How to get certain matrix/fiber volume fractions?

[yanchen291]

Hello there,

I am trying to model a 2D plain weave with certain matrix/fiber volume fractions, i.e. matrix volume fraction 68% and fiber 32%. I was able to get the desired volume fractions by the following steps.

1. Create a 2D plain weave with given fabric thickness, yarn width and yarn spacing data. (Those values are fixed, so the x, y dimensions of the RVE are fixed.)
2. Query the yarn volume in the Python console and calculate the RVE volume needed.
3. Modify the default RVE in the z direction to get the calculated RVE volume.

The material properties I got from this model appear to be incorrect, especially in the z direction. For instance, the unit cell z direction CTE was calculated to be 157 ppm/degree, while the matrix CTE was at 126 ppm/degree and fiber CTE at 5 ppm/degree. I suspect my method of changing the RVE z dimension to achieve desired volume fractions was wrong.

Does anyone have any other suggestions? Thanks in advance.

Regards,
Yan

PS: the fabric thickness input in the 2D weave wizard is the thickness of 1 ply of fiber glass without the matrix, correct?

[louisepb]

Hi Yan,

It would be worth looking at how closely the model represents your actual textile. The fabric thickness in the 2D wizard produces a textile with each yarn with a height of thickness/2. The reality is probably that your composite would have been compacted and you may therefore need to modify your model to give a more realistic representation of your finished composite; the thickness of the laminate would probably end up being less than the thickness of 1 ply of the dry woven material.

The attached paper looks at sheared textiles but may help.

Best regards,
Louise

[yanchen291]

Hi Louise,

Thank you for your reply and the reference paper.

What I am trying to do is similar to what was presented in the paper. I need a model to predict CTE in x, y, z of a laminate made from a certain glass fabric (Vf = 32%) and matrix (Vf = 68%). I try to start with a single layer model with PBC in all directions, but I am having doubts.

1. I have measurements for yarn spacing, width and the final laminate thickness. The single layer unit cell based on those values did not have the correct ratio of fiber and matrix. I tried to change yarn height, but that created large gaps between the yarns, which was not the case of the actual cross section of the laminate. What else can I change to get the volume fraction desired, but maintain the yarn spacing and width?

2. In my model, I used fiber properties as yarn properties. However, in the paper, it stated that yarn properties were derived from another FEA model. If possible, can you kindly provide some references on the model that was used for yarn property calculations?

Regards,
Yan

[louisepb]

Hi Yan,
The main thing that you can do is to change the cross-section along the length of the yarn, either at the nodes or at specified points along the yarn.

If you know the height of the yarn in the original textile then one thing you could try is to use the 2D wizard, select 'Sheared textile' but leave the angle at 0. Specify the fabric thickness as the final laminate thickness and then in the Weave Pattern dialog change the height of the yarns as described here: http://texgen.sourceforge.net/index.php ... e_2D_Weave Choose the refine option and then it will change the cross-sections to fit the overall thickness. I'm not sure if that will help in your case but it could be worth a try.

I've attached a paper which gives some information about obtaining the yarn properties.

Hope that helps,
Louise

[yanchen291]

Hi Louise,

Thank you for your help.

I figured since Abaqus treats the yarns as solid sections with fiber fraction of 1, it is very difficult for me to capture an exact volume fraction ratio between the yarn and the matrix. I have thus decided to closely capture the geometry of the yarns instead. The volume fraction ratio issue can be compensated by using calculated yarn properties that have taken into consideration of the embedded matrix inside the yarns.

While attempting to model the yarns as measured, I have come across another issue. The laminate I am modeling has a pressed thickness of 0.130 mm and the height of the yarn cross sections are measured to be 0.0381 mm. After creating the plain weave with 0.130 mm as the fabric thickness, I went back to reduce the height of the yarns to be 0.0381 mm. This creates a large gap among the warp and weft yarns, see attached. How can I get rid of the gap?

Regards,
Yan

[louisepb]

Hi Yan,
What does the cross-section of your actual textile look like? With the dimensions you have specified for thickness and yarn heights there must be a gap somewhere. If the laminate has resin on either side of the textile then you could specify 0.0762 as the thickness and then increase the domain size to 0.13mm later.
Alternately you could move the nodes. It may be easier to write a script to specify your textile. Either way, I think you need to know exactly what you want your textile model to look like so that know what you're aiming at.

Best regards,
Louise

[yanchen291]

Hi Louise,

Thank you for the quick reply again.

The actual cross section of the pressed laminate does not have noticeable gaps. I tried method 1 for a quick evaluation, which is to create the weave with a thickness of 0.0762mm and increase the domain size to 0.13mm afterwards. I am running into the same problem as mentioned in the first post of this thread again. The resin has a CTE of 127 ppm/c and the yarn has a CTE of 5 ppm/c, but the resultant z-direction CTE is 148 ppm/C for the laminate. Since this is just a quick evaluation, I understand that the inaccuracy in geometry and material properties will cause inaccurate predictions for the final laminate properties, but the z-CTE of the laminate should at least be within the bound of (5, 126) ppm/c. I am perplexed of what is causing such an unrealistic value for the z-direction CTE. Do you have any insights on this?

Regards,
Yan

[louisepb]

Hi Yan,

We can't really think of an explanation for this. The thickness of your composite seems very small - is this correct? Are you sure that you have assigned the correct CTE's for your yarns? Do you have different values for the transverse and longitudinal directions?

As the thickness is so small it might not be appropriate to use the periodic boundary conditions. You could do a check of your method by assigning the same CTE for both yarn and matrix and checking whether the result of your simulation gives the same value. If not then you would need to look at the method you're using.

I hope that helps,
Louise

[yanchen291]

Hi Louise,

Thank you for your help. I did a check with same material properties for both the yarn and the matrix. The simulation returned correct results, the same input CTE value. In regards to your questions,

1. The thickness of your composite seems very small - is this correct? Yes, it is correct. The laminate only has one ply of glass fabric, so it is very thin after being pressed.
2. Are you sure that you have assigned the correct CTE's for your yarns? Yes, I did assign the correct properties.
3. Do you have different values for the transverse and longitudinal directions? I tried to simulate with both isotropic and orthotropic material models for the yarn. Both came back incorrect.

I did notice that once the thickness of the model falls under a certain value, the CTE in z-direction will be out of bound. It is very likely that the PBCs are not appropriate for thin models.

The reason I need such a thin model is for building a CTE predictor for laminates made from different glass fabrics. I want to use those thin models to build a library of material properties for various glass styles with certain resin content. Then use them for laminate assemblies, see attached. I am new to this field. As an expert in the field, maybe you can give me some advice on whether this is a good approach. Thank you.

Regards,
Yan

modeling process

process.PNG (14.78 KiB) Viewed 13061 times

[louisepb]

Hi Yan,
It sounds like you will need to look at your boundary conditions and method of extracting the CTE from your FE results.

As far as building up a library of models for single layers goes this method would not take into account nesting between the layers. When the material is compressed the layers don't stay independent as would be indicated by the stacking of individual layers. I think you would need to try it and then validate your method against experimental data. It would be interesting to see how this method works.

Best regards,
Louise

[yanchen291]

Hi Louise,

I used the PBCs and the pyton script from TexGen for extraction of material properties in all three directions. I will need to do more research to see what I need to do to get a better z-direction prediction. Please let me know if you have any suggestions.

Thank you!

Regards,
Yan