Your video is really great, and professional! Not many researchers share their professional knowledge on RUclips. I nearly watched every of your videos, extremely helpful! Thanks a lot!
Dear Dr Michael I'm kind of confused about the boundary conditions. In case I(Z tension), according to another reference book, BCs are XSYMM on XFront&XBack, YSYMM on YTop&YBottom, ZSYMM on ZBack and Z displacement on ZFront, which are a bit different from those used in this video. The results are also slightly different. Could you please explain this and tell what BCs should be used in similar situations. Thanks!
Hi, I am not sure of the reference you mentioned here. The approach shown here is quite standard for an RVE setup for uniaxial loading. The 3 faces were constrained in one axis so that they can contract in the other directions. It is true uniaxial loading. I will expect the results to be different if the boundary conditions are not the same. Please look at a few other publications and see what's going on.
Good question. I wanted to do what most users can do without relying on a dedicated script. To undertake correctly PBC on 3D domain as this, you would often need resources not implemented yet in ABAQUS. This will be a python script that can read the surface nodes, their positions and then using a kinematic coupling approach to link directly opposite and corresponding faces. I know most users would have not have this, but would want to still undertake this sort of study. So, in effect, the best was was to show what is possible using resources within ABAQUS. I hope the explanation makes sense to you.
@@MichaelOkereke Thanks for your reply!!! I am a beginner in this area. My following question is: Based on your reply, does that mean we can use these BCs you proposed to replace the periodic BCs? Thanks
Hi Dr. Michael! I found really useful your video, I am doing my final degree project about RVE modelling and testing different materials. I followed all your videos and it helped me a lot. I don´t know if you could help me and say me what are the constraints and the BC for shear in plane XZ. That is what I have left. Thank you very much!
Glad it was helpful! Thanks for your kind comments. I have made an extensive video where I showed how to impose the shear loads (Pure shear). You can check out the whole video but if you start from the 13th minute, you will see description on imposing pure XZ shear through this link: ruclips.net/video/uzUQz4NPkEs/видео.htmlsi=IWr6puvLgTwuEZE9&t=792. Please let me know if you have any other queries. You can check other videos on shear loads, they might not be for UD composites as here but the implementation is still correct irrespective of the material under investigation.
Hello Dr Michael, i was wondering if you could please make a video about 3-point bending SENB testing following ASTM standard E1820. Regards, Aqsa UoS,England
Hello dear doctor. In the student version, I am trying to create a 3D dentin model consisting of hollow tubules of variable diameter, that is, hollow fibers. While doing this, I follow the same logic you showed in this application. I mount the hollow fibers into the matrix consisting of hollow holes. But I'm having serious problems creating the mesh and I keep getting error returns from the program. As you know, the student version only allows 1000 knot points. I would be very grateful if you could make a suggestion about this.
Hello, unfortunately there is no way getting around the 1000 nodes limit for the student license of ABAQUS. You will have to find a full license to do so. Alternatively, you may then try using a 2D model of dentin and that might not require so much meshing. Since your 3D dentin model will have a lot of irregular geometries, that will lead to high mesh density hence you will easily exceed the 1000 nodal limit of the student license ABAQUS. Sorry, I could not be any more helpful.
Hello @Anis, thanks for the query. I am not sure about your question. There are many ways of applying displacement loads on structures in a micromechanical model. I chose this type as it helps me extract the history outputs by tracking displacement and reaction forces variables in the reference nodes. If I use a direct application of displacement on the face, then my history variables will be too much and so unable to generate the stress-strain graphs. As regards comparing it with some other type of loading, what do you have in mind. I will need some further explanations.
Hi Mr. Okereke, I'm enjoying your classes. If you don't mind, I have a question: Do we need to create a symmetric mesh to achieve these results? Did you intentionally use a symmetric mesh in all of these examples, right?
Hello, I assume what you mean by symmetric mesh is meshes that match between the corresponding parallel faces. If this is so, then this is not a requirement for this sort of work. Symmetric meshes tend to be needed when working with Periodic Boundary Conditions. I am not using that type of boundary condition so this is not the case. I hope it makes sense.
@@MichaelOkereke thanks for replying my message. I believe that your channel is truly an oasis of good discussion and content. Regarding to your answer, could you please clarify just one more question? Is it possible to obtain trustworthy results using Abaqus standard BCs - instead of PBCs and a symmetric mesh? Or is the latter condition necessary for achieving good results?
Hello, thanks for your kind comments about the channel. I want it to be exactly that, an oasis of good discussion where I get challenged on these questions and hopefully, I am able to provide some kind of guidance to my audience. Regarding your question, of course, yes you can obtain excellent and trustworthy answers using ABAQUS standard Boundary conditions. Those are mainly the Dirichlet or Neumann Boundary conditions and they work quite well. Before the mathematics of PBCs were advanced, most analysis were based on these standard BCs. So, please feel confident to use them for your problem. In fact, for UD composites, many years ago I did a study to compare effective properties for PBC-based analysis and Dirichlet analysis and the results were comparable. You can see the paper here:doi.org/10.1016/j.commatsci.2012.12.036 The requirement for PBCs is important if you are working on a small RVE where the constituents have distinct constitutive behaviours and you have to capture those clearly. It gives a better result as the effect of boundary conditions on homogenized results is minimal where PBCs are involved compared to the other types. However, there is still confidence in the other approach. If you are not able to implement PBCs for unsymmetric meshes (as this is quite challenging, I am working on a code that can help), then do not worry, it is okay and trustworthy going with ABAQUS standard BCs.
Hi Dr Michael Thank you very much for providing this fruitful video and clear explanation. I have one question related to the case III which is compression. My question is why you set FixedZBack, RollerYBottom, and XBackRoller for the BC instead of only set the FixedYBottom and give the load in Y direction because this is a compression test? If my assumption is wrong/ not a right standard, could you please provide me the source of a standard uniaxial loading condition? Thank you!
Hello, this is just a different way of imposing loads compared to what you suggested. It is often better to do it this way, as this is a micromechanical study based on an RVE. The chosen RVE represents one-eighth of the quadrant of the actual virtual domain. To simulate effective such an RVE, it is customary to impose loads on 3 free surfaces and attach dirichlet boundary conditions on the 3 other remaining surfaces. I hope the above makes sense.
Thank you for your response@@MichaelOkereke , thats make sense. One more, I've done exactly like you did in this tutorial, but I encounter the "too many attempts in this increment", do you have any suggestion for me?
Thanks Sir, just a doubt, i really need to know how it will effect if instead of creating a ref. point and constrained it with Equation, i directly create the set of the deforming plane, apply displacement on the very set and check the history output for it.
Thanks for the query. If you do as suggested i.e. extract history outputs (say S11 and E11) from the nodal set on say XFront, you will still get a result without the reference point approach that I used here. The limitation here is that it will be only giving you the history outputs on those XFront sets which is not representative of the full model. It will not be an accurate stress-strain plot. This is because of the heterogeneous nature of the virtual domain. The best approach is to extract all the stress and strain for all elements in the domain and homogenize over them. This can work but the problem is that for a heavily meshed domain under consideration, you will have to write a lot of history variables to memory of the computer. ABAQUS typically runs out of history output memories once you have more than 10,000 history data. This is a problem and this is why I used this method as it requires only
@@MichaelOkereke Thanks for the clarification, just for confirmation, If we constrain a surface (say XFront) to a ref- point, would it be correct to say that we are constraining the nodes of that surface to the ref-point. Does it also means that we are lowering down the history output data for every nodes on the surface to only one node i.e. for ref-Point. If so, then the history output for the ref-point as explained in the Video is for the XFront nodes or for the whole model.
Hello @Haowei Huang, thanks for your interest in the channel. 1. I don't have any contacting parts in the model so no contacts were specified. In some models you may introduce contacts between fibres and matrix but this is not standard. What you may want to do is introduce a cohesive zone between fibre and matrix but that's a different problem. 2. The boundary conditions are Dirichlet or displacement boundary conditions. I haven't used PBC here. I hope the above is okay with you. Let me know.
Hello, I really enjoyed your video and I'd like to thank you for the clarity of the exchanges. Could you tell me how you came up with the coordinates of the points to represent the fibres location in the matrix?
Hello, I used a code I developed called MontCarlGen2D to create that using the Monte Carlo method of random positioning of fibres within an RVE window.
Hello @leperroquet9045, this code is certainly available - just check it out here: www.cmvideos.org/shop/, which is the company website. Let me know if there is any difficulty in getting the code.
Thanks for the kind comment. I am not sure about the problem you describe. It is quite complex and without better understanding it, it is difficult to be able to give any meaningful suggestion. Filament winded flywheel suggests it could be a textile composite or maybe some kind of angle-ply laminated composite system. Sorry, I am not able to help with such limited information. Good luck.
sir, while calculating stress value you multiply the result by 10^-6. But since as per our unit was N/(micro meter)^2 So to convert this into Mega pascal I think we need to multiply by 10^6. So can you give me clarification over this !
Here is a video that explains my philosophy of dealing with units in ABAQUS, especially for micrometre lengthscales: ruclips.net/video/YeCBDkjEmDU/видео.html
Hi Humberto, thanks for your interest in the channel. I have a MATLAB script that I developed for the Monte Carlo procedure. The principle of it is quite well known. I will consider making a video about this in future maybe to highlight the process and show how my MATLAB script works. Is that something you are interested in?
Hello eng, thanks for your interest in the channel. Finding fibre coordinates randomly is, as you rightly pointed out, best done using the Monte Carlo Method. I have already done this in a publication: www.sciencedirect.com/science/article/pii/S0927025612007744 . If you look at section 2.1 on *Geometric modelling* then, you can gain some insights into the thinking about the Monte Carlo Method. I will be making a series of videos on the Monte Carlo method, as this is a frequently requested video. Please keep coming back to the channel to see when the video becomes available.
Sir, I'm facing another problem right now. I've several questions: 1. I've taken the length of the RVE in meter (m). What will be the unit of Stress before multiplying by 1 * e^ -6 as you've done in this video at 47:50? 2. What will be the unit for the Force (F) and the Displacement (D) in F-D curve whenever I'm taking the length, width & height in meter unit? (In this video u can find it at 47:04) 3. During stress calculation you've multiplied 100*100. Are they width*height or, length*width or, what? (In this video u can find it at 47:28) 4. You've taken the length, height & width in micrometer but in this video at 47:47 , during stress calculation you've calculated the stress as Pascal and then you've converted it to Megapascal. Was it correct whenever you calculated the stress unit as Pascal? I'm a liitle bit confused about that because Nm^-2 represents Pascal but in your case the measurements were in micrometer! Kindly help me by answering all 4 questions sir.
It is a different way of extracting stress-strain plots but based on a reference node which is kinematically connected to a surface. I explained this in more details in this video. It is the third approach: ruclips.net/video/gCQzzF8Few8/видео.htmlsi=LICNUGB7j2SmayDJ
Dear Dr. Michael , Thanks for your helpful video , its really great job , however , would you please advise how did you generate the coordinate data of fibers by Montecarlo Algorithm in periodic style , Which software you had used , I would appreciate your help if you have a tutorial video or text explain this issue.
Hello Mortda, thanks for your interest in the channel. I used a monte carlo implementation to do this. See this video for the Monte Carlo method: ruclips.net/video/e6LK7YB3d_I/видео.html If you want to get the actual software, then please wait for future videos when I release the code and video of how to use the software.
Hello @Ha-Linh Nguyen, the concept of dispersion curves arises during dynamic assessment behave of heterogeneous materials or multilayered composites. This is because the differences in effective properties (Modulus and strengths) inside the heterogeneous materials leads to different propagation of wave dynamics within the heterogeneous material. As continuum has to be maintained, such differences in propagation will lead to dispersions of the wave as it hits different materials in the medium. There are a bunch of equations proposed by C. T. Sun in *Continuum Theory for Laminated Medium, Journal of Applied Mechanics, 1968* which helps in tracking this dispersion curves and plotting them. I believe this is the sort of thing you are interested in. Having said all these, I have actually never done this sort of modelling and so will not be able to help you with this query. I will have to research into it, understand it better, implement it before I am able to make a video about it. For now, this is all the help I can offer and good luck with the research.
Sir, in case of load application, if I use concentrated force for example 1200 N, the force displacement curve cannot be generated. Is it only generated when we apply the percentage strain? My another question is how can we convert the concentrated force for example, 1200 N to percentage strain?
Hello @Ashfakur, Your observations are right. If you use force, it will be overwritten by the underlying kinematic constraint equations involving displacements only. I think this is what's going on here. On converting from force to displacement, you can use the force displacement equation for a uniaxially loaded system where the equation to use is: Displacement, D = FL/AE Where F = Concentrated force, L Edge length of RVE, A = Surface area and E = Youngs Modulus. Good luck with this.
Sir, if there presents two type of fibers and one matrix in the system , what will be the value of E ? Which Young Modulus should we take as the value of E? Or, how can we calculate the value of E?
Hello @Adarsh Shah, no, I did not use the volume averaging method. This was based on extracting reaction forces on a face and dividing the force with the area of the edge. This is similar to what is called the direct macro-micro approach but adapted for cases where periodic boundary conditions have not been implemented. I hope you understand.
Will it work if fibers are modeled in such a way that they touch each other, In other works each circles we draw touch/intersect each other at single point(externally)
I certainly think it will, but the right mechanical behaviour will involve having some kind of matrix media (no matter how thin in-between the two closest fibres). If there is a requirement for fibres touching, then I will suggest you incorporate some kind of contact mechanics which will mean during loading, when fibres touch, there is effective load transfer between the interacting fibres. This is technically the right approach but it is computationally very expensive as your contact will be All contacts which will cause the simulation to take too long. Alternatively, you can bind the fibres to the matrices in node-to-node coupling which might mean you do not have to deal with contact effects. It is my long-winded way of answering the question to say, generally it will work with contact mechanics between touching fibres inclue but no, as this is a very computationally expensive way to study such composites.
Hello @Nag Mk, thanks for your interest in the channel. Three-point bending test is better studied using a macrostructural model rather than an RVE model as I did with this UD composite example. Why would you want a three-point bending test on an RVE? This is possible, but I am not sure you learn a lot from such a model but there could be a research objective, so please share that with the channel.
Great video and illustration. Could you make a video showing the orthogonal cutting of this composite? Could be of great help. There is very little detailed work about it on this platform. Thank you
Hello Haris, thanks for your interest in the channel. I have not actually modelled orthogonal cutting of composites but that should be an interesting little project to tackle. I will consider it and maybe in future make a video on it. Thanks for the suggestion.
Sir, first of all I would like to thank you to your video. I wanna ask a question for you. Now, I am working on blast loading responce of a plate (only plate, not any fiber). I did some analysis via Abaqus. But I have a error. My error was; Excessive incremental rotation of the elements in element set ErrElemExcessIncrementalRotation-Step1. There is only one element with excessive rotations. After I took this error, I increased my mesh elements number. But after this, I took another error. Now my error is this; Element 21, node-1: the distance between the node and the source of the explosion is smaller than the charge radius (5.30000E-02). The CONWEP model is not valid at this distance. What can you say about my error? How can I fix this error? Best regards
Hello @Ali, great that you were able to solve the first problem. I think the second problem is due to the location of the charge that you are using in the CONWEP system. I will suggest you increase it and check to see if the second error is still happening. For example, it is saying the charge radius is 0.053 (I assume that is meters), hence you may have to increase it to 0.150 and see if that improves it. Do not that the further away the charge is from the plate (i.e the location of the explosive), the lesser the impact the charge/explosive will have on the plate so you cannot just increase it arbitrarily. A rule I normally will follow is to start with a distance as far back as the half the edge length of square plate and then possibly reduce or increase it depending on the deflection you see on the plate. If you have an experimental data you are working with, then no guess work here, simply use the distace specified in the experiment. Good luck. Here is a paper that might help you: doi.org/10.1016/j.ijimpeng.2011.02.005
Hello Adedotun Banjo, definitely its part of my plan to show how you can use this method for a particulate composite. I will make a video like this. Please watch out for this video.
Hello @Haowei Huang, thanks for your interest in the channel. I have not used any contact in this model. The fibre is tied permanently to the matrix at the interface. In future, I will make a video where I include an interphase which is an improvement in this video.
@@MichaelOkereke that would be awesome! Looking forward to your new video. I am doing the comparison at the moment, with “hard contact”, “soft contact”, cohesive contact and without contact. The behaviours are all different. So without any contact case. In linear stage, can I assume that the contact stiffness is actually the matrix modulus? Does this make sense?
Hello Karan, thanks for your interest in the video. Actually, the 10% is chosen randomly, just to get an idea about how the model will respond. Because I am doing a microscale analysis, where the matrix is explicitly modelled as an elastoplastic material (without damage included), then it is bound to show significant plastic deformation. It makes sense then to go for a higher displacement of 10% as against about 3% for most composites. To do the later however, we need to have a damage model included in the simulation so that we can predict experimentally recoverable data. I hope the explanation makes sense and that is what you are asking about.
@@MichaelOkereke Thanks a lot for such a good explanation, I have one more query regarding meshing in Abaqus for complex geometry, as I saw you are good in Abaqus, if okay can I mail you the geometry and ask my query? I am sure solving this complex part will be a engaging )
@@karangupta120 - thanks for your interest. I will suggest you subscribe to my CM Videos Insider Group (here cmvideos.ck.page/16a99bd0e3) and we can take the discussion further from there.
Hello Subrat, thanks for your interest in the channel. No, I have not used a periodic boundary condition for this video. I simply used a Dirichlet boundary condition but I will consider making some videos in future where periodic boundary conditions will be used. As you may know, a periodic boundary condition is always preferred for such micromechanical analysis but its implementation within common FEA solvers is not easy. Is this something you are interested in Subrat?
hello doctor, how are you? i did exactly according to the movie you uploaded , but unfortunately when i try to make the first submission in case1 i get a warning message that says : the following parts have some elements without any section assigned : UdComposite. thank you for this problem i will fix it because this is the subject of my senior dissertation . thank you.
Very insightful video sir. Sir I have a query. What modification do we need to make in HISTORY OUTPUT REQUEST so that we can calculate shear modulus from shear forces in the 4th and 5th Case. And finally what should we choose in place of Reaction force and Displacement in the ODB history output to find shear modulus?
Hello @Sandy, thanks for your interest in the channel. For shear modulus you simply have to follow the same approach but of course change the boundary conditions to make sure you are simulating shear responses in I believe you mean YZ and XZ planes. For example if it is Shear YZ you are simulating, then you are simulating shear on the plane where Y-axis is the outward pointing normal and the shear forces are oriented along the Z-axis. Therefore, for this you need to extract shear force, RF3 (z-axis reaction force) and divide this by the area of the YZ plane where the shear force acts. This will give you a shear behaviour. I have made videos where such shear cases were investigated. Here are some of them: RVE Modelling of BD composites (ruclips.net/video/zHUC0cVkdVM/видео.html) and Hybrid Composites - an RVE Modelling Abaqus tutorial (ruclips.net/video/1rdbKix_biw/видео.html). Good luck with the work.
Hello Sandy, either is okay... Celsius or Kelvin is okay. Just make sure your units are consistent. Ideally best to use Kelvin but in interpreting your results make sure you convert back to drg Celsius.
Hello @Ramesh, thanks for your interest in the channel and am glad you found the video nice. I have not done any video on natural frequency and fatigue issues. This can be an inspiration for a future video but I am working on other priorities now. Thanks for the sugggestion though.
Hello @Jahnavi, this is common experience of most students when working with the Student license edition of ABAQUS. There is no way around it, except you refine your mesh by increase the seed size (during the seeding of the model). You have to use the Query (or i-button) to check the number of nodes in the model as you keep refining it. For this specific video you are considering, irrespective of your refinement or coarsening of the mesh, I will not believe it is possible to stay within the 1000 nodes limit. You can however try modelling in 2D in which case you can. Good luck with the work.
I didn't use a contact property here. I assume perfect bonding between fibre and matrix. To be more accurate in your modelling, you probably need to incorporate some form of interface effect. I will not go with a contact property but this is possible, if you so desire but the better approach is to use an interface (cohesize zone/surface) effect.
I am trying to model a RVE with transversely isotropic fibre. After completing the relevant steps, I have added a local orientation. For the meshing part, using the same steps, the process is stuck in "Propagating mesh seeds". I am relatively new to ABAQUS. Kindly help me.
can you please let me know why only you did analysis for volume fractions 0.35. Could you help me please ?...... benefit of choosing this particular volume fractions.
Hi, good question. I have always worked with this composite from my PhD days. So, I have experimental and modelling data about it to help with validation. Also, due to the high matrix volume fraction of about 65%, then the matrix response will dominate the composite behaviour. This way the elastoplastic matrix model I use will model the composite reliably. It can be any volume fraction really but there is no reason why it must be this composite that I have used here.
@@MichaelOkereke okey even your nice explanation give me clarification little bit....... thankyou for your nice content help me a lot..love and blessing from INDIA.
Hello @Masfikun, I am not sure about doing this at RVE modelling scale. It can be done from the principle of Direct Stiffness Method. I have an extensive discussion of this in my textbook (see Chapter 3) where you can learn about this. Here is a link to the textbook: link.springer.com/book/10.1007/978-3-319-67125-3
Since you have 6 load cases results, I will suggest you create 5 more viewports on top of the current view port. Then tile them either vertically or horizontally so they can be arranged for a simultaneous video. Make sure for each view port, you open the right ODB file (associate it with it) so that result shows on that view port. Hopefully this makes sense. YOu can also watch this video: ruclips.net/video/6k0ZAacuAEo/видео.html
@@MichaelOkereke I am still worried about how to manipulate the z-tension plot, whose results are way bogus and out-of-scale for the other 5 plots/cases.
Of course, I am happy to chat about your problems but do connect with me via the CM Videos Insider Group link which is given at the description section of this video.
Hello Sir, I need your help, i am finding difficulty in applying periodic boundary condition to RVE in comsol multiphysics. Thank you for providing such nice content to the students
Hello @Kamalpreet, thanks for your interest in the channel. I have actually never done any modelling of PBCs in Comsol. I will need to try it out first understand the model and then see how to help. You can share with me your model and I can have a look. Thanks.
To model an RVE with Comsol multiphysics, if the version is from 5.0 upwards, all you need to do is go to physics, select domain > cell periodicity. Apply the necessary conditions to the model Thanks
Hello @Adarsh Shah, since you are looking at plotting for Shear on XY plane, it means that the deformation you are looing at is on the plane where X- axis is the outward acting normal and the direction of shear is along the Y-axis. Therefore, the reaction forces you should be looking at is the RF2 which is reaction due to displacement in the 2 or Y-axis. The shear deformation will be defined by the U2 displacement which is basically y-axis displacement. So, the correct variables are RF2 versus U2 to give you shear Tau12 versus Gamma12. Hope this makes sense.
Hello Floyd, yes in deed, this can be demonstrated for a multilayered composite. In fact, I showed a similar implementation recently for a bidirectional composite here: ruclips.net/video/zHUC0cVkdVM/видео.html Similar approach can be used for a multilayered composite.
@@MichaelOkereke thank you for the quick response. Just wanted to confirm, is the model used in this video and bi- directional composite video are multi-layered composites. One more question, does this video have static or dynamic behaviour results.
Hello.i can use these steps to calculate at high temperature .but i don't use the degree of temperature i change the module de young for example the temperature 250 the module de young decrease
I think what you are suggesting is to use the same implementation but for thermomechanical analysis. If so, then yes, it is possible. I will simply suggest that you make your step to be a coupled temperature-displacement step so within your simulation you will have both mechanical response (activated via displacement step) and thermal response (active via temperature step). As regards how you do the mechancial temperature-dependent response, you are right, some people would have to specify a range of Young's Modulus which change with temperature. It will be a reference/look up table that the simulation will use to know what do do as temperature change during the simulation. I hope these ideas help.
Hello sir, how to simulate the model of bcc structure used in this paper"Development of an Elastic Material Model for BCC Lattice Cell Structures Using Finite Element Analysis and Neural Networks Approache"
Hello @Fernanda Lima, thanks for the comment. The current version of MontCarlGen2D software that you refer to only produces 2D RVEs. The clue is in the name and there is a clear information about this on the purchase page on the site. A future development of the code will allow for recreating 3D RVEs as you refer to in this video however that version of the code is not available yet. The essence of this current video is to show how you can manually do this without needed a code for now but I do understand the ease that comes with having such a code for 3D domains but that will come in the future. Good luck with the work.
@Jabran Saroia, it normally is an error due to the model not seeing a material linked to the section. Check that the section is linked to the material.
Your video is really great, and professional! Not many researchers share their professional knowledge on RUclips. I nearly watched every of your videos, extremely helpful! Thanks a lot!
Thanks for the kind comments.
One of the knowledgable and helpful video that I found ever for my Mtech thesis. love from india ❤
Happy to hear that!
Very interesting video thanks from Algeria.
So nice of you
I would like to great thanks and appreciated you for your great contribution with this tutorial. It help me a lot to make my composite model
Hello @Eng Hussein, am glad it was helpful. Good luck with the research.
@@MichaelOkereke thank you so much.. I will need your support in my PHD
Nice series of videos on RVE modeling of UD composites
Glad you liked it Muhammad and thanks for your interest in the channel.
Hello Dr. Can we use the RVE homogenization technique to model the plastic and fractur response of composite material
Of course, you can. Certainly!
Dear Dr Michael
I'm kind of confused about the boundary conditions. In case I(Z tension), according to another reference book, BCs are XSYMM on XFront&XBack, YSYMM on YTop&YBottom, ZSYMM on ZBack and Z displacement on ZFront, which are a bit different from those used in this video. The results are also slightly different. Could you please explain this and tell what BCs should be used in similar situations.
Thanks!
Hi, I am not sure of the reference you mentioned here. The approach shown here is quite standard for an RVE setup for uniaxial loading. The 3 faces were constrained in one axis so that they can contract in the other directions. It is true uniaxial loading.
I will expect the results to be different if the boundary conditions are not the same. Please look at a few other publications and see what's going on.
I would like to great thanks and appriciated you for your great contribution with this tutorial. It help me a lot to make my composite model.
Glad it helped @Mehmet Korkmaz! Good luck with the research.
Very thanks for your video. May I ask a quick question? Why you didn't use periodic boundary condition
Good question. I wanted to do what most users can do without relying on a dedicated script.
To undertake correctly PBC on 3D domain as this, you would often need resources not implemented yet in ABAQUS. This will be a python script that can read the surface nodes, their positions and then using a kinematic coupling approach to link directly opposite and corresponding faces. I know most users would have not have this, but would want to still undertake this sort of study. So, in effect, the best was was to show what is possible using resources within ABAQUS.
I hope the explanation makes sense to you.
@@MichaelOkereke Thanks for your reply!!! I am a beginner in this area. My following question is:
Based on your reply, does that mean we can use these BCs you proposed to replace the periodic BCs? Thanks
No, they are different.
Hi Dr. Michael! I found really useful your video, I am doing my final degree project about RVE modelling and testing different materials. I followed all your videos and it helped me a lot. I don´t know if you could help me and say me what are the constraints and the BC for shear in plane XZ. That is what I have left. Thank you very much!
Glad it was helpful! Thanks for your kind comments.
I have made an extensive video where I showed how to impose the shear loads (Pure shear). You can check out the whole video but if you start from the 13th minute, you will see description on imposing pure XZ shear through this link: ruclips.net/video/uzUQz4NPkEs/видео.htmlsi=IWr6puvLgTwuEZE9&t=792.
Please let me know if you have any other queries. You can check other videos on shear loads, they might not be for UD composites as here but the implementation is still correct irrespective of the material under investigation.
@@MichaelOkereke Thank you very much, you have solved my doubt. It was very kind of you. Thanks!!!!!
Hello Dr Michael, i was wondering if you could please make a video about 3-point bending SENB testing following ASTM standard E1820.
Regards,
Aqsa UoS,England
Hello Aqsa, thanks for the comment. I will take note of this and if I get the time, I can make a video about it. Good luck with the research.
Hello dear doctor. In the student version, I am trying to create a 3D dentin model consisting of hollow tubules of variable diameter, that is, hollow fibers. While doing this, I follow the same logic you showed in this application. I mount the hollow fibers into the matrix consisting of hollow holes. But I'm having serious problems creating the mesh and I keep getting error returns from the program. As you know, the student version only allows 1000 knot points. I would be very grateful if you could make a suggestion about this.
Hello, unfortunately there is no way getting around the 1000 nodes limit for the student license of ABAQUS. You will have to find a full license to do so. Alternatively, you may then try using a 2D model of dentin and that might not require so much meshing. Since your 3D dentin model will have a lot of irregular geometries, that will lead to high mesh density hence you will easily exceed the 1000 nodal limit of the student license ABAQUS. Sorry, I could not be any more helpful.
Why are you using this boundary condition? can we evaluate this boundary to another one? thanks for all your useful videos
Hello @Anis, thanks for the query. I am not sure about your question. There are many ways of applying displacement loads on structures in a micromechanical model. I chose this type as it helps me extract the history outputs by tracking displacement and reaction forces variables in the reference nodes. If I use a direct application of displacement on the face, then my history variables will be too much and so unable to generate the stress-strain graphs.
As regards comparing it with some other type of loading, what do you have in mind. I will need some further explanations.
@@MichaelOkereke thanks for your answer. This was my answer .👏👏 good luck
Great job Michael 👏👏👊
Thanks Ijeoma. Appreciate your interest in the channel.
Hi Mr. Okereke, I'm enjoying your classes. If you don't mind, I have a question: Do we need to create a symmetric mesh to achieve these results? Did you intentionally use a symmetric mesh in all of these examples, right?
Hello, I assume what you mean by symmetric mesh is meshes that match between the corresponding parallel faces. If this is so, then this is not a requirement for this sort of work.
Symmetric meshes tend to be needed when working with Periodic Boundary Conditions. I am not using that type of boundary condition so this is not the case.
I hope it makes sense.
@@MichaelOkereke thanks for replying my message.
I believe that your channel is truly an oasis of good discussion and content.
Regarding to your answer, could you please clarify just one more question? Is it possible to obtain trustworthy results using Abaqus standard BCs - instead of PBCs and a symmetric mesh? Or is the latter condition necessary for achieving good results?
Hello, thanks for your kind comments about the channel. I want it to be exactly that, an oasis of good discussion where I get challenged on these questions and hopefully, I am able to provide some kind of guidance to my audience.
Regarding your question, of course, yes you can obtain excellent and trustworthy answers using ABAQUS standard Boundary conditions. Those are mainly the Dirichlet or Neumann Boundary conditions and they work quite well. Before the mathematics of PBCs were advanced, most analysis were based on these standard BCs. So, please feel confident to use them for your problem. In fact, for UD composites, many years ago I did a study to compare effective properties for PBC-based analysis and Dirichlet analysis and the results were comparable. You can see the paper here:doi.org/10.1016/j.commatsci.2012.12.036
The requirement for PBCs is important if you are working on a small RVE where the constituents have distinct constitutive behaviours and you have to capture those clearly. It gives a better result as the effect of boundary conditions on homogenized results is minimal where PBCs are involved compared to the other types. However, there is still confidence in the other approach.
If you are not able to implement PBCs for unsymmetric meshes (as this is quite challenging, I am working on a code that can help), then do not worry, it is okay and trustworthy going with ABAQUS standard BCs.
Thanks, Continue, please!
Hi Dr Michael
Thank you very much for providing this fruitful video and clear explanation. I have one question related to the case III which is compression. My question is why you set FixedZBack, RollerYBottom, and XBackRoller for the BC instead of only set the FixedYBottom and give the load in Y direction because this is a compression test? If my assumption is wrong/ not a right standard, could you please provide me the source of a standard uniaxial loading condition? Thank you!
Hello, this is just a different way of imposing loads compared to what you suggested.
It is often better to do it this way, as this is a micromechanical study based on an RVE. The chosen RVE represents one-eighth of the quadrant of the actual virtual domain. To simulate effective such an RVE, it is customary to impose loads on 3 free surfaces and attach dirichlet boundary conditions on the 3 other remaining surfaces.
I hope the above makes sense.
Thank you for your response@@MichaelOkereke , thats make sense.
One more, I've done exactly like you did in this tutorial, but I encounter the "too many attempts in this increment", do you have any suggestion for me?
Not really
Thanks Sir, just a doubt, i really need to know how it will effect if instead of creating a ref. point and constrained it with Equation, i directly create the set of the deforming plane, apply displacement on the very set and check the history output for it.
Thanks for the query. If you do as suggested i.e. extract history outputs (say S11 and E11) from the nodal set on say XFront, you will still get a result without the reference point approach that I used here. The limitation here is that it will be only giving you the history outputs on those XFront sets which is not representative of the full model. It will not be an accurate stress-strain plot. This is because of the heterogeneous nature of the virtual domain.
The best approach is to extract all the stress and strain for all elements in the domain and homogenize over them. This can work but the problem is that for a heavily meshed domain under consideration, you will have to write a lot of history variables to memory of the computer. ABAQUS typically runs out of history output memories once you have more than 10,000 history data. This is a problem and this is why I used this method as it requires only
@@MichaelOkereke Thanks for the clarification, just for confirmation,
If we constrain a surface (say XFront) to a ref- point, would it be correct to say that we are constraining the nodes of that surface to the ref-point. Does it also means that we are lowering down the history output data for every nodes on the surface to only one node i.e. for ref-Point.
If so, then the history output for the ref-point as explained in the Video is for the XFront nodes or for the whole model.
Amazing video. Thank you !!
Glad you liked it @Suyog Dani.
great video, just have 2 questions: 1. what is the contact properties you are using? 2. Are those boundary conditions periodic boundary conditions?
Hello @Haowei Huang, thanks for your interest in the channel.
1. I don't have any contacting parts in the model so no contacts were specified. In some models you may introduce contacts between fibres and matrix but this is not standard. What you may want to do is introduce a cohesive zone between fibre and matrix but that's a different problem.
2. The boundary conditions are Dirichlet or displacement boundary conditions. I haven't used PBC here.
I hope the above is okay with you. Let me know.
Great job! very clear and helpful. expecting for more!
Thanks, will do!
Hello, I really enjoyed your video and I'd like to thank you for the clarity of the exchanges. Could you tell me how you came up with the coordinates of the points to represent the fibres location in the matrix?
Hello, I used a code I developed called MontCarlGen2D to create that using the Monte Carlo method of random positioning of fibres within an RVE window.
@@MichaelOkereke Hello, Thank you for your reply. If I want to obtain this software, is it available over the counter?
Hello @leperroquet9045, this code is certainly available - just check it out here: www.cmvideos.org/shop/, which is the company website. Let me know if there is any difficulty in getting the code.
great work!
Thanks
Thank you for a very useful video! How would you simulate a unit cell for filament winded flywheel made of composite material?
Thanks for the kind comment. I am not sure about the problem you describe. It is quite complex and without better understanding it, it is difficult to be able to give any meaningful suggestion. Filament winded flywheel suggests it could be a textile composite or maybe some kind of angle-ply laminated composite system. Sorry, I am not able to help with such limited information. Good luck.
Yes, it is very complex, so it is understandable! But thank you anyway for answering:)@@MichaelOkereke
wow thanks for make a video of it .. i am really need this...
Most welcome 😊
sir, while calculating stress value you multiply the result by 10^-6. But since as per our unit was N/(micro meter)^2 So to convert this into Mega pascal I think we need to multiply by 10^6. So can you give me clarification over this !
Here is a video that explains my philosophy of dealing with units in ABAQUS, especially for micrometre lengthscales: ruclips.net/video/YeCBDkjEmDU/видео.html
@@MichaelOkereke thankyou sir,, its clear now
Great video, Michael! It'd be nice to show the Monte Carlo procedure to generate the arrangement of the fibers!
Hi Humberto, thanks for your interest in the channel. I have a MATLAB script that I developed for the Monte Carlo procedure. The principle of it is quite well known. I will consider making a video about this in future maybe to highlight the process and show how my MATLAB script works. Is that something you are interested in?
@@MichaelOkereke Hi Michael, it would be great to have such a video and I am interested in it. Great channel, by the way, congrats!
I’m interest and waiting for monte carlo video please
thats great video
i have one question, how can i find the fiber coordinates using Monte Carlo method
Hello eng, thanks for your interest in the channel. Finding fibre coordinates randomly is, as you rightly pointed out, best done using the Monte Carlo Method. I have already done this in a publication: www.sciencedirect.com/science/article/pii/S0927025612007744 . If you look at section 2.1 on *Geometric modelling* then, you can gain some insights into the thinking about the Monte Carlo Method. I will be making a series of videos on the Monte Carlo method, as this is a frequently requested video. Please keep coming back to the channel to see when the video becomes available.
Sir, I'm facing another problem right now.
I've several questions:
1. I've taken the length of the RVE in meter (m). What will be the unit of Stress before multiplying by
1 * e^ -6 as you've done in this video at 47:50?
2. What will be the unit for the Force (F) and the Displacement (D) in F-D curve whenever I'm taking the length, width & height in meter unit? (In this video u can find it at 47:04)
3. During stress calculation you've multiplied 100*100. Are they width*height or, length*width or, what?
(In this video u can find it at 47:28)
4. You've taken the length, height & width in micrometer but in this video at 47:47 , during stress calculation you've calculated the stress as Pascal and then you've converted it to Megapascal. Was it correct whenever you calculated the stress unit as Pascal? I'm a liitle bit confused about that because Nm^-2 represents Pascal but in your case the measurements were in micrometer!
Kindly help me by answering all 4 questions sir.
Here is the video you need that answers your queries: ruclips.net/video/YeCBDkjEmDU/видео.html
hello .dr .thank you for this import videos .what type of conditions at the limits used
The boundary conditions are normal Dirichlet (Displacement) boundary conditions. Nothing too special.
The reaction forces and displacements extracted from the model to excel are from the reference node? I didnt understand that part very well. Thanks!
It is a different way of extracting stress-strain plots but based on a reference node which is kinematically connected to a surface. I explained this in more details in this video. It is the third approach: ruclips.net/video/gCQzzF8Few8/видео.htmlsi=LICNUGB7j2SmayDJ
@@MichaelOkereke thank you!! I will check it out 💪
Dear Dr. Michael , Thanks for your helpful video , its really great job , however , would you please advise how did you generate the coordinate data of fibers by Montecarlo Algorithm in periodic style , Which software you had used , I would appreciate your help if you have a tutorial video or text explain this issue.
Hello Mortda, thanks for your interest in the channel. I used a monte carlo implementation to do this. See this video for the Monte Carlo method: ruclips.net/video/e6LK7YB3d_I/видео.html
If you want to get the actual software, then please wait for future videos when I release the code and video of how to use the software.
@@MichaelOkereke thank you so much
Could you please advise how to obtain the dispersion curves of an RVE unit cell in Abaqus?
Hello @Ha-Linh Nguyen, the concept of dispersion curves arises during dynamic assessment behave of heterogeneous materials or multilayered composites. This is because the differences in effective properties (Modulus and strengths) inside the heterogeneous materials leads to different propagation of wave dynamics within the heterogeneous material. As continuum has to be maintained, such differences in propagation will lead to dispersions of the wave as it hits different materials in the medium. There are a bunch of equations proposed by C. T. Sun in *Continuum Theory for Laminated Medium, Journal of Applied Mechanics, 1968* which helps in tracking this dispersion curves and plotting them. I believe this is the sort of thing you are interested in. Having said all these, I have actually never done this sort of modelling and so will not be able to help you with this query. I will have to research into it, understand it better, implement it before I am able to make a video about it. For now, this is all the help I can offer and good luck with the research.
Sir, in case of load application, if I use concentrated force for example 1200 N, the force displacement curve cannot be generated. Is it only generated when we apply the percentage strain?
My another question is how can we convert the concentrated force for example, 1200 N to percentage strain?
Hello @Ashfakur,
Your observations are right. If you use force, it will be overwritten by the underlying kinematic constraint equations involving displacements only. I think this is what's going on here.
On converting from force to displacement, you can use the force displacement equation for a uniaxially loaded system where the equation to use is:
Displacement, D = FL/AE
Where F = Concentrated force, L Edge length of RVE, A = Surface area and E = Youngs Modulus.
Good luck with this.
Sir, if there presents two type of fibers and one matrix in the system , what will be the value of E ? Which Young Modulus should we take as the value of E?
Or, how can we calculate the value of E?
You have to use all to something like this
E = Ef1Vf1 + Ef2Vf2 + EmVm.
The Ef's are for the two fibre phases.
Thanks.
Dear Sir, Did you use volumetric averaging method of homogenization for finding out the stresses and strains in the last?
Hello @Adarsh Shah, no, I did not use the volume averaging method. This was based on extracting reaction forces on a face and dividing the force with the area of the edge. This is similar to what is called the direct macro-micro approach but adapted for cases where periodic boundary conditions have not been implemented. I hope you understand.
@@MichaelOkereke Yes Sir, I understood. Thank you very much for helping!
Will it work if fibers are modeled in such a way that they touch each other, In other works each circles we draw touch/intersect each other at single point(externally)
I certainly think it will, but the right mechanical behaviour will involve having some kind of matrix media (no matter how thin in-between the two closest fibres). If there is a requirement for fibres touching, then I will suggest you incorporate some kind of contact mechanics which will mean during loading, when fibres touch, there is effective load transfer between the interacting fibres. This is technically the right approach but it is computationally very expensive as your contact will be All contacts which will cause the simulation to take too long.
Alternatively, you can bind the fibres to the matrices in node-to-node coupling which might mean you do not have to deal with contact effects. It is my long-winded way of answering the question to say, generally it will work with contact mechanics between touching fibres inclue but no, as this is a very computationally expensive way to study such composites.
@@MichaelOkereke Thank you for the prompt reply.
Can you suggest the type of contact that I can use in ABAQUS?
I mentioned already that surface-to-surface is ideal but there will be a lot of it... so try the all-contacting pairs option.
Sir, please make a video on the three-point bend test for the same problem.
Hello @Nag Mk, thanks for your interest in the channel. Three-point bending test is better studied using a macrostructural model rather than an RVE model as I did with this UD composite example.
Why would you want a three-point bending test on an RVE? This is possible, but I am not sure you learn a lot from such a model but there could be a research objective, so please share that with the channel.
Great video and illustration. Could you make a video showing the orthogonal cutting of this composite? Could be of great help. There is very little detailed work about it on this platform. Thank you
Hello Haris, thanks for your interest in the channel. I have not actually modelled orthogonal cutting of composites but that should be an interesting little project to tackle. I will consider it and maybe in future make a video on it. Thanks for the suggestion.
Sir, first of all I would like to thank you to your video. I wanna ask a question for you. Now, I am working on blast loading responce of a plate (only plate, not any fiber). I did some analysis via Abaqus. But I have a error. My error was;
Excessive incremental rotation of the elements in element set ErrElemExcessIncrementalRotation-Step1.
There is only one element with excessive rotations.
After I took this error, I increased my mesh elements number. But after this, I took another error. Now my error is this;
Element 21, node-1: the distance between the node and the source of the explosion is smaller than the charge radius (5.30000E-02). The CONWEP model is not valid at this distance.
What can you say about my error? How can I fix this error?
Best regards
Hello @Ali, great that you were able to solve the first problem. I think the second problem is due to the location of the charge that you are using in the CONWEP system. I will suggest you increase it and check to see if the second error is still happening. For example, it is saying the charge radius is 0.053 (I assume that is meters), hence you may have to increase it to 0.150 and see if that improves it. Do not that the further away the charge is from the plate (i.e the location of the explosive), the lesser the impact the charge/explosive will have on the plate so you cannot just increase it arbitrarily.
A rule I normally will follow is to start with a distance as far back as the half the edge length of square plate and then possibly reduce or increase it depending on the deflection you see on the plate. If you have an experimental data you are working with, then no guess work here, simply use the distace specified in the experiment. Good luck.
Here is a paper that might help you: doi.org/10.1016/j.ijimpeng.2011.02.005
Great work Doc, please can you do this for particulate composites?
Hello Adedotun Banjo, definitely its part of my plan to show how you can use this method for a particulate composite. I will make a video like this. Please watch out for this video.
Hi, Dr Micheal. Just wondering what is the contact property between the fibers and matrix? Or it does not matter in this case?
Hello @Haowei Huang, thanks for your interest in the channel. I have not used any contact in this model. The fibre is tied permanently to the matrix at the interface. In future, I will make a video where I include an interphase which is an improvement in this video.
@@MichaelOkereke that would be awesome! Looking forward to your new video. I am doing the comparison at the moment, with “hard contact”, “soft contact”, cohesive contact and without contact. The behaviours are all different. So without any contact case. In linear stage, can I assume that the contact stiffness is actually the matrix modulus? Does this make sense?
It shouldn't be the matrix Modulus but something above it but below the fibre Modulus. It is this interphase Modulus.
Great Stuff, Thanks a lot for making this video, I have one query why 10% of strain was taken in the model as a displacement boundary condition ?
Hello Karan, thanks for your interest in the video. Actually, the 10% is chosen randomly, just to get an idea about how the model will respond. Because I am doing a microscale analysis, where the matrix is explicitly modelled as an elastoplastic material (without damage included), then it is bound to show significant plastic deformation. It makes sense then to go for a higher displacement of 10% as against about 3% for most composites. To do the later however, we need to have a damage model included in the simulation so that we can predict experimentally recoverable data. I hope the explanation makes sense and that is what you are asking about.
@@MichaelOkereke Thanks a lot for such a good explanation, I have one more query regarding meshing in Abaqus for complex geometry, as I saw you are good in Abaqus, if okay can I mail you the geometry and ask my query? I am sure solving this complex part will be a engaging )
@@karangupta120 - thanks for your interest. I will suggest you subscribe to my CM Videos Insider Group (here cmvideos.ck.page/16a99bd0e3) and we can take the discussion further from there.
great job sir....Are you applying the periodic boundary condition at 16:18??
Hello Subrat, thanks for your interest in the channel. No, I have not used a periodic boundary condition for this video. I simply used a Dirichlet boundary condition but I will consider making some videos in future where periodic boundary conditions will be used. As you may know, a periodic boundary condition is always preferred for such micromechanical analysis but its implementation within common FEA solvers is not easy. Is this something you are interested in Subrat?
Great job, Thank you
Thanks for watching!
Perfect, Thank for very through vedio
Glad it was helpful!
hello doctor, how are you? i did exactly according to the movie you uploaded , but unfortunately when i try to make the first submission in case1 i get a warning message that says : the following parts have some elements without any section assigned : UdComposite. thank you for this problem i will fix it because this is the subject of my senior dissertation . thank you.
It looks like you missed out section assignment to the process. That is why you are not getting the simulation to run.
I did that part! Can you guide me and explain more so that I can solve my problem, of course I use the 2020 version
Very insightful video sir. Sir I have a query. What modification do we need to make in HISTORY OUTPUT REQUEST so that we can calculate shear modulus from shear forces in the 4th and 5th Case. And finally what should we choose in place of Reaction force and Displacement in the ODB history output to find shear modulus?
Hello @Sandy, thanks for your interest in the channel. For shear modulus you simply have to follow the same approach but of course change the boundary conditions to make sure you are simulating shear responses in I believe you mean YZ and XZ planes. For example if it is Shear YZ you are simulating, then you are simulating shear on the plane where Y-axis is the outward pointing normal and the shear forces are oriented along the Z-axis. Therefore, for this you need to extract shear force, RF3 (z-axis reaction force) and divide this by the area of the YZ plane where the shear force acts. This will give you a shear behaviour. I have made videos where such shear cases were investigated. Here are some of them: RVE Modelling of BD composites (ruclips.net/video/zHUC0cVkdVM/видео.html) and Hybrid Composites - an RVE Modelling Abaqus tutorial (ruclips.net/video/1rdbKix_biw/видео.html). Good luck with the work.
@@MichaelOkereke Thanks a lot sir😊
@@MichaelOkereke Sir if we want to apply a temperature field then the temperature value should be in Kelvin or Celsius scale?
Hello Sandy, either is okay... Celsius or Kelvin is okay. Just make sure your units are consistent. Ideally best to use Kelvin but in interpreting your results make sure you convert back to drg Celsius.
Hello, it is. Nice videos, how to determine natural frequency and fatigue property, no of cycle to failure
Hello @Ramesh, thanks for your interest in the channel and am glad you found the video nice. I have not done any video on natural frequency and fatigue issues. This can be an inspiration for a future video but I am working on other priorities now. Thanks for the sugggestion though.
sir, how do you have a fine mesh, when i submit the job it says student edition is restricted to 1000 nodes
Hello @Jahnavi, this is common experience of most students when working with the Student license edition of ABAQUS. There is no way around it, except you refine your mesh by increase the seed size (during the seeding of the model). You have to use the Query (or i-button) to check the number of nodes in the model as you keep refining it.
For this specific video you are considering, irrespective of your refinement or coarsening of the mesh, I will not believe it is possible to stay within the 1000 nodes limit.
You can however try modelling in 2D in which case you can. Good luck with the work.
Didnt you set up contact properties ? or not necessary ?
I didn't use a contact property here. I assume perfect bonding between fibre and matrix. To be more accurate in your modelling, you probably need to incorporate some form of interface effect. I will not go with a contact property but this is possible, if you so desire but the better approach is to use an interface (cohesize zone/surface) effect.
@@MichaelOkereke Thanks for the answer
I am trying to model a RVE with transversely isotropic fibre. After completing the relevant steps, I have added a local orientation. For the meshing part, using the same steps, the process is stuck in "Propagating mesh seeds". I am relatively new to ABAQUS. Kindly help me.
Hello Srinjoy, I think it's with your mesh. Increase the mesh/seed size and see what happens.
Yes, thanks and also thank you so much for uploading such wonderful videos on the topic. It is very hard to find such topics on the internet.
@@NoobGamerSJ Thanks
Dear Sir, Can we apply load on a reference point at a particular angle? or it will always be a linear loading?
Kindly help!
Yes Adarsh, this is possible. Just apply loads at angles by specifying say ux and uy values.
can you please let me know why only you did analysis for volume fractions 0.35. Could you help me please ?...... benefit of choosing this particular volume fractions.
Hi, good question. I have always worked with this composite from my PhD days. So, I have experimental and modelling data about it to help with validation. Also, due to the high matrix volume fraction of about 65%, then the matrix response will dominate the composite behaviour. This way the elastoplastic matrix model I use will model the composite reliably.
It can be any volume fraction really but there is no reason why it must be this composite that I have used here.
@@MichaelOkereke okey even your nice explanation give me clarification little bit....... thankyou for your nice content help me a lot..love and blessing from INDIA.
Damn! You're good!
Thanks.
how to find stiffness matrix of a structure (preferably in RVE modelling) in Abaqus?
Hello @Masfikun, I am not sure about doing this at RVE modelling scale. It can be done from the principle of Direct Stiffness Method. I have an extensive discussion of this in my textbook (see Chapter 3) where you can learn about this. Here is a link to the textbook: link.springer.com/book/10.1007/978-3-319-67125-3
Doc, can you please give me some advice on performing the Tie Viewports. I have all 6 load cases opened. On seeing Tie Vert./Hor., I see only one view
I fixed it. Thanks
Since you have 6 load cases results, I will suggest you create 5 more viewports on top of the current view port. Then tile them either vertically or horizontally so they can be arranged for a simultaneous video. Make sure for each view port, you open the right ODB file (associate it with it) so that result shows on that view port. Hopefully this makes sense. YOu can also watch this video: ruclips.net/video/6k0ZAacuAEo/видео.html
Thank you very much
@@MichaelOkereke I am still worried about how to manipulate the z-tension plot, whose results are way bogus and out-of-scale for the other 5 plots/cases.
Thank you for the video. However I just wanted to know what is MCRVEGen? Is it any software tool? I came across this in the paper that you have shared
Yes @Reshma, it is a software for creating random distribution of inclusions within an RVE.
@@MichaelOkereke thank you for your response. I'm working on RVE model. In case of doubts I would like to discuss with you.
I'm unable to find the software in browser. Could you please tell me the source to download
Hello Resma, I have not released the code yet for public use. It will become available soon.
Of course, I am happy to chat about your problems but do connect with me via the CM Videos Insider Group link which is given at the description section of this video.
Hello Sir,
I need your help, i am finding difficulty in applying periodic boundary condition to RVE in comsol multiphysics.
Thank you for providing such nice content to the students
Hello @Kamalpreet, thanks for your interest in the channel. I have actually never done any modelling of PBCs in Comsol. I will need to try it out first understand the model and then see how to help. You can share with me your model and I can have a look. Thanks.
@@MichaelOkereke
Thank you for your reply...
How can i share my comsol file with you Sir?
Are you signed up to my CM Videos Insider Group? If so, the sign up email has my email... send it through that.
To model an RVE with Comsol multiphysics, if the version is from 5.0 upwards, all you need to do is go to physics, select domain > cell periodicity.
Apply the necessary conditions to the model
Thanks
Dear Sir, In case IV-model (Shear-XY), from which value we will draw the plot: RF1, U2 or RF2, U2 or RF2, U2? Kindly help.
Hello @Adarsh Shah, since you are looking at plotting for Shear on XY plane, it means that the deformation you are looing at is on the plane where X- axis is the outward acting normal and the direction of shear is along the Y-axis.
Therefore, the reaction forces you should be looking at is the RF2 which is reaction due to displacement in the 2 or Y-axis. The shear deformation will be defined by the U2 displacement which is basically y-axis displacement. So, the correct variables are RF2 versus U2 to give you shear Tau12 versus Gamma12.
Hope this makes sense.
@@MichaelOkereke Thank you very much Sir.
Hi sir. can you do video on Directionally distributed steel fibre reinforced concrete
Will try
ok teacher and sorry for the late replay.
sorry sir for the late reply, because I was outside the city and i haven't checked your message
Hello, thank you really for your videos, I appreciate so much. Could you show us or u can give me some instructions to define a choesive zone Please?
Hello, I will do this in a future video.
Is it possible to do some changes and do it multi-layered UD composite
Hello Floyd, yes in deed, this can be demonstrated for a multilayered composite. In fact, I showed a similar implementation recently for a bidirectional composite here: ruclips.net/video/zHUC0cVkdVM/видео.html Similar approach can be used for a multilayered composite.
@@MichaelOkereke thank you for the quick response. Just wanted to confirm, is the model used in this video and bi- directional composite video are multi-layered composites. One more question, does this video have static or dynamic behaviour results.
非常好
Hello, thanks for the comment.
Hello.i can use these steps to calculate at high temperature .but i don't use the degree of temperature i change the module de young for example the temperature 250 the module de young decrease
I think what you are suggesting is to use the same implementation but for thermomechanical analysis. If so, then yes, it is possible.
I will simply suggest that you make your step to be a coupled temperature-displacement step so within your simulation you will have both mechanical response (activated via displacement step) and thermal response (active via temperature step).
As regards how you do the mechancial temperature-dependent response, you are right, some people would have to specify a range of Young's Modulus which change with temperature. It will be a reference/look up table that the simulation will use to know what do do as temperature change during the simulation.
I hope these ideas help.
Hello sir, how to simulate the model of bcc structure used in this paper"Development of an Elastic Material Model for BCC
Lattice Cell Structures Using Finite Element Analysis
and Neural Networks Approache"
Thanks Amit for sending the paper reference. I will consider it for making a video on the topic.
@@MichaelOkereke hello sir,kindly look into my matter
Nice video sir
Thanks
I just bought the plugin and when I use the script generated in matblab, it gives me only a 2D rve. Not a 3D one like in the video....
Hello @Fernanda Lima, thanks for the comment. The current version of MontCarlGen2D software that you refer to only produces 2D RVEs. The clue is in the name and there is a clear information about this on the purchase page on the site. A future development of the code will allow for recreating 3D RVEs as you refer to in this video however that version of the code is not available yet. The essence of this current video is to show how you can manually do this without needed a code for now but I do understand the ease that comes with having such a code for 3D domains but that will come in the future. Good luck with the work.
ErrElemMissingSection Error is coming again and again.. kindly help
@Jabran Saroia, it normally is an error due to the model not seeing a material linked to the section. Check that the section is linked to the material.
God bless you
Thanks 😊
doc can you do pores structure?
Of course Red Riot, I can. What kind of pore structure do you mean?
Thx a lot Dr. very helpful. Just hated the background music. It was very distracting.
Hi @Abdalla, thanks for the feedback. I will bear that in mind.
Maybe you can promote your channel to your students!
Yes, I do already. They were the motivation for starting the channel I initially so they are fully invested.
Great job! thank you
Thank you too!