
section of the ingot). In such case, the inlet velocity should be equal to the
translation velocity, applied to the downstream domain and the bottom block.
Run Parameters
Finally, the FLOW = 1 Run Parameter should be set. In addition, the value of
LVSURF should be set to a higher value than 1 (e.g. 1.10). As mentioned above,
the value of PREF should be set to the same value as the one used in the Pressure
BC. Finally, one should set FREESF = 0.
The mesh should contain a small extra volume which has the same area as the real
inlet (see the picture below).
This allows to specify exactly the inlet velocity corresponding to the casting speed
(by making the ratio of the surfaces).
Then, a Wall BC should be applied to the rest of the top surface (see below).
Finally, the Pressure BC should be applied on one node of the inlet surface.
Thermal + Stress
This section describes the setup of a Thermal-Stress case. For the setup of the
Thermal part, please refer to the "Non Steady State cases (MiLE) section").
Stress properties
The stress properties should be specified as usual in the Materials/Stress menu.
One should remind that it is mandatory that the solidification front does not
"cross" the Accordion interface (i.e. the "upstream domain" should remain fully
liquid).
Please note that at this stage, it is not possible to perform a stress calculation in the
bottom block. However, the mold(s) can be considered as Rigid, Vacant or a
stress calculation can be performed.
Boundary conditions
No Displacement BC should be set on the Casting domain. This is due to the fact
that the liquid nodes are automatically set with a zero displacement. Thus, the
model will be enough constraint for the stress calculation.
If stresses are also calculated in the mold, it is necessary to set the usual
Displacement BC on the mold.
Run Parameters
Beside the STRESS = 1 Run parameter, a special care has to be taken with the
SCALC value.
Due to the MiLE algorithm, the number of elements in which the stress
calculation will be done will increase with time. It was observed that the
maximum timestep (dtmax), the casting velocity (V), the thickness of the
accordion elements (E) and the SCALC values should be defined in order to
satisfy the following relationship (in order to have a good accuracy of the results)
:
CORE BLOWING
ProCAST allows to model the core blowing process.
Introduction
When modeling the sand injection process, the air and sand are modeled as a
homogeneous fluid in which the air-sand mixture is treated as a single phase.
Thus, the process is considered as a filling with ad-hoc material properties and
boundary conditions.
where,
Model set-up
Material properties
The viscosity model should be defined using the Carreau-Yasuda input panel with
the following correspondence table:
Please note that the "Yasuda" parameter is not used in the Core blowing viscosity
model. Thus, this value (which is recommended to be set to 0) will be ignored.
Remark :
It is important to notice that when a core blowing calculation is performed
(NNEWTON = 7), the equation behind the Carreau-Yasuda model (shown above)
is not the same as the one used for non-newtonian fluid.
In all cases, the viscosity is never allowed to decrease below the µ0 value. The
last term of the viscosity relation measures the amount of contraction of the
velocity field. Only positive contractions are used to increase the effective
viscosity. This contraction term is modeling the increased resistance to flow that
occurs when intra-particle spacing decreases as the velocity field contracts.
Beside the viscosity, it is necessary to define the density, the specific heat and the
thermal conductivity. Please note that as no thermal model is performed, the
specific heat and the thermal conductivity are not explicitly used in the
calculation. However these two values should be defined (dummy values can be
used).
A pressure condition should be applied to the sand in the upper container to force
the sand down through the nozzle assembly and into the core box. Usually, this
pressure is around 4 bars.
Vents should be specified in order to allow the air to escape. The two following
figures are showing different vent selections.
Run parameters
The Standard Flow Run parameters should be set as usual. One should not forget
to set GAS = 1 to activate the gas model (mandatory) :
In addition, the core blowing model is activated by setting the following flow run
parameters: HIVISC = 2 and NNEWTON=7 (in the Flow/Advanced 1 Run
Parameter tab).
Experimental comparison
The following figure is showing the comparison between experimental
measurements (left) and the ProCAST calculation (right) at 3 different times.
Remarks
Please note that this Hot Cracking model (HCS) can be used only in steady-state
conditions, as encountered in continuous (or DC) casting. Thus, this model can
not be used for shape casting (as it is non-steady by essence).
Moreover, the HCS model can be used only for straight continuous casting.
INTRODUCTION
In many continuous casting processes, the appearance of hot tears has a
detrimental effect on the productivity, especially for some very sensitive alloys.
The new hot tearing criterion, recently derived by Rappaz, Drezet and Gremaud
[1] has been implemented in calcosoft for 3D geometries and steady state
situations in order to model the run conditions of a continuous casting process.
The strategy adopted here is very similar to the Stationary Inverse module of
calcosoft : once the temperature field reaches steady state, a set of profiles are
drawn parallel to the casting speed in order to determine the position of the Tcg
(temperature at coalescence of grains) isotherm and the value of the components
of the thermal gradient at that position. Then the hot tearing sensitivity (HCS) is
calculated using the so-called RDG criterion for each profile. Hence in 3D
situations, HCS is represented in a section perpendicular to the casting speed.
grow in a given thermal gradient, G, and with a velocity given by vT. Above a
certain volume fraction of grains, mass feeding can no longer compensate for
shrinkage, the specific mass of the solid being larger than that of the liquid for
most metallic alloys. Therefore, the liquid has to flow from right to left in a
packed bed of solid grains.
If the dendritic network is submitted to a tensile deformation perpendicular to the
thermal gradient, the flow should also compensate for that deformation if no hot
tears form. The pressure in the interdendritic liquid is schematically represented at
the bottom of Figure 1: it decreases from the metallostatic pressure, pm, near the
end of mass feeding.
Above the mass feeding temperature, Tmf, the grains have not yet coalesced and
are free to move within the liquid. On the other hand, below the temperature at
which coalescence of the grains takes place, Tcg, all the grains form a coherent
solid network which can transmit the thermal stresses induced by cooling. Note
that the temperature at which coalescence between two grains occurs depends on
their misorientation. Between Tmf and Tcg, the film of liquid can only resist up
to a cavitation pressure at which a void is nucleated and can develop into a hot
tear. Any opening of the continuous interdendritic liquid film present in the
packed bed of grains can hardly be compensated for by feeding from the upper
region of the mush because of the high volume fraction of solid (i.e., low
permeability).
The RDG criterion is therefore based on the derivation of the two pressure drop
contributions associated with deformation and shrinkage respectively. To do so, a
mass balance is performed at the scale of a small volume element of the mushy
zone in a reference frame attached to the isotherms [1].
Assuming no porosity formation, the volume fraction of liquid, fl, is equal to (1-
fs) and the specific masses of the two phases, rs and rl, are assumed to be
constant, but not equal (solidification shrinkage factor b). The velocity of the
liquid is related to the pressure gradient in the liquid via the Darcy equation and
the permeability of the mushy zone is given by the Carman-Kozeny
approximation [1].
Considering that the fluid moves along the thermal gradient only, whereas the
solid deforms in the transverse direction, one can calculate the pressure within the
mush (eq. 1):
where pa is the atmospheric pressure, ρgh the metallostatic contribution. ∆psh and
∆pmec are the pressure drop contributions in the mush associated with the
solidification shrinkage and the deformation induced fluid flow, respectively. In
steady state conditions and assuming an uniform mechanical deformation rate
throughout the mush, , these two contributions are given by (eqs. 2):
Tcg is the temperature at which coalescence (solid bridging) of the dendrite arms
between grains occurs and m the viscosity of the liquid. λ is the mean grain size
for equiaxed structures or the secondary dendrite arm spacing, λ2, for columnar
structures. The two parameters A and B depend only on the nature of the alloy and
Eventually, if the pressure, P, given by equations (1) and (2), falls below the
cavitation pressure, Pc, a hot tear forms. This condition can be rewritten in terms
of depressions (eq. 3) :
where the cavitation depression is defined as Dpc = pa - pc. This condition allows
the calculation of the maximum strain rate sustainable by the mushy zone, ,
and a hot cracking susceptibility, HCS, can be defined as . The higher
HCS, the more susceptible the alloy.
fraction, Fcg, at which coalescence of grains occurs. Note that the integration:
• can be performed between Tcg and Tliq instead of Tmf (mass feeding
temperature) without modifying much their values
• is performed for each solid phase if needed (e.g. eutectic as well as peritectic
reactions can be treated)
To compute the maximum strain rate the mush can sustain, the following data are
required:
computed. In case [1/s] is negative, the mush cannot sustain any tensile
deformation, thus a crack is to be formed and HCS is set to a high value, 1.e6.
If the maximum temperature difference computed between the last two saved
steps is larger than the given tolerance (stationary.t.tol) beyond which steady state
is assumed, the tolerance and/or the number of computation steps to get the steady
state temperature field should be increased. To decrease the computation time, it
can be useful to restart a calculation from a first approximation of the steady state
temperature field by using the restart option.
The steady state hot tearing module needs the following input files :
prefixg.unf (geometry file)
prefixt.unf (results file - temperature)
prefixfs.unf (results file - fraction of solid)
n, Nphase
where n is the number of lines and Nphase the number of solid phases (a, b, g, ...)
(which may be present together with the liquid phase) that are considered (1 for
eutectic reactions, 2 for peritectic reactions). are the solid fractions
(varying between 0.0 and 1.0) at temperature Ti for solid phases a, b, g , … (only
the solid phases which appear together with the liquid should be considered, i.e.
no solid phase transformations are considered). Note that the total solid fraction is
the sum of these quantities. These data are required to compute the parameters A
and B.
IMPORTANT
• the temperatures Ti are stored in the order of increasing values, thus the solid
fraction are decreasing and Tn is the liquidus temperature of the
alloy at which (or the mass feeding temperature at which
).
• the number of values, n, should be large enough in order to precisely compute
the two integrals A and B, notably at high solid fractions.
• the integration of A and B are performed from Tcg to Tn (liquidus or mass
feeding temperature).
• the solid fractions have no unit and should not be expressed in %
• T1 is the solidus temperature at which
Domain number [-] : Domain in which the HCS calculation is performed (it
should be in one and one domain only)
liquid height [m] : coordinate of the height of the top of the liquid pool (it can be
outside the domain). This height will be taken parallel to the gravity vector.
number of solid phases [-] : the number of solid phases should be defined (e.g. :
2 for a peritectic reaction)
Notes
Stationary.t.tol: Maximum acceptable variation in temperatures (°C)
The steady state regime is considered to be reached when the maximum variation,
between the two last saved steps, of the temperature at all nodes is smaller than
the value, stationary.t.tol. The HCS value will be calculated on the last step which
is supposed to correspond to steady state condition. A reasonable value of
stationary.t.tol. is 0.1-1.0 °C.
Cavitation depression:
The cavitation depression depends on the alloy composition and on gas and
impurities content. Typical values for metallic alloys vary between 10 to 100 kPa.
RUNNING A CASE
The HCS calculation can be launched from the Manager. The "ProCAST" button
gives access to the following panel (if activated in the Installation Settings) :
The computed values of the hot tearing sensitivity are stored in the file
prefixhcscrit.usf. In addition, using the Excel csv (comma separated values)
format, these values are also stored in the two files prefix_hcs-column.csv and
prefix_hcs-array.csv, in column and array formats, respectively.
If there are more than one domain (e.g. casting and mould), the values of HCS in
the mould are set to -1.
One can also represent the results with the help of Excel by drawing some
surfaces. The two files prefix_hcs-column.csv and prefix_hcs-array.csv, can be
opened in Excel which automatically recognizes the csv (comma separated value)
format. The file prefix_hcs-column.csv contains the following data, stored in
columns as indicated in the following figure :
x (x coordinate), y (y coordinate), z (z coordinate of the iso-Tcg surface), rdg (hcs
value obtained with the rdg model), Gx, Gy, Gz (components of the thermal
gradient at the location (x,y,z)).
Note that the coordinate system is the system defined in the mesh (see the Figure
after the next).
The file prefix_hcs-array.csv contains the same data, but this time stored in arrays
as indicated in the next Figure. The five quantities, z-tcg, rdg, Gx, Gy and Gz are
given as a function of x and y. This allows the user to get 3D graphics of each
array following the procedure described in the following Figures.
Viewing the arrays in the file prefix_hcs-array.csv with the help of Excel.
Clear the number nx*ny (60 here) before getting a 3D graphs of the array.
REFERENCES
[1] M. Rappaz, J.-M. Drezet and M. Gremaud : "A new hot tearing criterion",
Met. Trans. Vol. 30A, pp-449-455, Feb. 1999.
[2] J.-M. Drezet and M. Rappaz: "A New Hot Tearing Criterium for Aluminium
Alloys", proceedings of the First EsaForm Conference on Material Forming,
Ecole des Mines de Paris, CEMEF, Sophia Antipolis, France, March 1998, pp.
49-52.
[3] J.-M. Drezet, M. Gremaud, R. Graf and M. Gaümann, A new hot tearing
criterion for steel, proceedings of the 4th European Continuous Casting
Conference, IOM communications, Birmingham, UK, October 2002, pp. 755-763.
INPUT-OUTPUT FILES
This section is listing the different files which are created and used by the
different modules of ProCAST.
MeshCAST files
PREFIX.gmrst IO Binary Restart file for meshing
PREFIX.sm IO ASCII Surface mesh
PREFIX.stlsm O ASCII Surface mesh from STL file
PREFIX.igs I ASCII IGES file
PREFIX.iges I ASCII IGES file
PREFIX.stp I ASCII STEP file
PREFIX.step I ASCII STEP file
PREFIX.xmt_txt I ASCII Parasolid file
PREFIX.x_t I ASCII Parasolid file
PREFIXd.dat I ASCII ProCAST restart file
PREFIX_pre_sh.sm I ASCII Shell meshing recovery file
PREFIX.patran I ASCII Patran file (volume mesh)
PREFIX.out IO ASCII Patran file (surface mesh)
PREFIX.ideas I ASCII I-DEAS file (volume mesh)
PREFIX.unv I ASCII I-DEAS file (surface mesh)
PREFIX.nastran I ASCII Nastran file (volume mesh)
PREFIX.ansys I ASCII ANSYS file (volume mesh)
PREFIX.stl I ASCII STL ASCII file
PREFIX.bstl I ASCII STL Binary file
PREFIX.mesh IO ASCII Volume mesh
PREFIX_sub.sm O ASCII Surface mesh extracted from volume mesh
PREFIX_sub_act.sm O ASCII An active subset of original surface mesh
PREFIX.elem O ASCII Ansys element file
PREFIX.node O ASCII Ansys nodal file
PREFIX.wrk O ASCII Temporary working file for surface mesh merge
PREFIX.ceg O ASCII Temporary working file for surface mesh merge
usr_cmnds IO ASCII User defined icons
*.gif GIF capture
mesh.print
PREFIX.psm
PreCAST files
PREFIXd.dat IO ASCII General model setup (including the geometry),
also used for Restarts
PREFIXp.dat IO ASCII Run Paramerters file, also used for Restarts
PREFIX.ini IO Binary Extracted initial conditions
PREFIX.enc IO Binary Element number correspondance file in case of
Add/Delete Material
PREFIX.nnc IO Binary Node number correspondance file in case of
Add/Delete Material
PREFIXg0.dat IO ASCII Cut-off values per domain for Thixo casting
bc.db I ASCII Boundary conditions database
matl.db I ASCII Material properties database
stress.db I ASCII Stress properties database
intf.db I ASCII Interface heat transfer database
proc.db I ASCII Process database
Miscellaneous
(Some files are created for internal purposes)
PREFIX.cm
PREFIX.mtx
bem.out
PREFIXc.out
debug.out
dbg.out
mem_usage
poros.out
PREFIX.rhs
PREFIX.lhs
PREFIX.sol
PREFIXrf.out
PREFIX.rhs_v
sconv
PREFIXf.out
PREFIX.phs
PREFIX_rho.dat
PREFIX_k.dat
PREFIX_mu.dat
PREFIX_tdd.dat
PREFIX.err
PREFIXt.out
INLET VELOCITY
When a velocity Boundary condition is set in order to define an inlet of liquid
metal, the question of the value of the velocity and the value of the inlet diameter
of the liquid jet is important.
The following graph is showing the free fall velocity after a given distance of free
fall (a value of gravity of 10 m/s2 was used for this calculation).
One can see on the above graph that after 0.5 mm, the falling velocity is already
10 cm/s. After 1.25 cm, it is 0.5 m/s and after a falling height of 10 cm, the falling
velocity is 1.4 m/s. After 20 cm, the falling velocity is 2 m/s.
In usual casting conditions, the ladle (or crucible) is very often at a minimum of
10 or 20 cm above the top of the pouring cup. Thus, physically, it is advised to set
an inlet velocity (at the limit of the model) between 1 and 2 m/s and not smaller.
As a consequence, the inlet diameter should be set accordingly in order to have
the right inflow of metal.
If a too small inlet velocity is defined (e.g. 10 cm/s), the inlet diameter should be
too large and due to the strong acceleration on the first centimeters, the diameter
of the accelerating liquid jet will decrease very much. This situation is quite
difficult to be modeled and should be avoided.
Finally, one should always be careful that the mesh size below a falling liquid
stream should be fine enough in order to have 2-3 elements through the thickness
of the liquid jet.
When an "inlet BC" is defined, one should always be careful to set the right inlet
surface so that the corresponding inlet velocity is not too small (i.e. not smaller
than 1 m/s). To do so, the resulting inlet velocity should be checked at step 0 of
the simulation. If this velocity is too small, one should reduce the inlet surface.
LPDC RECOMMENDATIONS
The design of the ingate in LPDC is especially important in order to have good
filling results.
Although the real final casting has not anymore an ingate cylinder (see 1
hereafter), it is necessary to model an ingate cylinder which is long enough (see 2
hereafter). If this is not done, the pressure solution is not stable and the results will
not be good. There is not an absolute rule about the length of this ingate cylinder,
but it is advised to add between 6 to 10 cm.
Concerning the inlet boundary conditions, a pressure curve should be set at the
bottom of the ingate cylinder (see figure below).
Then, the pressure curve which should be applied must be a pressure ramp,
starting with a zero pressure (it is supposed that PREF is set to 0 bar) and going up
to the metallostatic pressure corresponding to the height h (see the figure above).
Thus, the maximum pressure (see 1 in the figure below) should be equal to the
metallostatic pressure of the height h (Pressure = density * g * h), at the time (see
2 in the figure below) corresponding to the filling time. The shape of the curve
does not need to be a straight line as shown in the graph below.
Please note that if the bottom of the ingate cylinder does not correspond to the
liquid surface in the furnace, the pressure ramp which is measured in the real
process should be shifted to correspond to the one sketched above.
All the remaining liquid (up to the fraction of solid defined by DRAINFS) which
is connected to the inlet pressure boundary condition location will be removed at
the time specified by the value of DRAINTIME.
The following figure is showing the results of two different calculations with
different values of DRAINTIME. As you can see the temperature in the casting
and in the mold (taken at the same time) are different, as the "heating effect "of
the liquid in the inlet tube is different in both cases.
The following figures are showing two examples with different values of
DRAINTIME. One can well see the effect of the DRAINTIME value on the final
shape of the wheel near the inlet tube. In the 20 s. example, it is obvious that the
draining was performed too early.
In the draining algorithm, all the liquid which is between the injection point
(where the inlet pressure BC is applied) and the DRAINFS isoline will be
removed at the time of DRAINTIME, regardless of the geometry of the casting. In
the following example, the draining being applied very early in the solidification
process, almost all the cavity is drained (see next figure).
However, in a normal situation, the volume below the dashed line (see next
figure) should remain in the mold and should not be drained. This situation is not
taken into account in the algorithm as it is normally not intended in the reality to
have such situation.
The draining algorithm is also applicable in the case of cycling calculations. In the
following simple example, one can see very well that due to the heating up of the
die during the first cycles, the shape of the casting at the drained location is
changing.
RESTART
What should we do when we would like to restart a calculation after some data
modifications ?
• If the content of the d.dat file was not modified, one should not run
DataCAST and it is possible to run directly ProCAST. Typically,
this happens if the content of the p.dat file only was modified. If
PreCAST is loaded, the d.dat file is not modified only if the Run
Parameters are changed. If anything else is changed in PreCAST, it
is necessary to run first DataCAST and then ProCAST.
• When the d.dat file is modified, DataCAST has first to be run and
then ProCAST should be launched.
CONVERGENCE PROBLEMS
If a calculation does not converge or has problem to converge (this can be seen in
the p.out file - please refer to the Troubleshooting section for more details), the
following operations can be tried :
STRESS CALCULATIONS
• Stress calculations might be rather long (in terms of CPU time), as the six
components of the stress tensor have to be computed, in addition to
Temperature, Pressure and the three components of the velocity.
• Stress calculations are starting as soon as the fraction of solid is larger than a
critical fraction defined by the Run Parameter CRITFS. By default, CRITFS =
0.5.
The model which is used assumes that radiation and heat conduction through an
air gap is progressively replacing the transfer due to the contact (see the model in
the figure below). If VACUUM is set to 1, the heat conduction part is deactivated.
When there is a contact, the heat transfer coefficient is increased as a function of
the pressure.
The figure below illustrates the change of the resulting interface heat transfer
coefficient, as a function of the air gap width. Please note that the air thermal
conductivity corresponds to a temperature of about 800°C.
By default, the gap is filled with air (with the above properties) or with Vacuum.
However, in special cases, the user has the possibility to define other properties
for the gap (e.g. flux continuous casting of steel, or water in a bottom block of DC
casting of Al).
To do so, the "interhtransfer.c" user function should be used. When there is a gap
(and the interhtransfer.c function is activated), the value returned by the function
does correspond to h0 in the above equation (thus, if there is no gap, h = h0).
Moreover, the values of k (the "gap conductivity") and of the Stefan-Bolzmann
constant (which is used to calculate the equivalent radiative heat transfer
coefficient hrad) can be changed within the interhtransfer.c user function. To
modify k, one should assign the desired value to the "gap_cond" variable and for
the Stefan-Bolzmann constant, one should assign the "stefan" variable (see
function hereafter).
By this means, it is possible to specify in the user function a gap equivalent heat
transfer coefficient as a function of the location for instance. For instance, above a
given height, there is some flux, thus, the value of k can be set to a "high" value
corresponding to the flux and the Stefan-Bolzmann constant can be set to zero to
cancel the radiation. Below a given height, we can model the effect of the water
by setting a very large value of h0 (the values of the conductivity and
Stefan-Bolzmann will be negligible with respect to water cooling heat transfer
#include <stdio.h>
#include <stdlib.h>
#include "common_dll.h" /* please note that this line was added in
2009.0 */
/* version to handle these gap definitions */
#ifdef WIN
#define EXPORT _declspec(dllexport)
EXPORT real func_interhtransfer(char*, int, real, real, real, real,
real, real, int);
#else
real func_interhtransfer(char*, int, real, real, real, real, real, real,
int);
#endif
/*
* interface heat transfer coefficient
*/
real func_interhtransfer(
char prefix[], /* case name */
int dimension, /* 2 = 2D ; 3 = 3D */
real temp, /* current temperature */
real fs, /* current fraction of solid */
real time, /* current time */
real x_coor, /* local coordinates: x */
real y_coor, /* local coordinates: y */
real z_coor, /* local coordinates: z */
int numBC) /* number of boundary condition */
{
/* ------------- Do not change anything above this line -------------
*
* ------------- Program your function below this line -------------
*/
real h0;
real hmax;
real hmin;
else
{
h0 = 500. ; /* default value for h0 */
/* default values for air gap , nothing has to be specified */
/* default value : stefan = 1.3541845E-12 (this value should be
always specified in these CGS units) */
/* default value (air) : gap_cond = 0.079 W/m2K (the units should
correspond to the "units.dat" ones) */
}
return h0;
/* only h0 is returned, however, the values of gap_cond and stefan
defined above are also returned */
/* and will be used for this location upon exit of the function and
for the next location, it will be */
/* changed at the next call of the function. */
}
STRESS VISUALIZATION
Effective Stress
There are several ways to evaluate the local stress state with a single value. The
formula below shows how the Effective Stress in ViewCAST is calculated (e.g. in
one dimension, the effective stress corresponds to the stress itself). To simplify,
one could consider that it corresponds to some kind of "average" value of the
stress. As shown in the equation below, the Effective Stress is always positive and
thus, it does not allow to determine whether the state of stress is in compression or
in tension. To determine that, the other viewing possibilities should be used.
The stress components which are in the square root correspond to the principal
stresses.
The Average Normal Stress corresponds to the average of the three normal
components. This value can be either negative (mainly compressive state) of
positive (mainly tensile state).
The shear stresses components are highlighted in red in the figure above. The
"maximum shear stress" is defined by the equation below (where the stress
components corresponds to the maximum and the minimum normal stresses).
Principal Stresses
Considering the stress state of a given unit volume, it is possible to find a unit
cube with a different orientation, where all the shear stresses are zero. Thus, only
the normal stresses are remaining and they are called principal stresses. The
principal stress 1 is the larger value and the principal stress 3 is the smaller value.
Positive values show a tensile state, whereas negative values show a compressive
state.
RESULTS COMPARISONS
BETWEEN VERSIONS
The goal of this section is to present results of a set of examples which have been
calculated with the current version, with comparisons with the previous one.
Comparisons between scalar and parallel (DMP) calculations are also presented.
One should note that in some cases (especially in filling), the physical solution
can be somehow "unstable". As a consequence, the results of the calculations
made with slightly different conditions can exhibit some noticeable differences.
For instance, as the solvers are different between the scalar and the DMP versions,
the convergences are different and thus the timesteps may be different. This can
affect for instance the free surface calculation and thus the results may differ.
However, when such behavior will happen in the calculations, it will also happen
in reality and vice-versa.
Please note that these cases were calculated with the Default Pre-defined Run
parameters.
Gravity-1
Comparisons of the filling sequences at different times of a gravity test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Gravity-2
Comparisons of the filling sequences at different times of a gravity test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Gravity-3
Comparisons of the filling sequences at different times of a gravity test casting,
including a filter. Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Filling time
HPDC-1
Comparisons of the filling sequences at different times of a HPDC test casting.
Fluid velocity magnitude and flow front are displayed.
LPDC-1
Comparisons of the filling sequences at different times of a LPDC test casting.
Temperature and flow front, filling time and shrinkage porosity are displayed.
Temperature
Temperature
Temperature
Filling time
Shrinkage Porosity
Investment-1
Comparisons of the cooling and solidification sequences at different times of an
investment test casting. Temperature are displayed.
Temperature
Temperature
Temperature
Stress-1
Comparisons of the thermal, porosity and stresses of an LPDC test casting.
Temperature, shrinkage porosity, Effective stress, Average Normal Stress and
Principal stress 1 are displayed.
Temperature
Temperature
Shrinkage Porosity
Effective stress
Principal stress 1
Microstructure-1
Comparisons of the cooling, solidification, microstructure, porosity and density of
a cast iron test casting. Temperature, nodule count, shrinkage porosity and
densities (as a function of temperature at different locations) are displayed.
Please note that in this case, as the algorithm for Fading in the presence of
graphite expansion was significantly improved in version 2007.0, the comparison
with 2006.1 is not relevant.
Temperature
Nodule count
Shrinkage porosity
Thixo-1
Comparisons of the filling sequences at different times of a thixo test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Centrifugal-1
Comparisons of the filling sequences at different times of a centrifugal test
casting. Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Temperature
Please note that these cases were calculated with the Default Pre-defined Run
parameters.
Gravity-1
Comparisons of the filling sequences at different times of a gravity test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Gravity-2
Comparisons of the filling sequences at different times of a gravity test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Gravity-3
Comparisons of the filling sequences at different times of a gravity test casting,
including a filter. Temperature and flow front are displayed.
Temperature
Temperature
Filling time
HPDC-1
Comparisons of the filling sequences at different times of a HPDC test casting.
Fluid velocity magnitude and flow front are displayed.
LPDC-1
Comparisons of the filling sequences at different times of a LPDC test casting.
Temperature and flow front, filling time and shrinkage porosity are displayed.
Temperature
Temperature
Temperature
Filling time
Shrinkage Porosity
Investment-1
Comparisons of the cooling and solidification sequences at different times of an
investment test casting. Temperature are displayed.
Temperature
Temperature
Temperature
Stress-1
Comparisons of the thermal, porosity and stresses of an LPDC test casting.
Temperature, shrinkage porosity, Effective stress, Average Normal Stress and
Principal stress 1 are displayed.
Temperature
Temperature
Shrinkage Porosity
Effective stress
Principal stress 1
Microstructure-1
Comparisons of the cooling, solidification, microstructure, porosity and density of
a cast iron test casting. Temperature, nodule count, shrinkage porosity and
densities (as a function of temperature at different locations) are displayed.
Please note that in this case, as the algorithm for Fading in the presence of
graphite expansion was significantly improved in version 2007.0, the comparison
with 2006.1 is not relevant.
Temperature
Nodule count
Shrinkage porosity
Thixo-1
Comparisons of the filling sequences at different times of a thixo test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Centrifugal-1
Comparisons of the filling sequences at different times of a centrifugal test
casting. Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Temperature
Please note that these cases were calculated with the Default Pre-defined Run
parameters.
Gravity-1
Comparisons of the filling sequences at different times of a gravity test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Temperature
Gravity-2
Comparisons of the filling sequences at different times of a gravity test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Gravity-3
Comparisons of the filling sequences at different times of a gravity test casting,
including a filter. Temperature and flow front are displayed.
Temperature
Temperature
Temperature
HPDC-1
Comparisons of the filling sequences at different times of a HPDC test casting.
Fluid velocity magnitude and flow front are displayed.
LPDC-1
Comparisons of the filling sequences at different times of a LPDC test casting.
Temperature and flow front, filling time and shrinkage porosity are displayed.
Temperature
Temperature
Filling time
Shrinkage Porosity
Investment-1
Comparisons of the cooling and solidification sequences at different times of an
investment test casting. Temperature are displayed.
Temperature
Temperature
Temperature
Stress-1
Comparisons of the thermal, porosity and stresses of an LPDC test casting.
Temperature, shrinkage porosity, Effective stress, Average Normal Stress and
Principal stress 1 are displayed.
Temperature
Temperature
Shrinkage Porosity
Effective stress
Principal stress 1
Microstructure-1
Comparisons of the cooling, solidification, microstructure, porosity and density of
a cast iron test casting. Temperature, nodule count, shrinkage porosity and
densities (as a function of temperature at different locations) are displayed.
Please note that in this case, as the algorithm for Fading in the presence of
graphite expansion was significantly improved in version 2007.0, the comparison
with 2006.1 is not relevant.
Temperature
Nodule count
Shrinkage porosity
Thixo-1
Comparisons of the filling sequences at different times of a thixo test casting.
Temperature and flow front are displayed.
Temperature
Temperature
Centrifugal-1
Comparisons of the filling sequences at different times of a centrifugal test
casting. Temperature and flow front are displayed.
Temperature
Temperature
Temperature
TUTORIALS
The Tutorials are in a separate document.