I have one more question. Why you choose discrete? I mean is it because of the cylinder not cube shape of the geometry or you get used to it? I want to know in which cases we should use discrete?
Here i had a lot of trouble, the "discrete" approach is the only one that worked reliable for cylindrical structures. There may be other ways but i dont know them
Hello Sir, I have a study of column strengthened with 2, 4 and 6 layers of unidirectional CFRP. 1)- is the number of layers refer to the number of plies, or I should divide each layer into 2 plies at least..?! 2)- when we say unidirectional, does it mean all plies are in the same direction..?!
in short: because they are derived in different ways, have different range of applications and different requirements. for more details you should probably read this book: www.wiley.com/en-us/Programming+the+Finite+Element+Method%2C+5th+Edition-p-9781119973348
Hallo Mr. Wagner, Thank you for this informative tutorial, especially the third case. I have a few questions: 1) Could you explain what and why did you add an additional rotation of 90deg in composite layup at 3:35 and in all cases? 2) What is the difference between continuum shell vs continuum solid shell, these options which you get in continuum shell case? 3) What is the scenario in the third case where you have stacked solid elements, but instead, you do it with shells and not solids? So it will be stacked shell laminate? 4) In third case, is there no need of adding cintact interactions or tie constraints between adjacent solid plies so as to restrict slipping?
1) So the 0 layer is in axial direction of the cylinder 2) Don’t know, will check it. Have wondered it myself. But never used it. 3) It’s a solid example where each layer is modeled individually, there are no shell elements in the third example only solid elements. 4) Such interaction are not need because all the layers were merged to one shell structure
@@hnrwagner Thank you for the answers. In a nutshell, the shell is prefered to solid elements as they need fewer nodes, less mesh and lesser integration points. The solid element needed 27 integration points to be as effective as shell elements. Is there any specific case where solid is prefered for composite laminate analysis? For instance, for fabric laminates? Also, did C3D8R failed to simulate accurately because it is a linear element and hence is very stiff for bending and exhibits locking under high bending?
I am working right now on a problem with a composite cylinder which has R/t = 11 & L/R = 21. For such a thick shell, the shell elements do not deliver useful results. For this problem (a bicycle frame - th-cam.com/video/Ctqa66QDLl0/w-d-xo.html), I use solid and continuum elements. Do not know why the C3D8R not work correctly. But your explanation seems sound to me.
@@nithishprabhu C3D8R has reduced integration point (1 point only) while C3D20 is full integration. C3D20 will have significant shear locking issue while C3D8R will not have shear locking issue. But C3D8R will have hourglassing issue while C3D20 don't have the hourglassing issue.
Hi sir, can I use the solid element (3rd) method to model a impact composite crush tube simulation with Abaqus Dynamic/Explicit? Also, for mesh type, can I use continuum shell elements instead of 3-D stress? Thank you sir, your videos are really helpful!
hello, am usind 2d elements (conventinal shells) for a composite laminate of 8 plies, i used cohesive interaction to model cohesive behavior between plies. however when i try to add boundary condition I can seem to be able add the load/ biundary condition to the ege of the 8 plieas, as I can't see the thickness of each ply. Hence it feels as if the load is not added equaly to the edges of all of my plies. so my question is how to add the boundary conditions of a tensile test to an 8 ply (conventional shell) composite laminates ? thanks in advance
Hello Ronald, Thank you very much for the video. I had one question. The composite layup can also be defined for a single solid element through the thickness. So can we use just one quadratic solid element through the thickness and still get comparable results with the shell elements? This way the the computational cost of using solid elements can be kept to minimum and since it is a quadratic element we are not going to have shear locking problem. Another advantage with using layered solid element is that we get to also see thickness stresses sig_33 which might be more relevant in other cases. Moreover, I would like to model buckling in composite structures post low velocity impacts (which results in subsurface delaminations/inter fiber failures and thereby reduces the bending stiffness of the laminate). Do you think that laminates with already present delmainations can be modelled using the homogenized approach with the element types you presented?
Hard to say, it depends on the problem. In one of my recent problems I had high local stresses and needed multiple elements through the thickness in order to come close to test data. So far I have not yet dealed with a delamination problem, so for the second question. I cant say.
Hello. You have added material properties for shell as lamina i.e. 6 engineering constants and for solid 9. Is it an error to provide 9 properties for shells or it doesn't make any difference?
Dear sir, I am making a 2d-shell-planar model. In the meshing, I do not get a option to select, continuum shell...because of its absence, I choose, plane stress option. But then, I get a error saying, layered properties can not be assigned to this element type. Please help.
I have just created a shell model in abaqus (Part, Shell, Planar) and could assign regular shell elements, which can be used for layered properties. Maybe you could send me your problem in detail via email. ro.wagner@tu-braunschweig.de
The thickness of the shell is based on the geometry and the relative thickness determines the thickness of the plies relative to each other. So if the relative thickness for all plies is 1 it means they all have the same thickness
@@hnrwagner so if the plies had different thickness then how would you enter the values? Should I assume one of the values to be one and the rest would be a factor of it? For example: I have 5 plies such that the first and the fifth have the same thickness. Also the second and the fourth have the same thickness but the third one has a completely different value. In this case, how would I enter the element relative thickness?
@@hnrwagner it would have been interesting to see how you did it for the brick element shell, where all 6 plies are combined into one part, as I believe you would have to assign the mesh stack direction for each ply?
@@Pishgul yes but you have to rotate the ply angles using the 2 axis, you should test it in order to see it.There may be other ways to do it. But this works best for me.
by far the best abaqus tutorial channel out there...
Thanks
Another great video, when I have to study a new topic, your videos are always vere useful, thank you
what kind of tutorial you want to see next ? comment below !
Carbon composite shell with laminated solid core.
Hello Professor Wagner, how about RVE viscoplasticity in Abaqus?
Impact on laminate composites: dif. between continuum shell and solid elements
I have one more question. Why you choose discrete? I mean is it because of the cylinder not cube shape of the geometry or you get used to it? I want to know in which cases we should use discrete?
Here i had a lot of trouble, the "discrete" approach is the only one that worked reliable for cylindrical structures. There may be other ways but i dont know them
Hello Sir, I have a study of column strengthened with 2, 4 and 6 layers of unidirectional CFRP.
1)- is the number of layers refer to the number of plies, or I should divide each layer into 2 plies at least..?!
2)- when we say unidirectional, does it mean all plies are in the same direction..?!
Hi, Wagner, thanks, very interesting video. why the results are different for different element type?
in short:
because they are derived in different ways, have different range of applications and different requirements.
for more details you should probably read this book:
www.wiley.com/en-us/Programming+the+Finite+Element+Method%2C+5th+Edition-p-9781119973348
Hallo Mr. Wagner,
Thank you for this informative tutorial, especially the third case. I have a few questions:
1) Could you explain what and why did you add an additional rotation of 90deg in composite layup at 3:35 and in all cases?
2) What is the difference between continuum shell vs continuum solid shell, these options which you get in continuum shell case?
3) What is the scenario in the third case where you have stacked solid elements, but instead, you do it with shells and not solids? So it will be stacked shell laminate?
4) In third case, is there no need of adding cintact interactions or tie constraints between adjacent solid plies so as to restrict slipping?
1) So the 0 layer is in axial direction of the cylinder
2) Don’t know, will check it. Have wondered it myself. But never used it.
3) It’s a solid example where each layer is modeled individually, there are no shell elements in the third example only solid elements.
4) Such interaction are not need because all the layers were merged to one shell structure
@@hnrwagner Thank you for the answers. In a nutshell, the shell is prefered to solid elements as they need fewer nodes, less mesh and lesser integration points. The solid element needed 27 integration points to be as effective as shell elements. Is there any specific case where solid is prefered for composite laminate analysis? For instance, for fabric laminates?
Also, did C3D8R failed to simulate accurately because it is a linear element and hence is very stiff for bending and exhibits locking under high bending?
I am working right now on a problem with a composite cylinder which has R/t = 11 & L/R = 21. For such a thick shell, the shell elements do not deliver useful results. For this problem (a bicycle frame - th-cam.com/video/Ctqa66QDLl0/w-d-xo.html), I use solid and continuum elements.
Do not know why the C3D8R not work correctly. But your explanation seems sound to me.
@@nithishprabhu C3D8R has reduced integration point (1 point only) while C3D20 is full integration. C3D20 will have significant shear locking issue while C3D8R will not have shear locking issue. But C3D8R will have hourglassing issue while C3D20 don't have the hourglassing issue.
Hi sir, can I use the solid element (3rd) method to model a impact composite crush tube simulation with Abaqus Dynamic/Explicit? Also, for mesh type, can I use continuum shell elements instead of 3-D stress? Thank you sir, your videos are really helpful!
hello,
am usind 2d elements (conventinal shells) for a composite laminate of 8 plies, i used cohesive interaction to model cohesive behavior between plies.
however when i try to add boundary condition I can seem to be able add the load/ biundary condition to the ege of the 8 plieas, as I can't see the thickness of each ply. Hence it feels as if the load is not added equaly to the edges of all of my plies.
so my question is how to add the boundary conditions of a tensile test to an 8 ply (conventional shell) composite laminates ? thanks in advance
Hello Ronald, Thank you very much for the video. I had one question. The composite layup can also be defined for a single solid element through the thickness. So can we use just one quadratic solid element through the thickness and still get comparable results with the shell elements? This way the the computational cost of using solid elements can be kept to minimum and since it is a quadratic element we are not going to have shear locking problem. Another advantage with using layered solid element is that we get to also see thickness stresses sig_33 which might be more relevant in other cases.
Moreover, I would like to model buckling in composite structures post low velocity impacts (which results in subsurface delaminations/inter fiber failures and thereby reduces the bending stiffness of the laminate). Do you think that laminates with already present delmainations can be modelled using the homogenized approach with the element types you presented?
Hard to say, it depends on the problem. In one of my recent problems I had high local stresses and needed multiple elements through the thickness in order to come close to test data.
So far I have not yet dealed with a delamination problem, so for the second question. I cant say.
Hi. What is the difference between shell and continuum shell elements?
Hello. You have added material properties for shell as lamina i.e. 6 engineering constants and for solid 9. Is it an error to provide 9 properties for shells or it doesn't make any difference?
you can also provide 9 for the 2D shells but it will probably make no difference (compared to 6) for very thin shells
Sir, also in solid case we have dynamic, implicit step but in others it was static. Would you please explain it?
cant remember anymore, but i think i used implicit for the solid elements because it leads to better convergence of the results.
Dear sir, I am making a 2d-shell-planar model. In the meshing, I do not get a option to select, continuum shell...because of its absence, I choose, plane stress option. But then, I get a error saying, layered properties can not be assigned to this element type. Please help.
I have just created a shell model in abaqus (Part, Shell, Planar) and could assign regular shell elements, which can be used for layered properties. Maybe you could send me your problem in detail via email.
ro.wagner@tu-braunschweig.de
Also, in the solid why you used 1 as a relative thickness? since in shell you used the main 0.125 mm but in solid you used relative thickness of 1?
The thickness of the shell is based on the geometry and the relative thickness determines the thickness of the plies relative to each other. So if the relative thickness for all plies is 1 it means they all have the same thickness
@@hnrwagner so if the plies had different thickness then how would you enter the values? Should I assume one of the values to be one and the rest would be a factor of it? For example: I have 5 plies such that the first and the fifth have the same thickness. Also the second and the fourth have the same thickness but the third one has a completely different value. In this case, how would I enter the element relative thickness?
@@devenmhadgut6805
1
0.75
0.5
0.75
1
@@hnrwagnerDoes it have to be these particular numbers or do the numbers have to be in proportion to my ply thickness values?
Why did you not assign a mesh stack orientation?
Dont know it seems that i did it anyway but its not in the video because the results are correct
@@hnrwagner it would have been interesting to see how you did it for the brick element shell, where all 6 plies are combined into one part, as I believe you would have to assign the mesh stack direction for each ply?
Sir, why you chose 2 in primary axis?
The primary axis 2 is choosen so that the 0 degree layer is oriented in axial direction of the cylinder
@@hnrwagner But isn't the axial direction the z axis? I just confused
@@Pishgul yes but you have to rotate the ply angles using the 2 axis, you should test it in order to see it.There may be other ways to do it. But this works best for me.
@@hnrwagner Thank you I hope you did not bother of lots of questions
@@Pishgul
that is ok, this platform is for questions and discussion. That is the spirit of science !