Boundary effect at X=L/10

1. 

   Fei Tao @ 3:26 pm on 06 Dec 2017

   Dear Prof. Yu,

   I have done the comparison of the axial stress between Swiftcomp and DNS at X = L/10. However, the result shows Swfitcomp and DNS get almost the same result. The errors are within 1%. I can not observe the boundary effect. I have attached a figure to show the comparison. I also tried to compare the axial stress at X = L/100. I observed the boundary effect at that X.

   Therefore, should we change the problem to compare the axial stress at a smaller X?

   Thank you for your time!

   Fei

   

   Report abuse

2. 

   Kunal Samel @ 6:56 pm on 06 Dec 2017

   Hi Fei! 

   how did u extract the data from ABAQUS? If I use a node path I am getting some difference between the SwiftComp and DNS results but I see the values on the contours, they are exactly the same.

   also how many elements did your model have?

    

   thanks

   Kunal

   Report abuse

3. 

   Kunal Samel @ 6:59 pm on 06 Dec 2017 — Edited @ @ 7:01 pm on 06 Dec 2017

    

    

   

   Report abuse

4. 

   Fei Tao @ 7:58 pm on 06 Dec 2017

   Hi Samel, 

   I used 4 elements for each layer. The total number of nodes are 2262897. 

   When you use the path to plot your result, did you choose the model type as undeformed? I have attached what I choose in Abaqus for your consideration. 

   For the S11 at X=L/10, is your maximum value from Swiftcomp is about 27.2 MPa?

   Best,

   Fei

   

   Report abuse

5. 

   Kunal Samel @ 10:14 pm on 06 Dec 2017

   thanks for your reply Fei, yes my max value from swiftcomp is 27.2MPa.

   I have also chosen undeformed in abaqus. Which method did u use to model the layers?

   I entered orthotropic engineering constants and assigned two different material coordinate systems to the top and bottom layers. Did you do this differently?

    

   thanks

   Kunal

   Report abuse

6. 

   Fei Tao @ 10:32 pm on 06 Dec 2017

   I think I find your problem. In your method, what you get is the stress in the material coordinate. You can have a try to transform your stress into the global coordinate.

   What I did was, I transformed the material from its material coordinate into the global coordinate first. what I got was a monoclinic material. Then, I used anisotropic material in Abaqus. In this method, the stress I get is under the global coordinate.

   Report abuse

7. 

   Kunal Samel @ 10:42 pm on 06 Dec 2017

   thanks! I will attempt it with the anisotropic material tomorrow.

   Report abuse

8. 

   Fei Tao @ 10:44 pm on 06 Dec 2017

   You are welcome! Let me know if you have any problem or get any difference. 

   Report abuse

9. 

   Wenbin Yu @ 5:44 pm on 06 Dec 2017 — Edited @ @ 5:44 pm on 06 Dec 2017

   You and others are welcome to exam at some place closer to the boundary such as 0.01 L. However, it is not a requirement. ---- Emailed forum response from wenbinyu@purdue.edu

   Report abuse

10. 

    Fei Tao @ 12:47 am on 07 Dec 2017

    Dear Prof. Yu,

    Thank you for the clarification.

    Best,

    Fei

    Report abuse

11. 

    Kunal Samel @ 1:57 pm on 07 Dec 2017

    thanks Fei! I think u were right, I used pre-rotated stiffness matrices and I got a match with the Swiftcomp results

    Report abuse

12. 

    Fei Tao @ 9:22 am on 07 Dec 2017 — Edited @ @ 9:22 am on 07 Dec 2017

    You are welcome! I am glad that helped. On Thu, Dec 7, 2017 at 13:57 Kunal Samel @ cdmHUB - The Composites HUB < support@cdmhub.org> wrote: ---- Emailed forum response from cntaofei@gmail.com

    Report abuse