User login

Navigation

You are here

MSc+PhD position Fully Funded -- Massively parallel biomechanics simulation of brain surgery on HECToR

High Performance Computing MSc+Ph.D. position available at the
University of Glasgow on Massively Parallel Brain Surgery Simulation
with the extended finite element method (XFEM and FleXFEM)  (University
of Glasgow) -- funding body is EPSRC.

One year MSc in HPC in Edinburgh (all costs covered by funding) + 3 year Ph.D.  and access to HecToR,
one of the world's largest super-computer, including training with
experts in massively parallel simulation (10,000+ processors).

Supervisor: Dr Stephane Bordas,Dr Lee Margetts (Manchester)

Collaborators: Prof. Ray Ogden and Prof. Gerhard Holzapfel

Medical experts: two expert surgeons
in Belgium and medical imagery + computer aided/guided surgery
specialists, access to one in 30 interventional MRI scanners.

Eligibility: UK students have full funding. EU students will be discussed on a case-by-case basis.

 Non-EU students should apply to ORSAS http://www.orsas.ac.uk/ and must have an outstanding research activity with international publications in leading journals to be considered. 

EU Students should have a first class degree and preferably an MSc in numerical methods or/and computaitonal mechanics.

 UK students: A Good second class or first class degree (preferred) or MSc in any
field of engineering or science: mathematics, computer science,
physics, biomedical engineering, civil engineering, mechanical
engineering, aerospace engineering, electrical engineering.

Student should be self motivated and hard working with a strong
interest in biomedical engineering and an outstanding background in
Mathematics and Physics.  

Standard EPSRC Stipend + all fees covered.  

Aims and Objectives

                             This   project   aims   to   devise   and
                         validate  a uniquely effective fully parallel  surgery simulation tool for  use  in the
                          training, rehearsing and objective evaluation of surgeons to eradicate much of the
                          uncertainty   in   improving   existing   and   creating   new   surgical   procedures.   High
                          Performance   Computing   (HPC)   is   today   the   only   way   forward   to   simulate   the
                          effects   of   various   strategies   for   cutting   and   manipulating   brain   tissue   both
                          accurately and in real­time. Achieving this will provide a step­change in surgical

 Aim  The main long ­term aim of the research in which this studentship is inscribed is to  reconcile  
real­time and accuracy  in brain surgery simulation through cutting ­edge computational mechanics,
high­performance computing and realistic brain matter mechanical models. Achieving this aim is not
reasonably possible within a Ph.D.­level research, the proposed work will provide a detailed proof ­of­
concept required for further research by tackling the following four objectives.


Objectives   (1)  Optimise   accuracy   versus   computational   cost   through   cutting­edge   numerical  
methods (2) Develop and test massively parallel algorithms developed by leaders in HPC to achieve  
real­time   simulations  (3)  Utilise   rigorous   experimentally­informed   mathematical   models   of   brain  
tissue developed by world leaders in the field (4) Validate and verify the proposed simulation tool.
For the first time in the field of surgical simulation, the increased efficiency (objectives 1 and 2) will
allow realistic non­linear material models to be used (objective 3) without sacrificing accuracy.

What is there for you in this project?

Benefit to the student: Suitability of the project for training Through the unique HPC training, the
student will acquire cutting­edge skills in high­performance computing. By working closely with Lee
Margetts, and expert in HPC, the proposed project will allow him/her to build upon and hone these
skills beyond the MSc training. What is more, the proposed research is highly multi­disciplinary and
will  
train   the   Ph.D.   student   in   computational   biomechanics,  
which   is   a   leading   theme   in   today's research. More
specifically, two major sets of skills will be acquired:

Numerical methods for evolving discontinuities  [OpenXFEM++, FleXFEM, SB] XFEM is one of the
most highly researched fields in computational mechanics ­­5 articles in 1999­2000 and 74 in 2006­
2007 for a total of 138 journal papers between 1999 and 2008 (source: scirus) and more than 400
citations   of   the   original   paper   [8].   This   method   and   its   sibblings   such   as   the   Flexible   XFEM
(FleXFEM) is very likely to become an industrial standard in the coming decade. The student will join
SB's group, one of the most active and recognized developers of XFEM today both academically and
industrially, which will be essential in the transfer of his/her skills after the Ph.D.

Element­ by­ element   massively   parallel   architectures  [ParaFEM,   LM]   An   essential   point   for   this
research is that after his/her Msc, the training of the student in HPC will continue intensively, through
the   involvement   of   Lee   Margetts,   a   recognized   expert   in   massively   parallel   computing   and   an
experienced developer of HECToR supported packages. In particular, the student will be gradually
introduced to ParaFEM, developed and maintained by Lee Margetts.

Additionally,   the   student   will   benefit   from   the  
ongoing   work   of   the   team   of   Profs.   Ogden   and Holzapfel
in soft­tissue modelling and be exposed to some of the leading­edge
research by Prof. Ray
Ogden's group, of the host organisation, a world leading expert in fundamental theory in nonlinear
elasticity. These sets of skills are highly transferable and will provide the individual with a powerful
springboard for a successful career.

Please contact me for more details, this is only a preliminary advert

http://www.civil.gla.ac.uk/~bordas

stephane dot bordas at gmail dot com

See also:


http://www.epsrc.ac.uk/PressReleases/HPCStepsUpAnotherGear.htm
http://www.epcc.ed.ac.uk/training-education/hec/

 

Comments

Pooyan Broumand's picture

Hi ;

 I am modeling quasi-static crack propagation with x-fem . 

for calculation of SIFs I used the extrapolation method , but I found that the displacements at the tips are not very smooth and are turbulated. and as the result the SIF calculated are not correct . 

Why the displacements are not smooth, and there exist a high gradient in the tip element. is it in X-Fem nature or it is my code problem?!

another thing is that where should the points to be used in SIF calculation be located ?

In FEM they are located at the edges of the crack but in X-FEM there are turbulances near the crack edge and this will affect the results.

thanks 

Dear Pooyan,

 What you are describing is not a problem with your code (at least not necessarily). 

There are many papers on this topic, for instance, I recommend the following:

 http://doi.wiley.com/10.1002/nme.1601 by Karihaloo Improving the accuracy of XFEM crack tip fields using higher order quadrature and statically …

 Direct evaluation of accurate coefficients of the linear elastic crack tip asymptotic field by Karihaloo again

 XFEM for direct evaluation of mixed mode SIFs in homogeneous and bi-materials still by Karihaloo

 Derivative recovery and a posteriori error estimate for extended finite elements by Marc Duflot and Stephane Bordas

 http://www3.interscience.wiley.com/journal/119054430/abstract by Marc Duflot and Stephane Bordas

 A recovery-type error estimator for the extended finite element method based on singular+smooth … by Rodenas, etc.

 A Global Explicit Residual Based Error Estimator for the eXtended Finite Element Method in … by Geniaut and Delmas

 A simple error estimator for extended finite elements (Marc Duflot and Stephane Bordas and Phong Le)

 Numerical Aspects of the extended Finite Element Method by Peters and Hackl

Do let me know if you would like preprints of those articles, which I could provide.Our XFEM matlab code could be of interest to check your code as well. I wonder if you could describe what you mean by extrapolation technique specifically?    

Many thanks, 

Dr Stephane Bordas

http://people.civil.gla.ac.uk/~bordas

 

Subscribe to Comments for "MSc+PhD position Fully Funded -- Massively parallel biomechanics simulation of brain surgery on HECToR"

More comments

Syndicate

Subscribe to Syndicate