UPCOORD vs UPGEOM

[manarshhjot]

Hello,
I am solving the structural problem iteratively, and changing the material properties using an external logic. The boundary conditions do not change with iteration. For updating the node locations after every iteration I used the mapdl.upcoord(1). The results were fine but now I see that it is not recommended to use. It's possible that the code might work now for a test system but fail at real application. Hence, I tried using mapdl.upgeom() instead. The loop runs smoothly but the mesh does not undergo any deformation, not even for the first iteration.

Here is the code I have used:

   global data

    mapdl.clear()
    mapdl.prep7()

    data.mesh.NoElements = read_cdb(data.MeshFile, mapdl)    ### reads the mesh file containing the mesh and BCs

    mapdl.prep7()
    mapdl.fscale(rfact='100000', ifact='1')  

    data.Correction = np.zeros(data.mesh.NoElements)

    for i in range(data.sim.No_Iterations):
           mapdl.prep7()
    
           # Get initial Node coordinates
           if i==0:
                  nodes = mapdl.db.nodes
                  index, cord, angles = nodes.all_asarray()
                  data.mesh.NodeCord.append(cord)
                  data.mesh.NodeIndex.append(index)

             # Material Assignment
             temp_file = os.path.join(data.IterationFolder, 'Material_I_' + str(i + 1) + '.txt')
             ele_mat = create_element_wise_material2(temp_file, data, mapdl)       # create and read new material file for every iteration
             use_correction = [i.use_correction for i in ele_mat]
             data.Material.append(ele_mat)
             
             # Solution
             mapdl.slashsolu()
    
             mapdl.outres('ALL', 'ALL')
             mapdl.solve()
    
             # Post processing
             mapdl.post1()

              # Create Element table
              mapdl.etable('SED', 'SEND', 'ELASTIC')
              sed = mapdl.get_array("elem", 1, "ETAB", "SED")
              data.res.StrainEnergy.append(sed)

              # calculate new correction
              if data.sim.CorrectionMode:
              data.Correction = calc_correction_factor(sed, data.Correction, dens, use_correction)
        
              # Function to update coordinates ####
              mapdl.prep7()        
              mapdl.upgeom()

              # new node coordinates
              nodes = mapdl.db.nodes
              index, cord, angles = nodes.all_asarray()
              data.mesh.NodeCord.append(cord)
              data.mesh.NodeIndex.append(index)

[mikerife]

Hi @manarshhjot it really helps to see the whole picture when troubleshooting. I'm guessing that MAPDL was launched in distributed parallel mode (the default for later releases). If so we need to issue a SET command after entering the /POST1 processor to read in a set of results from the result file. Unlike shared memory parallel (SMP) DMP does not fully write results to the MAPDL database as part of the normal solve process. So we need to read in the result data.

Mike