Some debug variables.

In order to help the programming, we provide some debug variables. It is recommended to use them for programming but not to deliver them with your applicative product. Also, do not use them during performances tests.

• export  TRACE_EXPLICIT=1 will generate in the console the stack of the error messages generated by the field model API. This can be used for development phase.
• export  TRACE_EXPLICIT=THROW will generate in the console the stack of the error messages generated by the field model API and produce a crash of the application. This can be used for tests cases replay in order to detect potential regressions.

What can I do with a CATSamExplicitPtr?

When programming with the CATAnalysisExplicitObjects,  we work with Object and no pointers. In order to manage several objects, the ExplicitList or ExplicitListUsr can be used. If an ExplicitCharac may receive as values another object, the SetValue/GetValue methods use CATSamExplicitPtr. Any Field model object derive from the CATAnalysisExplicitData Class. To get the pointer use the GetPointer method, to build and object use LinkWithPointer method. 

...

Here is the sample of code I use to obtain the CATSamExplicitPtr :

//-----------------------------------------------------------
CATSamExplicitPtr AxisPtr = NULL;
CATAnalysisExplicitAxis Axis;

CATAnalysisExplicitChild Child = *(ChildrenList.First());
while(Child!=NULL_exp)
{
  if (Child.GetType() == CATSamDataTypeCharac) 
  {
    CATAnalysisExplicitCharac CharacAxis = Child ;
    CATSamValue ValueType;
    hr = CharacAxis.GetValueType(ValueType);
    if (SUCCEDDED(hr) && ValueType == CATSamValueAxis)
    {
      hr = CharacAxis.GetValues(AxisPtr);
      if (SUCCEDDED(hr)) { Axis.LinkWithPointer (AxisPtr,pExplicitModel); break}
    }
  }
  Child = *(ChildrenList.Next());
}
...
Note that the pointer on the CATAnalysisExplicitModel, if you do not have it from previous lines is retrieved with:
CATAnalysisExplicitModel *pExplicitModel = ChildrenList.GetModel();

[Top]

How to store values for a tensor in a CATAnalysisCharacCollector ?

The chapter describes also the way values must be stored in CATAnalysisExplicitCharac object.

As long as your tensor dimension is lesser or equal to three, the figure shows the way the values have to be put in the values of the CATAnalysisCharacCollector or the CATAnalysisExplicitCharac. The same order needs to  be used for Principal or symmetrical tensors.
For tensor with highest dimension, see the associated figure that illustrates a 6x6 tensor. Three cases are illustrated here: Full tensor Symmetric tensor Principal (Eigen Value) tensor.

[Top]

I don't find SAMHeaderGenerateImage anymore !

Before R11 this header was exported by the "SAMWorkshop", so you just had to insert it in your toolbar. It is no more the case. Now, you have to create your own header for accessing the generic command as follow:

 new MyCommandHeader ("xxxxxxxxxxxx","CATAV0UI" ,"CATSPMGenerateImageCmd",(void *) NULL );

 Some changes allows now to be completely independent of DASSAULT SYSTEMES libraries and commands names. For this, refer to the associated reference.

[Top]

How to work with CATSamDimension Object.

This functionality impacts also the collect of the data that use the CATAnalysisCharacCollector object. The impacts are located in the CATSamDimention class. This object have to manage the "repeat" information not only per position and SMI (as before Release 15), but also per Position, per Lamina and per SMI. It means, for one physical type of characteristics, that you can retrieve several values for each combining of Position, Lamina and SMI.

New preprocessing capabilities.

Before V5 Release 11, it was possible to define in the field model a physical characteristic on an Finite Element Face. But, for thermal loading applied on shell elements (for example), it makes sense to define such a temperature at on three intermediate positions: Superior face, Middle of the thickness and Inferior face. If not specified, the solver may assume that the temperature is constant inside the element.

To support this at the level of the processing data, the ApplyTo methods of the CATAnalysisExplicitEntity have been completed with some arguments.

The optional CATSamApplyQualifier enumerate allow to customize the apply type with specific behavior. By using it your are able to specify a loading on the same face number of an element with two sub positions. The enumerate is also used to define the number of the position on which the data are applied.

[Top]

Impact on CATSamDimension Class usage.

This functionality impacts also the collect of the data that use the CATAnalysisCharacCollector object. The impact are located in the CATSamDimention class. This object have to manage the "repeat" information not only per position and SMI (as before Release 15), but also per Position, per Lamina and per SMI. It means, for one physical type of characteristics, that you can retrieve several values for each combining of Position, Lamina and SMI.

Building a CATSamDimension

Let's take an example:

To create the associated dimension, when implementing the CATICharacCollector interface, the data of the dimension can be setup at the level of the constructor or later with the fill method. For the example:

• NbPosition = 3
• Position[NbPosition] = {0, 1, 3} where Position[i] is the node number of the element.
• NbLamina = 1
• Lamina[NbLamina] = {0} where Lamina[i] is the layer number of the element
• NbSMI = 2
• SMI[NbSMI] = {CATSamApplyQualifierUpper, CATSamApplyQualifierLower} where SMI[i] is the "SMI" Identifier of the element (see CATSamApplyQualifier )
• NbRepeat [NbPosition*NbLamina*NbSMI] = {R0, R1, R2, R3, R4, R5}
• Repeat [S(Ri)] = {l,m,...} where l, m, ... is the sequential number of the entity (load entity) that provide data or -1 if unknown (The Repeat can also be NULL)
• AggregationMode [S(Ri)] = {l,m,...} where l, m, ... describe the way the values are combined or CATSamAggregationMode if unknown (AggregationMode can also be NULL).

This dimension is now ready to be stored in a Charac-Collector.

[Top]

Getting access to a CATSamDimension

Two ways can be noted to extract the data.

1. The goal is to extract all the values without any knowledge about the way they are applied. In this case, you just need to loop on the total number of values get by CATSamDimension::GetNbValues() for a DimensionPointer[iEntite]. The values are in the memory allocation pointed by ValuesPointer[iEntite)
2. The goal is to extract values for a list of position, lamina, SMI....  It is recommended to use the CATSamDimension::GetValuesAdress() for using DimensionPointer[iEntite]. By Filling the memory allocation pointed by ValuesPointer[iEntite] and the index required of Position, Lamina and SMI location. This method will return the pointer in the memory corresponding to the index of Position, Lamina and SMI location. This method allow to scan all the values with the following sequence of loops:

...
for iLam = 0 -> DimensionPointeur->GetNbLamina() -1
  for iSMI = 0 -> DimensionPointeur->GetNbSMI() -1
    for iPos = 0 -> DimensionPointeur->GetNbPosition() -1
      for iRepeat = 0 -> DimensionPointeur->GetNbRepeat(iPos, iLam , iSMI) Note that Lamina as index is added in R15 Release.
        ValPtr = DimensionPointeur -> ValuesAdress(ValuesPointeur, iPos, iLam, iSMI, iRepeat) 
        FOR iComp = 0 -> DimensionPointeur->GetNbComponents() -1 
           cout <<" my value = " << ValPtr[i];
...

This way of programming is the only stable method because it gives a guaranty to access effectively all the values indexes, whatever the way the data are defined for a charac-collector.

Note that some methods supported until Release 10 are now deprecated and replaced by new ones. They correspond to all methods that were not able to support variable SMI. These methods will remain for two versions, they are programmed by using the new ones with the SMI index set to zero (the first one). The pointer of access is valid but only one SMI information is seen. It is mandatory to migrate on the new set of methods.

[Top]

CATAnalysisExplicitTopology usage.

In order to correct some bugs that may produce crash and extra-memory consumption.

The CATAnalysisExplicitTopology object is the central object that will allow to access all the information's related to the meshing. Before V5 Release 13, it was possible to use the CATAnalysisExplicitTopology constructor. When using this constructor, all the internal data were never released. Also, any changes that may happen on the mesh were not propagated in this new version of the topology.

To avoid those problem, the constructor is now private. To get access to the CATAnalysisExplicitTopology use CATAnalysisExplicitModel::GetExplicitTopology. Preparing new enhancements related to assembly of analysis, the Explicit Topology supports now an AddRef/Release mechanism. So the CATAnalysisExplicitModel::GetExplicitTopology increase the life cycle counter of this object and you have to release it. For assembly of the topologies, some internal data members are shared (for memory consumption reduction), for this an Addref /Release is mandatory.

[Top]

Field Model "MemberList" usage.

A new implementation of the Field model has been delivered in V5R14. Objectives of this action was memory consumption and performances. The previous implementation was based on list of chained objects by physical types. Those list were visible at programming level by using GetMemberList and FindData methods. Because of the new implementation, the lists do not exist anymore. Because of CAA compatibility the API is deprecated for V5R14 but the lists are now allocated on demand by scanning the complete field model. So performances of creation and search will be slower. Try to remove those API from your programming. Field model objects should be retrieved from their parent that is associated by a feature (for this use CATISamExplicit::GetExplicitObjects) and use the CATAnalysisExplicitParent::GetChildrenList to scan the associated children's.

In any case, the methods defined on  the CATAnalysisExplicitModel class will be removed 2 releases after V5R14.

[Top]

Meshing tasks (Update, Remove ....).

Some changes as been introduced in release R14 concerning meshing algorithms. This can impact, from a programming point of view,  the Update and Remove tasks or "Empty Mesh" contextual menu. The following schema will describe the way it's work. It describe the different steps and where you can introduce some applicative software.

Update In the new additional call of CATISamExplicitation, you can add some applicative software. Take care, that now an explicitation of an empty meshpart is no more an error, is this case you have to Create the set with physical type MESHSET, out it in the oNewExplicitList and call the UpdateExplicitList method.
MeshPart Life Cycle In the new additional call of CATISamAnalysisSetInit, you can add some applicative software.
"Empty Mesh" In the new additional call of CATISamExplicitation, you can add some applicative software.

Meshing Model programming rules.

In order to avoid any mismatch between finite element modifications (Elements creation or deletion) and the corresponding features modifications (Mesh Parts and Mesh Operators) we are going to put in place more severe controls when performing some operations. This rule are activated in V5R15.

This rule will impact the following tasks:

• Create an element without having a current domain before. The current domain is the last created or the last which has been set as current. A domain is created using: CATIMSHMesh::CreateMeshDomain (CATIMSHMeshPart * MeshPart) A domain is set as current using: CATIMSHMesh::SetCurrentMeshDomain (CATIMSHMeshDomain * Domain).
• Create an element in a Mesh Part which is not in edition. A Mesh Part is considered as being in edition in the two following situations:
  • During the execution of the Mesh method of its associated mesher: XXXMesher::Mesh (CATIMSHMeshPart * MeshPart).
  • After execution of the OpenEdition method: CATIMSHMeshPart::OpenEdition ()
• Delete an element in a Mesh Part which is not in edition. A Mesh Part is considered as being in edition in the two following situations:
  • During the execution of the Mesh method of its associated mesher: XXXMesher::Mesh (CATIMSHMeshPart * MeshPart).
  • After execution of the OpenEdition method: CATIMSHMeshPart::OpenEdition ().
• Load a document containing elements which are not belonging to any Mesh Part. This will happen in case of a document stored without elements created without a current domain.

As summary, the way of creating, deleting and modifying meshing entities must be limited into:

• Implementaion of CATIMSHMesher interface.
• Implementation of a CATIMSHOperator interface.

[Top]

CATAnalysisExplicitCharac Class.

This functionality completes the enhancement related to the dimension class started in V5R11. It impacts also the collect of the data that use the CATAnalysisCharacCollector object. The objective is to provide a generic way to describe that the values of a characteristic are variable. In fact, they can be variable inside an element (for example, on lamina or predefined position) or they can be variable per element. Before release R15, for the second point, you can use a negative repeat to have values per elements. We have enhanced the SetDefinition method of CATAnalysisExplicitCharac . See the appropriate enumerate: CATSamValuesDistributionMode. In order to apply at specific position inside a finite element, you can use:

• Predefined enumerate to set up different value at the node of the element: CATSamValuesDistributionModeRepeatedOnVertexOfElements CATSamValuesDistributionModeDistributedOnVertexOfElements
• The couple of characteristics "NB_POSITIONS" and "POSITIONS" associated to CATSamValuesDistributionModeRepeatedOnPositions  or CATSamValuesDistributionModeDistributedOnPositions. Those physical types have specific behavior for the collect. The values of position characteristic are collected in the UVW of the CATAnalysisCharacCollector, see GetPositionUVWCoordinates method. NB_POSITIONS is integer type and gives the dimension of the POSITIONS characteristics, defined generally as float or double.

Note that a negative repeat is supported for 2 releases after V5R15 but must be replaced as soon as possible by a positive repeat and  a distribution mode set to "CATSamValuesDistributionModeDistributed". This means that values are distributed per apply.

[Top]

Mechanism for backup startup.

Suppose you have an analysis document with instances of CAA analysis feature StartUp, and you send this document  to a supplier. In most cases, your suppliers do not have the code (dll) and StartUp's catalog which would enable them to find the behaviors that you defined on your CAA features. Your document is quiet useless in this context.

The proposed (called backup StartUp) allows to provide a minimum of behaviors when the CAA feature is used without its CATfct file. It informs you that installation is partial and some operation like edition or update are secured.

• The supported startup for this are : MSHPart, AnalysisSet and AnalysisEntity.
• If you need to check if association are missing in an existing catalog use CATSamCatalogFactory::UpgradeForBackup. And input argument allows to check or check an repair and existing catalog.
• An automatic association to the backup is performed when using the CATSamCatalogFactory class.
• All the new instances associated to existing startup will support this mechanism. For pre-existing instances, some tools will be provided in a next future.

[Top]

History

Copyright © 2002, Dassault Systèmes. All rights reserved.