massimopetracca / opensees Goto Github PK

• add setParameter to ShellMITC4, forward it to all sections

int ShellMITC4::setParameter(const char ** argv, int argc, Parameter & param)
{
	// this element does not have specific parameters
	// just forward it to all sections.
	// therfore we don't need to implement updateParameter
	int res = -1;
	int matRes = res;
	for (int i = 0; i < 4; i++) {
		matRes = materialPointers[i]->setParameter(argv, argc, param);
		if (matRes != -1)
			res = matRes;
	}
	return res;
}

Hello Dr. Petracca,

It seems the MPCO recorder result file does not support relative path that directs to upper level folders. For example, file path "../model/res/mpcoRes.h5" would lead to “/model/res/mpcoRes.h5.mpco” which is not intended (due to the split function on line 4596 in the MPCORecorder.cpp file). This is not big issue and just noted for your info.

Regards,
Junfei

material wrappers are missing updateParameter(useful??) and setResponse:

• PlaneStrainMaterial
• PlaneStressMaterial
• PlateFiberMaterial
• PlateFromPlaneStressMaterial
• PlateRebarMaterial

int PlaneStrainMaterial::updateParameter(int responseID, Information & eleInformation)
{
	return theMaterial->updateParameter(responseID, eleInformation);
}

Response * PlaneStrainMaterial::setResponse(const char ** argv, int argc, OPS_Stream & s)
{
	/** massimo petracca - 12/12/2019. call base class for those results known by the base class,
	otherwise call the material method
	*/
	if (strcmp(argv[0], "stress") == 0 ||
		strcmp(argv[0], "stresses") == 0 ||
		strcmp(argv[0], "strain") == 0 || 
		strcmp(argv[0], "strains") == 0 ||
		strcmp(argv[0], "TempAndElong") == 0 || 
		strcmp(argv[0], "TempAndElong") == 0 ||
		strcmp(argv[0], "Tangent") == 0 || 
		strcmp(argv[0], "tangent") == 0
		) {
		return NDMaterial::setResponse(argv, argc, s);
	}

	return theMaterial->setResponse(argv, argc, s);
}

some elements (trussSection corotTrussSection, zeroLengthSection) do not give correct output, and the MPCORecorder cannot guess the number of fibers

see the zeroLengthSection in our old code as an example

It looks like those shell elements invert the sign of curvature to give to the fibers in layered cross section

ShellMITC4 has convergence issues, probably due to the fact that it does not implement the update method.
Moreover, in linear elastic analysis it does not compute results, also for the same reason.

• MPCORecorder in our version has some minor changes, merge them.
• Update with new Elements

merge our own modifications in MUMPS_xxx files and tcl/commands.cpp