I can used this expression S = Df*sqrt( (N*pi) / (4*Vf) ). For FOUND DIMENSION OF MATRIX .AND I CAN USED THIS FORMULE FOR CONCRETE AND POLYPROPYLENE FIBER. THANK YOU SSO MUCH DR
Yes, you have to be careful. Lets assume, given that you know the diameter, Df of fibre, volume fraction of fibre, Vf and number of fibre desired, and the Length , L and with, W is desired (for square L = W), then you can use your equation where S = L = W. I think you can also follow the formula I used in this video at earlier part of the video.
Thank you very much, Dr. Michael, for these good lectures. I have a question. How do I determine the volumetric fraction of two materials in the case of preparing a slide or layer consisting of polymeric material and fibers at rates of 30% and 70% in the Abaqus program? Is there an explanatory video available?
Hello, thanks for your comment. I am a bit unsure about what the question is about. If you want to determine the volume fraction of two materials (composites), you can do so by finding the weight contribution of each of the phases that make up the system. The weights of say E-glass fibre and the weight of the matrix, say Epoxy. With those weights, you define the weight fractions. You also need to know the densities of the constituents that make up the composite. If you want to see how to make the conversions, a quick google search will help you as you can see from here: sites.google.com/site/compositematerialsdesign/home/weight-and-volume-fractions. I am not sure if this is what you are looking for. Good luck with the research.
Hello Doc, if I want to want to create a downward compression in the Y axis ONLY, do I need all THREE constraint equations, TWO or ONLY ONE? and which? Thank you
Hello, what you need to do for a RVE-based analysis as I showed in this video is to revert the tensile load (say in the X-axis) to simulate compression. That means you will have one constraint equation for the X-tensile load which in your case will be X-compression load. The other 3 Boundary conditions on three orthogonal surfaces is there to simulate a quadrant of the macroscale model simulation.
thank you doc, but now i am wondering about the interface (cohesive element) properties. How can i obtain those properties, like what method do we use, experimental procedure or anything you suggest.
Hello, it is always possible to determine the interface properties but not very straightforward. For a composite material, I had a colleague in the past explore this which you can pick up in this paper: doi.org/10.1016/j.compscitech.2009.04.021 It was a carefully designed set of experiment that explores the mechanical behaviour of the interface region by loading it to failure. This is one way. Some others have used a nano-indentation method to probe those regions. The challenge with interfaces is that they are usually a broad region around the fibre whose mechanical properties are neither those of the fibre/yarn or the matrix media. As a result, it is always subjective. In this modelling here, I assume some affected region properties that matches/compares with those of the matrix rather than the fibre. It is technically right but you need to identify or execute the right kind of experiment to generate comparable data for it. More importantly, you need to incorporate a failure model with it as failure of the interfaces always lead to failure of the composite.
Thanks, Doc. When I tried to create the 2nd ref. point at 12:18, it's asking me to replace the first one, which is clearly not it. How do I solve this please?
I have always wondered (in the 3rd and 4th parts of this video) why you calculate your stress and strain by dividing with AREA and LENGHT respectively instead of getting the stress and strain direct from Abaqus?
Hello, I can easily extract stress and strain directly from the simulations but this is not always the best approach due to localization effects. If I choose three different elements at different locations in the model, due to the heterogeneous effect of the microstructure, you will be reproducing different stresses. So, to eliminate this problem, a true homogenized stress stage (for a heterogeneous media) needs to be extracted. This means replicating what you do in real experiments but extracting using a load cell, the force experienced by the material and using a strain guage (or even cross-head displacement) the displacement experienced by the deforming material. Then subsequent to the experiment, you can then operate on the force-displacement plot to recover the appropriate stress-strain plot. I will suggest you watch this video of mine where I delved deeper into these ideas and hopefully this will convince you why I tend to follow the approach you have referred to: th-cam.com/video/gCQzzF8Few8/w-d-xo.htmlsi=fNOB7ZxP8K1ycGfU
Hello Doc. I've always had a deep question about the 3D UD-RVE model. Do you create a cut or space in the matrix where the fibre will stay or you just position the fibre in a solid matrix without considering spatial interference in terms of geometry?
There are two ways: 1. If you do not have an interface layer, then you do not need to create the hole. The boolean merge operation in ABAQUS allows for good merging without need for making a hole. However, it makes for node-to-node coupling. 2. If you do have an interface, fhen you need to make a hole. Hope this helps.
Hello Dr. I was following your video in practice. At 11:35 when you clicked on Query, your dialogue box came out full and complete. Mine is cut. I can't move or see the rest of it up. Please help
Thank you Doc, for such an insightful video. How do we obtain material properties when performing such micromechanical analysis, and how do we compare with the experimental results?
Hello @Johnson, normally with this sort of micromechanical analysis, it is better to obtain properties of the individual constituents. For example, test E-glass and Polypropylene (matrix) separately and then use those in the model. The fibre is elastic here so the manufacturer-quoted properties are typically alright. For the matrix, depending on how it is processed, the properties can vary. So, most at times, people will test the matrix under the conditions in which the composite is processed. For example, if you use prepregs that is compression moulded, then you have to compression mould the matrix to test it so you are sure you are testing under right processing conditions. If you use an autoclave to manufacture the composite, then you have to autoclave-manufacture the matrix. What is a bit more challenging to get is the properties of the interface as that is not always definite. Most at times, it has a gradation of properties varying with the stiffest/strongest closest to the fibre and as you move away from it, it changes to lesser values, approaching those of the matrix. For most composites that are brittle, at room temperature, the failure is brittle and the interface is best considered as a brittle media (so the fracture mechanics approach works well here). However, for thermoplastic ductile failure composites, especially at elevated temperatures, the approach I recommend here is accepted but getting its properties is hard. I think with comparing experimental results, you ahve to do uniaxial (tensile and compression, separately), then shear tests for all three planes. These will then be compared with predictions from this microscale model. You will have to homogenize though. I will make part 2 and 3 of this video to explain how you get effective properties which then you can compare easily with any typical experiment.
@@MichaelOkereke Thank you Sir for the reply. But the experiment is done at standard scale whereas the FEM is micro. How is the size difference reconciled?
Hello @Johnson, this is precisely why we do homogenization. Although the constituents are at the microscale, by invoking some kind of computation or asymptotic homogenization, you can generate properties that are comparable with experiments at the macroscale. The mathematics of homogenization makes it possible. This is why microscale approaches as this work, because modellers know that when they homogenize their model, they can compare it with real life experiments. It has its limitations as you will not be able to run structural tests using such microscale approaches. For example, you cannot use such microscale model (in its classic form) to run plate bending or three point bending tests. They tend to work for simpler loading studies such as uniaxial and shear loads. As load types becomes complicated, simple homogenization might not work well. You may have to read around the principles of homogenization for multiscale simulations and try to follow the mathematics behind it, if you want to truly understand these things.
@@johnsonezenwankwo3497 I also have that question. It's good to see your comment. By the way, could you tell us what CDMHUB is to perform homogenisation?
Calculation of the Vf with the given formula gives a different result than 35%. (pi * D^2)/(4*120*100) = 29%. If I use 100 in place of 120, then I get 35% as in the video.
Hello, the interface is a part of composite modelling. It is an affected region between the fibre and the matrix. It is always necessary and it contributes to failure in composite materials. Some modelling will negelect it with reasonable results. For your woven textiles, it is not always considered because its effect is seen most at the fibre-matrix-interface level. A woven composite modelling is at a yarn level where properties of the yarn results from volume-averaged response of the fibre-matrix-response. I will not bother about it for mesoscale woven textiles composites modelling but if you are conducting a microscale modelling of unidirectional composites, then you do need to consider it.
Hello, this is quite simple: It is variable = constant. The variable can be displacement, velocity, temperature. The constant here is a fixed number say Displacement = 20 mm. It is that straightforward.
![CAMPปลิ้น | EP.83[2/2] ประสบการณ์ขนหัวลุกจากตำนานวงการผี!](http://i.ytimg.com/vi/T_kFecQw7Ds/mqdefault.jpg)
Thank you so much Dr Michael, waiting for more videos on UD composites
More to come @Alino! Keep coming back!
Great video! Thank you for the series of Abaqus simulation
Glad you like them @Eslingena! Thanks.
Great Video, Thank you Dr. Micheal
Thanks @Walaa
I can used this expression S = Df*sqrt( (N*pi) / (4*Vf) ). For FOUND DIMENSION OF MATRIX .AND I CAN USED THIS FORMULE FOR CONCRETE AND POLYPROPYLENE FIBER. THANK YOU SSO MUCH DR
Yes, you have to be careful. Lets assume, given that you know the diameter, Df of fibre, volume fraction of fibre, Vf and number of fibre desired, and the Length , L and with, W is desired (for square L = W), then you can use your equation where S = L = W. I think you can also follow the formula I used in this video at earlier part of the video.
Thank you very much, Dr. Michael, for these good lectures. I have a question. How do I determine the volumetric fraction of two materials in the case of preparing a slide or layer consisting of polymeric material and fibers at rates of 30% and 70% in the Abaqus program? Is there an explanatory video available?
Hello, thanks for your comment. I am a bit unsure about what the question is about. If you want to determine the volume fraction of two materials (composites), you can do so by finding the weight contribution of each of the phases that make up the system. The weights of say E-glass fibre and the weight of the matrix, say Epoxy. With those weights, you define the weight fractions. You also need to know the densities of the constituents that make up the composite. If you want to see how to make the conversions, a quick google search will help you as you can see from here: sites.google.com/site/compositematerialsdesign/home/weight-and-volume-fractions. I am not sure if this is what you are looking for. Good luck with the research.
Hello Doc, if I want to want to create a downward compression in the Y axis ONLY, do I need all THREE constraint equations, TWO or ONLY ONE? and which? Thank you
Hello, what you need to do for a RVE-based analysis as I showed in this video is to revert the tensile load (say in the X-axis) to simulate compression. That means you will have one constraint equation for the X-tensile load which in your case will be X-compression load. The other 3 Boundary conditions on three orthogonal surfaces is there to simulate a quadrant of the macroscale model simulation.
thank you doc, but now i am wondering about the interface (cohesive element) properties. How can i obtain those properties, like what method do we use, experimental procedure or anything you suggest.
Hello, it is always possible to determine the interface properties but not very straightforward. For a composite material, I had a colleague in the past explore this which you can pick up in this paper: doi.org/10.1016/j.compscitech.2009.04.021
It was a carefully designed set of experiment that explores the mechanical behaviour of the interface region by loading it to failure. This is one way. Some others have used a nano-indentation method to probe those regions. The challenge with interfaces is that they are usually a broad region around the fibre whose mechanical properties are neither those of the fibre/yarn or the matrix media. As a result, it is always subjective. In this modelling here, I assume some affected region properties that matches/compares with those of the matrix rather than the fibre. It is technically right but you need to identify or execute the right kind of experiment to generate comparable data for it. More importantly, you need to incorporate a failure model with it as failure of the interfaces always lead to failure of the composite.
Thanks, Doc. When I tried to create the 2nd ref. point at 12:18, it's asking me to replace the first one, which is clearly not it. How do I solve this please?
Oh Doc, I forgot to create an assembly. It's sorted too. Thanks
Yes, you do it in assembly module not in part module. Glad it is sorted.
I have always wondered (in the 3rd and 4th parts of this video) why you calculate your stress and strain by dividing with AREA and LENGHT respectively instead of getting the stress and strain direct from Abaqus?
Hello, I can easily extract stress and strain directly from the simulations but this is not always the best approach due to localization effects. If I choose three different elements at different locations in the model, due to the heterogeneous effect of the microstructure, you will be reproducing different stresses. So, to eliminate this problem, a true homogenized stress stage (for a heterogeneous media) needs to be extracted.
This means replicating what you do in real experiments but extracting using a load cell, the force experienced by the material and using a strain guage (or even cross-head displacement) the displacement experienced by the deforming material. Then subsequent to the experiment, you can then operate on the force-displacement plot to recover the appropriate stress-strain plot.
I will suggest you watch this video of mine where I delved deeper into these ideas and hopefully this will convince you why I tend to follow the approach you have referred to: th-cam.com/video/gCQzzF8Few8/w-d-xo.htmlsi=fNOB7ZxP8K1ycGfU
Hello Doc. I've always had a deep question about the 3D UD-RVE model. Do you create a cut or space in the matrix where the fibre will stay or you just position the fibre in a solid matrix without considering spatial interference in terms of geometry?
There are two ways:
1. If you do not have an interface layer, then you do not need to create the hole. The boolean merge operation in ABAQUS allows for good merging without need for making a hole. However, it makes for node-to-node coupling.
2. If you do have an interface, fhen you need to make a hole.
Hope this helps.
Hello Dr. I was following your video in practice. At 11:35 when you clicked on Query, your dialogue box came out full and complete. Mine is cut. I can't move or see the rest of it up. Please help
I sorted this out. Thanks
Ok I can see you sorted it out.
Thank you Doc, for such an insightful video. How do we obtain material properties when performing such micromechanical analysis, and how do we compare with the experimental results?
Hello @Johnson, normally with this sort of micromechanical analysis, it is better to obtain properties of the individual constituents. For example, test E-glass and Polypropylene (matrix) separately and then use those in the model. The fibre is elastic here so the manufacturer-quoted properties are typically alright. For the matrix, depending on how it is processed, the properties can vary. So, most at times, people will test the matrix under the conditions in which the composite is processed. For example, if you use prepregs that is compression moulded, then you have to compression mould the matrix to test it so you are sure you are testing under right processing conditions. If you use an autoclave to manufacture the composite, then you have to autoclave-manufacture the matrix.
What is a bit more challenging to get is the properties of the interface as that is not always definite. Most at times, it has a gradation of properties varying with the stiffest/strongest closest to the fibre and as you move away from it, it changes to lesser values, approaching those of the matrix. For most composites that are brittle, at room temperature, the failure is brittle and the interface is best considered as a brittle media (so the fracture mechanics approach works well here). However, for thermoplastic ductile failure composites, especially at elevated temperatures, the approach I recommend here is accepted but getting its properties is hard.
I think with comparing experimental results, you ahve to do uniaxial (tensile and compression, separately), then shear tests for all three planes. These will then be compared with predictions from this microscale model. You will have to homogenize though.
I will make part 2 and 3 of this video to explain how you get effective properties which then you can compare easily with any typical experiment.
@@MichaelOkereke Thank you Sir for the reply. But the experiment is done at standard scale whereas the FEM is micro. How is the size difference reconciled?
Hello @Johnson, this is precisely why we do homogenization. Although the constituents are at the microscale, by invoking some kind of computation or asymptotic homogenization, you can generate properties that are comparable with experiments at the macroscale. The mathematics of homogenization makes it possible. This is why microscale approaches as this work, because modellers know that when they homogenize their model, they can compare it with real life experiments.
It has its limitations as you will not be able to run structural tests using such microscale approaches. For example, you cannot use such microscale model (in its classic form) to run plate bending or three point bending tests. They tend to work for simpler loading studies such as uniaxial and shear loads. As load types becomes complicated, simple homogenization might not work well.
You may have to read around the principles of homogenization for multiscale simulations and try to follow the mathematics behind it, if you want to truly understand these things.
@@MichaelOkereke Thanks a lot, I am using CDMHUB to perform homogenisation.
@@johnsonezenwankwo3497 I also have that question. It's good to see your comment. By the way, could you tell us what CDMHUB is to perform homogenisation?
Calculation of the Vf with the given formula gives a different result than 35%. (pi * D^2)/(4*120*100) = 29%. If I use 100 in place of 120, then I get 35% as in the video.
You may be right, @Johnson. You just have to determine the right number of fibre values for a combination of RVE lengths and fibre diameters.
Is there any specific reason to make interface?
And is it necessary to make interface, in case of making woven composite for mechanical properties.
Hello, the interface is a part of composite modelling. It is an affected region between the fibre and the matrix. It is always necessary and it contributes to failure in composite materials. Some modelling will negelect it with reasonable results. For your woven textiles, it is not always considered because its effect is seen most at the fibre-matrix-interface level. A woven composite modelling is at a yarn level where properties of the yarn results from volume-averaged response of the fibre-matrix-response. I will not bother about it for mesoscale woven textiles composites modelling but if you are conducting a microscale modelling of unidirectional composites, then you do need to consider it.
What is the mathematics notation of the Dirichlet BC?
Hello, this is quite simple: It is variable = constant. The variable can be displacement, velocity, temperature. The constant here is a fixed number say Displacement = 20 mm. It is that straightforward.
@@MichaelOkereke Thank you, Dr
Thank you! Please I want to know when we use dimensions in nanometers what should be the unit of Young's modulus ?
This is the video to help: th-cam.com/video/YeCBDkjEmDU/w-d-xo.html
@@MichaelOkereke in your video you used micrometer so the Young's modulus should be mutiplied by 10^-3 not 10^9.
Please confirm. Thanks
See the end where I justified the way I treat these issues