deal with non-prismatic beam in xc

[ebrahimraeyat]

Hi All. I wanted to work with xc and for starting i want to model an industrial structure with XC, my question:

can I model non-prismatic beam in xc. I saw this discution that seems it is not directly possible in opensees:

How_to_deal_with_non-prismatic_beam_in_Opensees

I have been modeled it in ETABS v19 and wanted to model it with xc and compare both of them:

[lcpt]

Hi @ebrahimraeyat

You can model the tapered beams discretizing its length (that's certainly the solution that ETABS uses). All the solutions proposed in your ResearchGate link seem adequate. On the other hand, if you need a very precise solution (for example in the context of a PhD dissertation, it's better to use shell or brick elements).

[ebrahimraeyat]

Thanks, you mean we don't have an element for non-prismatic at all (in theoretical and thus in none of FEM software). Thanks.
I want it for my engineering purpose.

[lcpt]

Well, as Mr. Balduzzi says, there are more accurate elements like the one in this article. I you want to implement those elements in XC (or OpenSees of both), I'll be happy to help.

On the other hand, for practical purposes, I would use the discretization mentioned before for the model of the whole structure, and make a more precise analysis when designing the connections by means of a CBFEM model.

[ebrahimraeyat]

Well, as Mr. Balduzzi says, there are more accurate elements like the one in this article. I you want to implement those elements in XC (or OpenSees of both), I'll be happy to help.

Unfortunately I don't have work experience with c++, but if I can develop it with python I will be happy to apply it in XC

[ebrahimraeyat]

Is it possible to share some of these models?

[lcpt]

The code that generates those models is not ready to be published yet. If you want to analyze how it works I can share the code privately. Ask me using l.pereztato@ciccp.es.

[ebrahimraeyat]

Hi @lcpt, How can I model general Steel I/Wide Flange sections, I mean without manually calculate the properties of sections like A, Iz or Ix?
thanks

[lcpt]

Hi @ebrahimraeyat

You don't need to manually calculate the section properties you can use the steel shape database (at present AISC and Arcelor). You can also compute automatically the mass properties of any polygon using the geom module. Furthermore, you can also model the steel shape using fiber sections if you want to reproduce the nonlinear behavior of the material.

[ebrahimraeyat]

Thanks @lcpt. I don't want to model standard I/Wide Flange. I wanted to model general, I mean whatever BF, TF, H and TW. is it possible?

[lcpt]

Why it will not be possible? The solutions I mentioned earlier are completely general.

[ebrahimraeyat]

Thanks, @lcpt. I don't want to perform a nonlinear analyzing, can you write it on one line? For example this is in examples:

shape= ASTM_materials.WShape(steel,'W18X50')

but I wanted to set (BF, TF, D, TW), not general polygon:

shape= ASTM_materials.WShape(steel,(20, 12, 30, 8)) # (BF, TF, D, TW)

[ebrahimraeyat]

Excuse me, I found it on materials/sections/section_properties.py. thanks.

from materials.sections import section_properties as sp
wdTopFlange = wdBotFlange = 20
thTopFlange = thBotFlange = 1.2
thWeb = .8
hgWeb = 15

isection = sp.ISection("isection",wdTopFlange,thTopFlange,thWeb,hgWeb,wdBotFlange,thBotFlange)

[ebrahimraeyat]

I created two functions for creating lines and section for nonprismatic section:

import xc_base
import geom
import xc

from materials.sections import section_properties
from materials import typical_materials as tm

creating lines giving 2 points

def create_lines(
    p1,
    p2,
    n: int,
    points,
    lines,
    point_id,
    )-> list:
    x1, y1, z1 = p1.getPos.x, p1.getPos.y, p1.getPos.z
    x2, y2, z2 = p2.getPos.x, p2.getPos.y, p2.getPos.z
    dx = (x2 - x1) / n
    dy = (y2 - y1) / n
    dz = (z2 - z1) / n
    
    first_point = p1
    new_lines = []
    for i in range(n):
        point_id += 1
        x = first_point.getPos.x + dx
        y = first_point.getPos.y + dy
        z = first_point.getPos.z + dz
        if all((x == x2, y == y2, z == z2)):
            last_point = p2
        else:
            last_point = points.newPntIDPos3d(point_id, geom.Pos3d(x, y, z))
        l = lines.newLine(first_point.tag, last_point.tag)
        new_lines.append(l)
        first_point = last_point
    return new_lines, point_id

creating sections, but I don't know how can I apply section on one line, not on Set:

def apply_nonPrismatic_section(sets, bf, tf, tw, hw_min, hw_max, \
                               coordTransf, modelSpace, material, prep):
    tag = 'isection'
    isection = section_properties.ISection(tag,bf,tf,tw,hw_min,bf,tf)
    linel_mat= tm.BeamMaterialData(name=tag, section=isection, material=material)
    linel_mat.setupElasticShear3DSection(preprocessor=prep)
    for Set in sets:
        tag = Set.getLines[0].tag
        # n = Set.lines.size
        # dhw = hw_max - hw_min
        # for line in Set.lines:
        #     z = line.getCentroid().z
        #     average_hw = hw_min + z / h_col * dhw
        #     isection = section_properties.ISection(f"isection{line.tag}",bf,tf,tw,average_hw,bf,tf)
        #     p1 = line.getP1().tag
        #     p2 = line.getP2().tag
        #     seedElemHandler.defaultMaterial = isection.sectionName
        #     seedElemHandler.newElement('elastic_beam3d', xc.ID([p1, p2]))
        modelSpace.createElasticBeams(Set, linel_mat, coordTransf, xc.Vector([0.0,1.0,0.0]), nDiv= 6)

creaitng lines:

col_frame_a, n = create_lines(pt0, pt3, 10, points,lines, n)
col_frame_c, n = create_lines(pt1, pt7, 10, points,lines, n)
col_frame_e, n = create_lines(pt2, pt11, 10, points,lines, n)

assign sections:

apply_nonPrismatic_section([col_frame_a_set, col_frame_c_set], bf_col_ae, \
                           tf_col_ae, tw_col_ae, hw_col_ae_bot, hw_col_ae_top, coordTransf, \
                           modelSpace, steel_W, preprocessor)

Thanks.

complete code:

sole_zeinali.zip

[lcpt]

Hi.

Materials (a section is a material) are assigned to elements when the element is created. You specify the material template to use when the genMesh() method is called using the defaultMaterialproperty of the seed element handler (see, for example, test_uls_checking_aisc_01.py). You can store the material name you'll like to use in the lines or anywhere else (you can use the setProp() and getProp() methods to store things in the C++ object that lies inside the Python wrapper provided by the boost.python interface).

The element material can also be modified after creation, but I don't use this very often.

The methods of the ModelSpace class are mostly convenience functions to make easier simple tasks. You can take a look inside to see how the task is performed when you need to do things that are not so simple.

[ebrahimraeyat]

Thanks, @lcpt. I wrote the function (assume developing line in z direction, sufficient for my case now!):

def apply_nonPrismatic_section(sets, bf, tf, tw, hw_min, hw_max, \
                               material, prep, element_handler):
    # dia_bounds = {
    #     'x': ['dia_bound.x', 'l_center.x', 'bnd.getZMin'],
    #     'y': ['dia_bound.y', 'l_center.y', 'bnd.getZMin'],
    #     'z': ['dia_bound.z', 'l_center.z', 'bnd.getZMin'],
    # }
    dhw = hw_max - hw_min
    for Set in sets:
        bnd = Set.lines.getBnd()
        dz = bnd.diagonal.z
        zmin = bnd.getZMin
        for line in Set.getLines:
            z_line = line.getCentroid().z
            percent = abs(z_line - zmin) / dz

            average_hw = hw_min + percent * dhw
            assert average_hw > 0
            # lines_len += l
            isection = section_properties.ISection(f"isection{line.tag}",bf,tf,tw,average_hw,bf,tf)
            xc_material= isection.defElasticShearSection3d(prep, material)
            element_handler.defaultMaterial = xc_material.name
            element_handler.newElement('ElasticBeam3d', xc.ID([0,0]))
            line.genMesh(xc.meshDir.I)
            for e in line.getElements:
                e.setProp('crossSection',isection)
        Set.fillDownwards()

But when I want to create a report, It gives me this error for elements that using ISection shape class:

DL Load combination: DL already exists, redefined. XC::ElasticBeam3d::getPyProp; Warning, property: 'ULS_normalStressesResistanceSect1' not found. Returning None. Traceback (most recent call last): File "/home/ebi/xc/my_models/sole_zeinali/xc_write_reports.py", line 66, in <module> report.checksReport(limitStateLabel,[],[],setsBmEl,argsBmEl) File "/usr/local/lib/python3.8/dist-packages/postprocess/reports/report_generator.py", line 69, in checksReport self.displayBeamResult(attributeName=limitStateLabel,itemToDisp=argS,beamSetDispRes=stV,setToDisplay=stV,caption=capt,fileName=fullgrFileNm+'.jpg') File "/usr/local/lib/python3.8/dist-packages/postprocess/output_handler.py", line 779, in displayBeamResult lstArgVal=[e.getProp(attributeName+'Sect1')(itemToDisp) for e in beamSetDispRes.elements] File "/usr/local/lib/python3.8/dist-packages/postprocess/output_handler.py", line 779, in <listcomp> lstArgVal=[e.getProp(attributeName+'Sect1')(itemToDisp) for e in beamSetDispRes.elements] TypeError: 'NoneType' object is not callable

But for other elements works fine.

allMemberSet = modelSpace.setSum('allMemberSet', [
                                #  beam1Set,
                                 beam2Set,
                                #  col_frame_a_set,
                                #  col_frame_e_set,
                                 col_frame_c_set,
                                 ])

The commented sets are sets that created with ISection shape.

[lcpt]

Hi @ebrahimraeyat

It seems that the problem is due to the fact, that the ISection class has not its counterparts in ASTM_materials.py and aisc_metric_shapes.py (see WShape, CShape, etc. in both files). Those classes implement AISC code checking for the standard shapes. To make this work with ISection we need to implement the corresponding AISC checking for «generic» I-Shaped sections.

[ebrahimraeyat]

Thanks @lcpt. Maybe inheritance and then write special method for built-up I-shaped sections? Thanks.

Is it better to open an issue?

[ebrahimraeyat]

Implemented as per commit #88 and here is an example:

https://github.com/ebrahimraeyat/xc_soleh