Extracting certain items of element values from result files

[arun-maniam]

Hi,

I have a plate which is subjected to harmonic excitation. Harmonic analysis was performed for a certain frequency range. The image below shows the element plot of the plate. The element enclosed by the red box is (for example) of interest in the post processing. The plate has 1000 elements and the area enclosed has 60 elements.

With the command mapdl.set() I defined the dataset that to be read. And then with mapdl.etable() all the element values(1000 elements) are written into the etable. However, I dont need all the values of the elements, I just want the values of elements enclosed by the red box(60 elements). I created a masking array, e_mask for those 60 elements. With mapdl.vmask('emask[1]') I specified the array parameter as a masking vector. In an attempt just to extract the elemental values of those 60 elements i used mapdl.starvget(). However, I still get an array with size 1000. Among those 1000 elements, only 60 elements have a value and the rest has a value of zero. I can use another masking vector in python to extract the non-zero values but it is not efficient as i need to do the same thing so many times for other element values that im extracting from the result file, making my simulation very slow.

I also tried creating a component (mapdl.cm()) for those 60 elements and selected the component(mapdl.cmsel()) after the mapdl.set() command, attempting to just retrieve the values of those 60 element when reading into element tables. However, it didnt work:(

So my question is, what is the most efficient way of getting the elemental values of those 60 elements only? Is there any commands in pymadl has can perform such extraction of certain elemental solutions. Thanks in advance!

Below you can find a section of my code:

mapdl.esel('S', 'ENAME' , '', 281)
mapdl.esel('R', 'CENT', 'X', Res_x1-4*e_size, Res_x1-e_size)
mapdl.get('e_mx' , 'ELEM' , 0 , 'COUNT' )
e_mx = mapdl.get('e_mx' , 'ELEM' , 0 , 'COUNT' )      # convert mapdl parameter into Python variable
e_mx = int(e_mx) 
mapdl.dim('emask' , 'ARRAY' , 'e_mx' , '1')
mapdl.starvget('emask(1)' , 'ELEM' , '1' , 'ESEL')



mapdl.finish()
mapdl.post1()
mapdl.set('last','last')
mapdl.get('n_lstp','active','','set','lstp')
mapdl.get('n_sbst','active','','set','sbst')
nfreq = mapdl.parameters['n_lstp']*mapdl.parameters['n_sbst']

n_lstp = mapdl.parameters['n_lstp']
n_lstp = int(n_lstp)
n_sbst = mapdl.parameters['n_sbst']
n_sbst =  int(n_sbst)
for ll in range(1, n_lstp + 1):
    for ss in range(1, n_sbst + 1):
        nn = (ll - 1) * n_sbst + ss

############# Import of velocities into element tables ######################
        mapdl.set(ll , ss, 'VELO' , 'IMAG' ,'' ,'')     # If Fact=VELO, the harmonic velocities (v) are calculated from the displacements (d)
        mapdl.etable( 'Vx_r_h', 'U', 'x'  )
        mapdl.etable( 'Vy_r_h', 'U', 'y'  )
        mapdl.etable( 'Vz_r_h', 'U', 'z'  )
        Vx_r_h1 = mapdl.get_array('ELEM', item1='ETAB', it1num='Vx_r_h')
        Vy_r_h1 = mapdl.get_array('ELEM', item1='ETAB', it1num='Vy_r_h')
        Vz_r_h1 = mapdl.get_array('ELEM', item1='ETAB', it1num='Vz_r_h')


        mapdl.vmask('emask[1]')
        mapdl.starvget('Vz_r_h1_test','ELEM','', 'ETAB','Vz_r_h') 

[germa89]

I see no red box... but whatever it is not relevant.

== Update ==

Regarding getting the arrays using mask. You are right, but you are missing one last step.

As you correctly did, you first need to create your masking array (emask):

mapdl.esel('S', 'CENT', 'X', 0, 0.5)
e_mx = mapdl.mesh.n_elem
mapdl.dim('emask' , 'ARRAY' , e_mx , '1')
mapdl.starvget('emask(1)' , 'ELEM' , '1' , 'ESEL')

Then you can proceed to retrieve the results into a new array.

mapdl.allsel()
mapdl.etable("UX", "U", "Z")
mapdl.vmask("emask")
mapdl.starvget("FULLARR", 'ELEM', item1='ETAB', it1num='UX')

Because the mapdl.vmask command does not affect the mapdl.etable command, your table has all the elements.

The mapdl.starvget does get affected by the mapdl.vmask but only on the "writing" part of the resultant array ("FULLARR"), therefore the results you get are the full array, but with only the values you specified throught the mask. If you were retrieving those results, you would get:

>>> mapdl.allsel()
>>> mapdl.etable("UX", "U", "Z")
>>> mapdl.vmask("emask")
>>> UX = mapdl.get_array('ELEM', item1='ETAB', it1num='UX')
>>> UX[:50]
array([[61.10726809],
       [52.17670105],
       [49.92137412],
       [49.29690002],
       [49.09603702],
       [ 0.        ],
       [ 0.        ],
       [ 0.        ],
       [ 0.        ],
       [ 0.        ],
       [52.17670105],
       [42.9878871 ],
       [40.6360634 ],
       [39.92410881],
       [39.6757068 ],
       [ 0.        ],
       [ 0.        ],
       [ 0.        ],
       [ 0.        ],
       [ 0.        ]])

If you want to get a condensed array and get rid of the zeros, you need to have some changes in the above code:

>>> mapdl.allsel()
>>> mapdl.etable("UX", "U", "Z")
>>> # with mapdl.non_interactive: # I would use a non-interactive command here. See later comment.
>>> mapdl.vmask("emask")
>>> # We do not use here `mapdl.get_array` so we do not retrieve the array from MAPDL (faster)
>>> # but also we know the array where the ETABLE data is stored `FULLARR`
>>> mapdl.starvget( "FULLARR", 'ELEM', item1='ETAB', it1num='UX')  
>>> mapdl.vmask("emask")
>>> mapdl.vfun("MYARR", "COMP", "FULLARR")
>>> UXCondensed = mapdl.parameters["MYARR"]
>>> UXCondensed[:50]
array([[61.10726809],
       [52.17670105],
       [49.92137412],
       [49.29690002],
       [49.09603702],
       [52.17670105],
       [42.9878871 ],
       [40.6360634 ],
       [39.92410881],
       [39.6757068 ],
       [49.92137412],
       [40.6360634 ],
       [38.3379692 ],
...

Also I would recommend you to run everything in non_interactive, because two reasons:

• You don't want to retrieve values until the end (you are being non-interactive)
• The *VMASK affects to the first array operation (V** or M*** commands) right after it. PyMAPDL issues commands under the hood to facilitates things, but it does not really check for *VMASK to have been issued or not. So you might be applying the masking to an internal PyMAPDL operation. So be careful. Running everything in non-interactive reduces this possibility (it does not avoid it all together).

Interesting question by the way. Thank a lot!

[arun-maniam]

Hi @germa89, I could see the red box in the previous attached image. By red box i meant the rectangular box as highlighted again below.

[mikerife]

Hi @arun-maniam & @germa89
I think the PyMAPDL "enriched output" to_list is a good solution here. MAPDL will list element tables for just the elements selected. So we just need to get that data. I created a one part test model; a 24x4 area meshed with shell elements 0.25 thick. Let's say we want the results at element centroid X at 15.5 (meshing with element size 1). This is a full harmonic analysis run at 20 frequencies.

mapdl.post1()
mapdl.esel('s','cent','x',15.5)
mapdl.nsle()

results = []

for sub in range(1,21):
    mapdl.set(1, sub, 'velo', 'ampl')
    etable = mapdl.etable('uyetab', 'u', 'y')
    etable_list = mapdl.pretab('uyetab').to_list()
    results.append(etable_list)
mapdl.allsel()

Since the mesh size is 1, and I'm selecting just one 'row' of elements in X, the element count is 4. The result of the loop gives (results):

16, 40, 64, and 88 are the selected element IDs. The result is UY. Anyway, would this be acceptable?

Mike

p.s. we could append the whole mapdl.pretab... command but I thought this was a little more readable.

[germa89]

This is a great solution too. I can't believe I forgot about to_list ... and I did implement it!

Thank you for the suggestion Mike!