Anyone used TrueGrid as a hex mesh generator 1

[irishcanadian]

Hi all,
Just looking to see if anyone has experience using TrueGrid as a pre-processor to Abaqus. I'm using CT images for a long bone model where density gradients are of particular importance. The hex meshing in Abaqus is not providing a good mesh and I want to do impact simulations therefore am wanting to use hexahedral elements (hence the need for trueGrid).

 

[IceBreakerSours]

I have used TrueGrid. It has some significant advantages: a) parametric meshing is very convenient with TrueGrid, b) possibility of hex meshing, and c) academic license is super cheap! But that being said, it is not the most convenient software code to use.

FYI:

A) I find the whole CT-based geometry modeling paradigm a little overblown. I can understand it if the geometry is quite complicated, say a cervical vertebra, but definitely not long bones when you could use SolidWorks or something similar to make a very clean geometry. This also reduces all the headaches that come in pre-processing including meshing. Moreover, doing a shape sensitivity analysis, will be straightforward.

B) Secondly, why do you even need hex meshing? If there is no contact involved, simply use quadratic tet elements that any mesher will generate very easily. Secondly, doing a mesh convergence study of, say, strain energy density or stress at a point of interest, will be easy. Besides, if contact is involved, you could use modified tet elements along with membrane elements at the interface. Hex meshing is highly desirable if the expected strains are high like in discs, ligaments, muscles etc.

C) Boundary and loading conditions, and material properties are probably the most important factors in a simple model of femur or tibia under a load. Isotropy vs. anisotropy, time dependence, etc. are highly relevant in such a situation.

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[irishcanadian]

Thanks IceBreakerSours for the reply,
Using the bone specific geometry is highly important in what I am trying to achieve with this study. The bone I'm modeling is a long bone of a horse leg where osteoarthritic changes including increses in density to site specific areas of the bone of interest occur. I am comparing the areas of high stress under impact and quasi-static loading between a healthy bone and an osteoarthritic bone. I will be doing in vivo testing then scanning the bones, therefore using the CT images is of high priority. From what I've heard with impact loading, hexahedral elements are usually the only acceptable/reliable under high speed loading. I am trying to simulate the loads seen with a galloping racehorse during different phases of the stride. And opinions to determine if hexahedrals are required?

 

[IceBreakerSours]

Simple answer: I am not sure.

But if I were to do it, I, most definitely, would not waste time on making the best geometry and the ideal mesh. Why? One of the reasons is that there are other inputs (like stress-strain curves of bones, cartilage, ligaments, etc. or stress-time or strain-time curves of each of these tissues, loading conditions, boundary conditions, etc.) whose variations will have a big impact on the results anyway! One of your inputs may be accurate to the sixth decimal place, but if the other inputs are only accurate to the first decimal, your results too are going to be accurate to the first decimal! So, as long as, there are no clear demonstrated downsides to tet mesh in a specific problem, I do not buy the hex mesh dogma anymore.

Take a look at this link. One thing is for sure that a more complicated tissue geometry under impact was studied and the conclusion is precisely what I've been suggesting.

For a general meshing overview, take a look at this link.

To me, the same is true for CT-based geometries. Almost always, one does NOT want a model so specific whose results are valid for one subject (in your case, one horse!) A good theoretical model is supposed to be as simple and as general as possible. Once the model is verified and validated (after a mesh convergence study), then one must perform sensitivity studies (material, shape, etc.).

Take a look at this example. One can even import a few CT scans in to SolidWorks and make an excellent geometry if it is such a huge concern!

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[irishcanadian]

IceBreakerSours:
Thank you very much for the links. This is very helpful information!

 

[IceBreakerSours]

You may also wish to check this link. These authors used CT data to get the femur geometry. However, they demonstrated sufficient accuracy (in their specific problem) with quadratic tet mesh.

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[burningDiesel]

> So, as long as, there are no clear demonstrated downsides to tet mesh in a specific problem

Well the downside is more computational time in using a tet mesh.
But of course some hours of computational time is less bad than trying to hex mesh your structure and spending 1 hour on it.

 

[IceBreakerSours]

That is correct, sdebock. But, as you probably allude to, it is not a concern in today's times.

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083