User login


You are here

Glasgow: 4-year Post-doc XFEM for 3D crack growth with goal-oriented error estimation including industrial applications

University of Glasgow, Civil Engineering, Mechanics and Materials

Increased Reliability for Industrially Relevant Automatic Crack Growth Simulation with the eXtended Finite Element Method (Funding for £483k)

A unique EPSRC funded 4 year post-doc to investigate

- goal-oriented error estimation in XFEM

- adaptivity for 3D crack growth

- industrial applications to damage tolerance assessment of complex structures

- incorporation into commercial codes

This project is in collaboration with 

- Rolls-Royce plc (UK)

- Bosch GmbH

- inuTech (diffPack, Germany)

- CENAERO (Belgium)

- Northwestern University (Ted Belytschko -- 3 month sabbatical at Northwestern)

and includes

- Hughes-Belytschko training course

- DiffPack training course

- 6 months work with our commercial partners (Bosch, Rolls-Royce, inuTech, CENAERO)

Details pertaining to salaries will be provided shortly.

The suitable applicant has

- a strong motivation to work in a dynamic team 

- a will to strive and succeed as a leading researcher in their field

- an interest in industrial applications of computational fracture mechanics

- an open mind

- C++ or object-oriented experience

- XFEM experience

- Fracture Mechanics or error estimation experience

 Applications should be sent to 

stephane dot bordas at gmail dot com (CV, list of publications, list of at least 3 references)

Lay summary of the research proposal: 

 This project will deliver new computational modelling tools that will allow engineers working on
safety critical structures to rationally assess the effects of crack initiation and crack propagation. Such
problems have to date remained intractable.

The research will permit unprecedented understanding of crack
propagation, thereby delivering less conservative designs, and, most importantly avoid unpredicted
catastrophic failures in service.

This is possible by building upon the recent success of the extended finite element
method (XFEM), which has emerged as a revolutionary simulation tool for modelling discontinuities and has the potential to require an order of magnitude less engineering time than conventional methods.

Yet, this new method requires much reliability improvements to invade industry. By leveraging recent theoretical
and numerical developments and working hand-in-hand with future users, this project has the potential to
provide XFEM with the accuracy and robustness it requires to become the new tool of choice for structural
integrity predictions and reconcile accuracy and computational tractability.

Cracks or defects are almost always present in engineering structures. In aerospace engineering for instance, during the life of the aircraft (take offs, flights and landings), these cracks will grow under the influence of the forces applied to the structure. How do engineers ensure that, despite these growing cracks, the aircraft can still be operated safely?

The idea is to regularly inspect the aircraft to monitor the major cracks. The next question is to know how often should an aircraft be inspected to prevent catastrophic failure between two inspections. To answer this question, engineers must be able to evaluate the time (number of flights) it takes for the cracks to become fatal to the structure. If it takes 1,000 flights, the maximum inspection interval should be less than 1,000. To estimate  the time to failure, engineers use computer methods, where they model the behaviour of the structure using various simplifications: this is known as "Damage Tolerance Analysis" (DTA).

However, today, existing software are still unable to provide engineers with a rational tool to assess the tolerance of a structure to damage. The proposed research has the long-term goal to provide this tool which could provide a paradigm shift in the way engineers think about simulating fracture, whereby sufficient accuracy would not be synonymous with intractable computational time or manpower.


phunguyen's picture

Hi Stephane,

It is a very interesting topic and a big chance for anyone interested in fracture mechanics to work with Prof. Belytschko.

I have one question. One of the main advantages of XFEM against other methods in crack modelling is the absence of remeshing since projection of history variables from the old mesh to the new one inevitably induces errors. So, why you do adaptivity in XFEM? It seems to me that you would make lost the pros of XFEM. No? 


Dear Phu,

 The group in Glasgow is really glad that you found this topic of interest, if you know someone who may be interested, please let me know, although I have already quite a few very good candidates. 

The idea behind adaptivity is not necessarily to do remeshing or p-adaptivity, but enrichment (e-) adaptivity, as we described, Marc and I in our papers as did Prof. Belytschko in his latest "parametric" enrichment paper. 

It is true that X-FEM can work without remeshing, but in 3D, it is not so obvious, especially for growing cracks. That is based on Marc's work and my work on XFEM for industrial structures that we found that out. We can discuss thatmore on the forum at some point if you feel like it.

  Derivative recovery and a posteriori error estimate for extended finite elements
Author(s): BORDAS, S; DUFLOT, M
Source: COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING Volume: 196 Issue: 35-36 Pages: 3381-3399 Year: 2007
Times Cited: 6
DOI: 10.1016/j.cma.2007.03.011

 Title: A posteriori error estimation for extended finite elements by an extended global recovery
Author(s): DUFLOT, M; BORDAS, S
Source: INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING Volume: 76 Issue: 8 Pages: 1123-1138 Year: NOV 19 2008
Times Cited: 2
DOI: 10.1002/nme.2332

 Title: A simple error estimator for extended finite elements
Author(s): BORDAS, S; DUFLOT, M; LE, P
Source: COMMUNICATIONS IN NUMERICAL METHODS IN ENGINEERING Volume: 24 Issue: 11 Pages: 961-+ Year: NOV 2008
Times Cited: 2
DOI: 10.1002/cnm.1001




Dr Stephane Bordas

Subscribe to Comments for "Glasgow: 4-year Post-doc XFEM for 3D crack growth with goal-oriented error estimation including industrial applications"

Recent comments

More comments


Subscribe to Syndicate