Export Fiber orientation tensor

[M.Brian]

Hello everyone,

I am wondering if it is possible (and if it is, how) to export the tensor or matrix used for the fiber orientation tensor and average fiber orientation results from Moldflow 2013 to Mathematica (or if there's another software that would work, I'm open to suggestions).

What I am trying to do is compare the fiber orientation predicted by Moldflow to the fiber orientation of an injection molded part found by micro-CT scan.  I started by copying the image to Mathematica and analyzing the colors displayed in a section of the image to get the degree of the average fiber orientation at that section; however, this doesn't tell me about the direction the fibers are oriented in.  It also only tells me the average fiber orientation through the thickness of the part; I would like to be able to analyze the fiber orientation at each level inside the part as well.

Any help or suggestions are appreciated.

Brian

[Alex.Bakharev]

Hi, Brian

You need to write a short ~10 line script using Moldflow API. In the help file menu (under the question mark in the top) there is API reference section. Start with Class List/PlotMgr/GetTensorData . The ID for 3D fiber orientation result is 4021.

If you have difficulties writing the script I could do it for you, just tell me the format of the output file you want. Or maybe if you would tell me there you have orientation data from your micro-CT I could extract the data from the same location.

BTW at Autodesk Moldflow we are quite interested in different validation experiments of our software. We might help you with the software in exchange for th experimental data. The right person to contact is Dr. Franco Costa (franco.costa@autodesk.com), he is coordinating R&D colloboration. You cal also contact me (alex.bakharev@autodesk.com). I have a particular interest in fiber orientation.

[tim.vanast]

Hi Brian,

Aside from Alex's answer on exporting the data, are you trying to see the fiber orientation through the layers instead of just the average?  If so, you can view them by animating the results for the "Fiber Orientation" result.  You can also edit the plot properties to change the view of the result to arrows, or straight lines, etc.

Good luck.  It's  a great topic, but definitely a challenge.

Tim

[M.Brian]

Firstly - thank you for your help and quick response.  That was exactly what I was attempting to find in Moldflow.

Secondly, it is not actually my data.  I am a research assistant for Professor Fariborz Parsi at Winona State University.  However, I have forwarded your response to him, and he will be in contact with you and Dr. Franco Costa with more information regarding the research and experimental data.

Thanks again for your help,

Brian

[franco.costa]

Rather than an API script, I wonder whether it would not be simpler for you to just export the result into a patran format text file and then read it into whatever software you are using. 

To do this:

1. Display the fiber orientation tensor result
2. If you are using the Ribbon UI: On the result tab of the ribbon menu, click on the “Export and Publish” pull-down and then select Patran. This will write the full fiber tensor result set into a text file.
3. If you are using the Classic UI: You will find the “Save Plot Data in Patran Format” item on the “Results” menu.

If you are running a 3D analysis, you will get a file which contains the 6 tensor components for every node.

If you are running a midplane or dual-domain analysis, you will get files which contains the 6 tensor components for every element. Each file will correspond to a different laminate through the thickness.

The API script could also be used to automate this export to text file step, or it can be used to access the data directly.

regards

Franco

[M.Brian]

I've been working wiht the patran export (I didn't know that was there- thanks), and have been importing the data into Excel for analysis.  However, as I was analyzing the data in Excel, I noticed that the number of nodes saved and imported into Excel was not equal to the maximum number of nodes in the part.  I am missing nodes- roughly 5% of the nodes on my three differently meshed trials disappeared between the part and Excel.

I know these nodes exist in the part and weren't just renumbered or deleted in mesh editting.

So- I'm wondering where these values are disappearing, and if they are retrievable.  It could be computer memory- the three trials I ran were a coarse mesh (edge length 5mm), fine mesh (edge length 1mm) and very fine mesh (edge length 0.5mm).  The number of nodes in the meshes were approximately 8000, 100000 and 400000 respectively.  If it is the computer memory, there's nothing I can do about it, so I'm hoping it isn't.

If it's relevant, I'm working in Windows 7, 64 bit, quad-core (Intel Xeon 2.67GHz) with 12GB (total) RAM.

Any ideas on where they could be / what is causing the problem?

Also, and on a different topic- when selecting a large number of nodes and triangles, it comes up with the list of Nxxx Nxxx Nxxx... <> for sets too large to display fully.  Is there a way (besides selecting smaller sets of nodes and triangles) that I can have all the values displayed?

Thanks for any help or suggestions

Brian

[franco.costa]

Hello Brian,

Does your study file include beam elements for the feed system (runner, gate, sprue)? Depending on your fiber analysis settings, these may not be included in your fiber orientation calculation.

Does your study file include cooling lines, part inserts, cores or mold boundary elements?. Obviously these will also not be included in the fiber orientation results output.

On your second question, it is possible to get the full selection list if you write an API script to iterate through the selection list.

regards

Franco

[thakua2]

Hello ,

I have worked with fiber filled materials and performed virtual DOE's to compare different design parameters for Plastic Part optimization.

My Objective was looking for results(Y's) which can be measured to evaluate my design concepts.(X's).

Fiber orientation tensor results were one of the things I tried to measure. I used the patran export option, read it using MS Excell and then analyse it .

Though excell has its limitation when the Model is having large no. of elements.

As I try to analyse the simulation data, I always check the normal distribution of the data and try to compare the mean,stdev and range results using statistical software.

This complete process is quite tedeous but helps us to reduce measurement noise in data analysis.

I would like to know, is it possible to get the statistical results(mean,stdev,range,percentile etc...., of the MOLDFLOW RESULTS(nodal as well as elemental) by any means in Moldflow itself.

Regards,

Aritra Thakur

[Alienware18]

Hello Alex,

I'm an FEM engineer and I use Ansys for structural simulations and Moldflow for plastic injection molding. I'm new with Moldflow and I'm trying to extract moldflow data, more exactly the fiber orientation tensor file.

The mesh has been initially a dual domain mesh in Moldflow afterwards I transformed it in a 3D mesh with 8 layers through thickness.

I was told I can use the mpi2ans command in order to generate the files I need because I need the mesh file also and by using that command the mesh file *.cdb file is created and also an initial stresses file *.ist is created but the fiber orientation tensor file with the extension *.nod or *.xml are not generated.

Please let me know what seems to be the problem and how can I generate the required file after running a Moldflow simulation

Thank you very much Alex.

Kind regards,

Daniel Ionut VERZES

[wasay.bradford.ac.uk]

Hey hi Brian,

I am also working on a similar task. Do you have a code or something which I can look as a reference. As a student its really hard imagining these things.

[wasay.bradford.ac.uk]

Hey hi Frank,

As you mentioned in your comment that we 'will get files which contains the 6 tensor components for every element', is it the fibre orientation tensor you are talking about or whether they are the principle eigen vectors in the local 1 and 2 directions because I read on the autodesk website that beginning in MPI 6.0, Autodesk Simulation Moldflow Insight directly outputs the principal directions of the material property set. Orientations are defined in job-name_principalDirections.xml by giving vectors defining the local 1- and 2-directions.

Can you please kindly explain this, I have a set of data from a previous student and I dont know whether it is the fibre orientation tensor or the eigen vectors in local 1 and 2 directions.

Many thanks,

Wasay

[franco.costa]

Hello Wasay,

This thread is talking about the export of the fiber orientation tensor, so my comments were about exporting the tensor result to a Patran file which will contain the 6 components of the tensor. You can confirm that these component values in the file correspond to the component values of the fiber orientation result which you can view in Moldflow Synergy.

The output of the principal directions which you mention is a whole different export mechanism of the composite mechanical properties and their principal directions - This is not related to an export of the fiber orientation tensor.

Are you talking about measurement data from either Bushra or Millan? I had involvement in both their projects. Let me know if you need a more detailed discussion about what you are trying to do with the software. Your supervisor has my contact details.

best regards

Franco

[wasay.bradford.ac.uk]

Dear Dr. Franko,

Many thanks for your answer, I was out travelling and couldnt answer you unfortunately. I think I was just confused as to the results I had,but now there is more clarity. No its not the same Project as Bushra and Millan.

I am actually working on a Project where I have results from a moldflow experiment and a CT scan. Now I want to develop a mechanism so as to validate the results of mold flow with the experimental results from CT scan. The Problem is that the mesh sizes are very different with a more fine mesh in Moldflow and a very coarse mesh in the scan. Can you give me some insight as to how I can validate the model ?

[franco.costa]

Hello Wasay,

the way we usually compare measured and predicted orientation data is to make an XY Plot of diagonal components of the orientation distribution tensor along a probe line through the part thickness are particular locations. This gives a nice easy to digest overview of the skin-core-skin nature of the orientation distribution through the thickness.

When I say the diagonal components of the tensor, it is usually more intuitive to look at the tensor as resolved in global Cartesian axes (rather than the principal tensor directions). Traditionally, direction 1 is the flow direction and direction 3 is the thickness direction. However, resolving the orientation tensor into the Cartesian directions is usually only reliable along a symmetry axis of the part. Elsewhere, the off-diagonal terms will be significant and in such locations looking at the principal directions of the tensor would be more reliable.

Whatever you do, make sure you are comparing like-for-like: Do not compare Cartesian components from simulation to principal directions from measurement (or vice-versa).

In the Moldflow simulation results, you can adjust which component/direction is being displayed in an XY-Probe Plot by editing the plot properties.

[wasay.bradford.ac.uk]

Dear Dr. Franko,

Many thanks for your answer.

Kind regards,

Wasay