3
$\begingroup$

I am trying to model the strain generated by inserting an electrode device into brain tissue. I've had some success in doing this in ANSYS Explicit Dynamics. Here's a link to a GIF showing the penetration.

My issue is that I had to do a very fast insertion (10 m/s) in order to have a faster computation time (~18 h). The actual insertion speed is on the order of 0.1 mm/s, which ends up being 5 seconds of simulation. 5 seconds of explicit dynamics simulation is days of computation on my computer! My question is, can I do this sort of slow penetration/ballistics model in ANSYS transient dynamics or some other implicit model? Or am I stuck with Explicit dynamics and the super long computation time?

$\endgroup$
2
  • $\begingroup$ Hi James, welcome to Engineering SE. You can use this link to edit your profile, if you'd like to change your display name from "user9405" to something more appropriate. $\endgroup$
    – Air
    Dec 30 '16 at 18:04
  • 1
    $\begingroup$ As part of any numerical solution, you should be doing grid and timestep convergence studies. That is, you make the grid/timestep coarser and finer and see how the solution changes. You might be able to get away with a coarser time step for your problem, which will cost fewer iterations (hence less wall-clock time). Keep reducing your time step by an order of magnitude at a time, and as long as it doesn't crash the program and doesn't change the solution much, then it should be ok. $\endgroup$
    – Carlton
    Dec 31 '16 at 0:02
3
$\begingroup$

Given the subject matter my gut tells me to take the time to compute once you've triple checked your conditions. What material properties are you using? You can alter the mesh to decrease cpu usage as well. Just food for thought, but that simulation reminds me of gelatin. If you can find an article that has investigated the hyperelastic material properties of the brain you may find that there is less displacement and therefore much less vibrational analytics. Then again It could be dead on. What are you using for boundary conditions? Crack propagation ?

$\endgroup$
14
  • $\begingroup$ My pleasure. Have you tried abaqus? $\endgroup$ Dec 30 '16 at 18:28
  • $\begingroup$ For the electrode, I used the silicon carbide parameters from the Explicit Materials Library. The real probe is silicon based. For the brain phantom, I used the following parameters, which are a combination of elastic and viscoelastic properties pulled from literature with a Johnson Cook failure mode. As far as boundary conditions, I have the brain phantom fixed at the bottom, and all of the faces of the electrode are going through a linear displacement downwards. Failure/crack propagation is defined by the Johnson Cook parameters $\endgroup$ Dec 30 '16 at 18:31
  • $\begingroup$ I really wanted to try Abaqus, but my lab has an ANSYS license and not abaqus... but, I've definitely seen more papers use Abaqus. $\endgroup$ Dec 30 '16 at 18:32
  • $\begingroup$ Sorry, real link to brain phantom parameters! $\endgroup$ Dec 30 '16 at 18:34
  • 1
    $\begingroup$ A couple of things might help you here if you are still working on this problem. The first is that the model looks symmetric about the centerline of the probe. Cutting the model in half and applying symmetric bcs on that new surface should cut the computation time approximately in half (explicit FE scales nearly linearly with element count). There is also quite a bit of stress wave propagation in the model. You mentioned that displacement bcs are applied linearly with time. Applying them in a smooth step should minimize accelerations. Finally, you might try adding some mass to the model. $\endgroup$
    – Marchi
    Jan 4 '17 at 1:47
0
$\begingroup$

I don't know anything about the material properties of brains, but if you are applying some loads/displacements at 0.1mm/s, I don't understand why you are doing dynamics at all.

I expect the inertia forces caused by slowly deforming the material will be completely negligible compared with the elastic forces. A nonlinear stress analysis (probably with large displacement capability switched on) should work fine. Try it, and compare the results with the dynamics model you already ran.

$\endgroup$
4
  • 1
    $\begingroup$ Think about the size of a neuron. Consider the chemical properties of an organic material. You are applying standard material science (metallic properties) to organic tissue. Also the electrode penetrates the material. Lots to consider. $\endgroup$ Jan 1 '17 at 9:08
  • $\begingroup$ certainly a good way to check, however crack propagation is not elastic deformation and does not follow hookes law. Given the organic makeup of the material you can make certain assumptions to prevent falling into the rabbit hole, however the assumptions must be somewhat plausible the molecular composition of metal vs organic tissue is not, in my opinion, a reasonable simplification. The concern here would seemingly be a totally different subject: the effect of penetrating gray matter and avoiding physical degradation of synapse/neural networks through incision! $\endgroup$ Jan 7 '17 at 6:35
  • $\begingroup$ @ShuddaBeenCodin "You are applying standard material science (metallic properties) to organic tissue" - and you are applying pure fantasy when reading what I actually wrote. $\endgroup$
    – alephzero
    Jan 7 '17 at 23:30
  • $\begingroup$ pure fantasy? I actually up voted your comment, even though you think that penetrating organic tissue is elastic. Tell me, how is the separation of tissue (breaking intermolecular bonds) elastic? I suppose you're considering the deformation about the penetrated tissue, which would be hyperelastic. Pure fantasy is thinking gray matter behaves like metal. The up vote was for suggesting a comparison as that is strong practice in general, even if it shows how flawed your approach is. $\endgroup$ Jan 8 '17 at 8:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.