Dear Yves,
Thank you for your response. It is then indeed strange that Abaqus never got around to implement it. I was asking just to make sure there no technical problems lurking around. Your point about quadratic partition of unity not needed makes
a lot of sense. Getfem is a great piece of software!
Regards,
Vikram
From: yves.renard@insa-lyon.fr <yves.renard@insa-lyon.fr>
Sent: Monday, June 5, 2023 1:44 PM
To: Bhamidipati, Vikram <vikram.bhamidipati@swri.org>
Cc: getfem-users <getfem-users@nongnu.org>
Subject: Re: XFEM with second order hexahedral and tetrahedral elements
[EXTERNAL EMAIL]
Dear Vikram,
There is a priori no problem using the Xfem with quadratic elements in Getfem, for both hexahedral or tetrahedral elements. The enrichement with the Heaviside like
function corresponds to cut-elements and is optimal in both situations. I suppose that the limitation in ABAQUS is simply due to the fact that it has not been implemented for hexahedral elements (the only trick is the decomposition of cut elements for integration
purpose which is different in the two cases).
Concerning now the enrichment with singular functions at the crack tip/ front, it depends on how the transition is made between the enriched zone and th non-enriched
zone. If nothing special is done (occurence of wath is called by some authors of blended elements), the partition of unity used will accomodate to approach a cut-off function, so that the fact to use quadratic elements can lead to a better convergence (quadratic
convergence). If the singular function for the enrichement are multiplied by a regular cut-off function whose support is inside the enriched zone, there is no need of a quadratic partition of unity (because what is to be approximated is more or less a constant
in that case).
Best regards,
Yves
De:
"Bhamidipati, Vikram" <vikram.bhamidipati@swri.org>
À: "getfem-users" <getfem-users@nongnu.org>
Envoyé: Samedi 3 Juin 2023 00:31:28
Objet: XFEM with second order hexahedral and tetrahedral elements
Dear Getfem-users,
I am asking this question here since there are developers here who also wrote some important papers on using XFEM for fracture mechanics applications. My question is concerning using second order (quadratic) tetrahedral
and hexahedral elements with XFEM. Can the partition of unity be chosen as linear Lagrange shape functions for the enrichment terms along with a fixed area asymptotic enrichment and expect a second order convergence? Or should the partition of unity be second
order as well? Also is there any difference in how XFEM applies to quadratic hexahedral elements vs quadratic tetrahedral elements? I ask this because it seems like Abaqus software does not seem to support XFEM for quadratic hex elements but does support quadratic
tet elements. Does anyone know why? Does this have anything to do with level set computation for 3D? For Getfem is XFEM supported for both hex and tet quadratic elements?
Thank you,
Vikram
-----------------------------------------------------------------------------
|
|
Vikram Bhamidipati, PhD Senior Research Engineer, Computational Materials Integrity Southwest Research Institute 210.522.2576 |
NOTICE: This email, including any attachments, is intended for the recipient
only and may contain proprietary and/or sensitive information. If you are not the intended recipient please notify the sender immediately, and please delete it. Do not copy or disclose the email to any other person or use it for any purpose. Southwest Research
Institute reserves the right to monitor all email communications through its networks.