Hello CEE, It is finally time indeed 😊. For me personally, and especially in situations when I continuously struggle to understand a particular subject, I get to a point where I “make peace” with that fact and “move-on”, however such things are almost always guaranteed to find their way back one way or the other and I often find myself having to apply or to use the same particular subject(s) over and over again irrespective of the supposedly “peace” that I had thought to have made with the matter. One of those subjects (for me) is the “internal workings” of the Finite Element Method. I would like CEE to know that all of your teachings, responses, and comments regarding the FE method so far in both past videos, including the current introductory video on meshing are a huge (gigantic) help for someone like me to begin to appreciate the “internal workings” and the sheer breadth, and possible limitations of the FE method. Thank you for your patience and for finding ways to try and simplify such involved issues in order to allow for essential appreciation of the main concepts, plus thank you also for the bonus provision of the key reference materials and links on the FE method. I am definitely never forgetting that a “rectangle” … (oh no, scratch that 😊) …I mean a “square” or “a square_ish” element shape is the ideal FE shape for best results with the FE method. Granted that triangular element shapes are also used in the FE method. 13:03 I find your comparison of the subtle differences in the results from the use of quad v/s triangular shaped finite elements to be quite enlightening; also, how one “gets the best of both worlds” with quad when one considers a quad as a shape made of two (or double) triangles. 19:55 Thank you for the practical introduction to available “adaptive mesh refinement” options in RSAP. Being “the average user 😊” myself, I can’t help to laugh at the undisputable fact which CEE have pointed out that “…the average user goes to job preferences and just select the mesh fineness and goes on with life….”; Yes, this is true, but at the same time, it also does really feel pretty good to me (one of the average users) to know the basics of how to use mesh refinement or adjustment options in RSAP. I think such refinements can actually become handy from time to time; uses at other possible irregular points of interest in a structure being analysed come to mind. 20:52 The stress (and moments) on a point load on an area: I find this concept to have been very well-illustrated and captured here by the CEE, thank you! It is interesting indeed and it does give food for thoughts about what the CEE have refer to here as a “…. numerical problem in the FE….”. 23:09 Thank you Mr. Editor for the note. 23:14 What an interesting (and yet spot on) choice of words “……how to detect if one is going into a numerical singularity…. with ones FE mesh….”, these are the kinds of questions and answers that (for me) makes the CEE channel unique. To the CEE team: kindly know that these kinds of Q&A come highly appreciated and are treasured. 25:30 Thank you for touching upon this issue. To add to this discussion, the differences are even more “visible” if one were to design the slab reinforcements based on the moments results from an unsymmetrical mesh on a symmetrical plate or structure. The resulting required reinforcement regions in this case are not symmetrical as one would ordinarily anticipate. I am satisfied with the current explanation offered by the CEE in this meshing video. My addition to the discussion is intended if it could also help others. 26:15 & 27:45 I guess there is “homework” here to read about shear locking, shallow angles, and NURBS. 28:35 If I recall correctly, the meshing (or non-meshing by RSAP in this case) around the point of penetration of a shell with a beam element is something that was also explored in the silos video series while it was being intended to generating specific located points in a shell to attach the “silos column support bracket”. Congratulations that your software STRADO actually does that. This is an important feature and it does have its uses in the FE method. 29:36 This is hilarious and educative at the same tine: …...if you don’t have things connected to a mesh at the node(s), then “…they don’t see, they don’t feel, they don’t care….” about each other 😂🤣😂. 29:57 and 30:42 Thanks for showing some of the common things to avoid with FE modelling. 32:30 I welcome the opportunity to explore further meshing options in future video(s). And within the context and requirements of the practical problem(s) being solved at that time. For example, CEE have identified certain mesh elements at 31:50 as “bad elements”, now if there are advanced meshing options to “correct or to adjust such elements” then this is where those options should be explored (in my opinion). The same principles could be considered to introduce and to explore further advanced meshing options on video or problem basis (or even an approach which looks at a collection of common problems and the associated advanced meshing options that can be implemented in order to deal with those problems could be considered?) By the way, can the “bad elements” mentioned in 31:50 be corrected by using some of the “adaptive mesh refinement” options discussed here at 19:55? More “homework” 😊 Delaunay triangulation and Voronoi diagram? On second thoughts, I think we definitely need the planned deep dive video by the CEE 😁. For me, this (today’s) meshing video is one of those “not your everyday kind of lessons”, It was a good introduction and an absolutely necessary one in the field of the FE method. Thank you so much for the video and keep up the good work. I am looking forward to the next CEE videos Regards, DK Ps… Tangent: at 02:39 I am curious about the presence of the Fx, Fy, reactions values (greater than zeros) on each of the individual support nodes. Considering that the structure and the mesh are (for all intents and purposes) perfectly symmetrical, and that the only load applied to the structure is a vertical load along the global Z-axis direction. Why are there non-zero reaction values in the X and Y (Fx and Fy) directions at all? What calculations or scenarios is RSAP considering which results in these reaction values being non-zeros? Apologies if this question seems a repetition of a question, I asked some time ago to which the CEE noted that materials have “Poisson’s ratio”, [are the Fx, Fy reactions due to the deformation in the material in a direction (X and Y) perpendicular to the direction (Global Z) of the applied load or force?], If this is related, can CEE expand a bit on this in the current context?
Hello Engr. DK, the topic of meshing is a really deep one. To be honest, the video actually scratches the surface, and we are lucky that RSA has such good mesh generation capabilities. It pleases me to know that you do actually want to understand some of the inner workings of the FEM. Please note that I did provide a rather simplistic explanation with some "key words" that would help interested viewers like you to research on their own. You know, typical CEE style. For some reason I messed up and kepy saying rectangles (LOL). The editor actually hated me for this and just gave up, I guess: "achievement unlocked: make the editor quit"? Also, the editor in 13:03 actually had a great addition, as he usually do. Look, I know the editor does sometimes roast me, but he is actually a cool dude who knows what to say when. When I was talking about mesh refinement options, I actually had in my mind ppl like you. Now truth to be said, I do not expect this part to appeal to a wide audience, but is actually a kind of "testing the waters" of a "Deep ocean". On "burning moments", I have been asked to give more clarification and recommendations about that, so I am searching for a reference I can cite before I just give an opinion. Otherwise, I will only have my own opinion to give. (I usually like to provide references, as in a discussion, you cannot just appeal to a youtube channel, but you could appeal to a book). Your feedback on those internal monologues and question I do have is highly appreciated. I do take an "academic" approach in my videos, but also coupled with my experience as a site engineer and design engineer in my past. Actually the length of videos is usually longer, because I usually speak my mind. Now sometimes, Mr. Editor cuts stuff because my thoughts just decent into chaos and gibberish. This is basically the creme-de-la-creme. I agree 100% with your observation about reinforcement and mesh symmetry. About "not seeing each other ^_^", I proving this with number and results.. I should have somehow opened or shown something to prove that those two things are not connected and do not share anything. I want to thank you with regard to your feedback about deeper meshing topics. I agree. I think I should try "catch" the best chance to include it in a practical and helpful way, rather than forcing the issue, because it is kind of deeper than what was presented in this video. To answer the question: Bad elements can be fixedd by both, either adaptive mesh refinement or mesh size. Of course the warning here is that mesh size affects the entire structure, which would increase significantly the calculation effort, whereas adaptive mesh refinement affects the position you are refining. I love delaunay triangulation, I know it because I struggled with it while programming STRADO's mesher. Now STRADO's mesher is NO WAY near RSA's, but I had to really work hard on the triangulation to make STRADO do anything. Oh, about the X and Y reactions, it is because you have indeterminate structures... The easiest way for me to explain this in written form is to remind you of what happens when you have a frame with as follows: Bay: 5meters Height: 3meters Gravity Loads Pin supported (2 pins, 1 for each column) So you have 3 elements, 2 columns and 1 beam. Let's imagine that the supports are not pins, but 1 pin and 1 roller. The frame under the gravity loads would "open up". i.e., the column with a roller would have the tendency to open up, as the beam wants to bend in a "smily face" pattern. Now if you have 2 pins, the pin will restrict this movement, and generate Fx, which causes the other pin to generate an opposite Fx. Now this concept can be extended to 3D structures. Please let me know if this explanation made sense, coz I know it is kind of odd to explain visual stuff using text. I tried to be as vivid as possible. Regards, CEE
@@CivilEngineeringEssentials Thank you for the feedback, it is very much appreciated. I agree with you 100% about the editor; that dude does know his stuff and he is quick-witted. I cannot think of anyone more suited for the task, we are all very fortunate to have him. Yes, I can happily confirm that the explanation about the reactions Fx and Fy is as vivid as it was intended and it does make good sense. Thank you for this. See you in part 4 of the slab series. Kind regards, DK
There was a nice discussion with Engr. Candea about mesh sizes. Here is an out-take of my comment to him: about mesh size, here is an article www.masterseries.com/blog/fe-mesh-size But to summarize, span / 10 or 1000mm, so in case of normal spans between 3 to 8 meters, you would end up with 0.3-0.8 so your college approximation is quite on spot.
@@CivilEngineeringEssentials What a beautiful 💐reference to complement the discussions and the corresponding content of part 3 of the slab series video (thanks to you both, Jesus Candea and the CEE). I especially love the resonating tone of this article and its practical guidance for the designer or the user to always aim to maintain an acceptable level of “balance” based on the designer or the user’s own informed judgment and also based on the specific requirements of the model that is being solved or that is being considered. I think that the author(s) have done well in trying to define words such as “too big” and “too small” into relatable terms for the purpose of FE mesh sizing. The suggested convergence testing for both global analysis and for localised areas of high-stress concentrations coupled with localised refinement of the mesh, merges well with similar concepts that were discussed by the CEE in today’s main video including localised mesh refinement options such as “adaptive mesh refinement” techniques, and the theoretical infinite stresses at points of singularity. This article is an ideal summary in a “Kurzgesagt” fashion on the introduction and basics of the FE mesh sizing, accuracy, required computational time and processing power, convergence testing to help the designer or the user to make informed judgment and decisions on the final adopted size of mesh for their model, or for their project. Of course, there are multitude of other factors such as geometry, constraints, loading, etc. which would require more in-depth investigations. But this article right here, is without doubt or questions, a very good starting point. By the way, I started watching some of the videos from the Kurzgesagt Channel and what I can say is ‘man, oh man!'🥰; there are some interesting points of views to life, to science, and to every other thing(s) out there. I can only now begin to grasp why you would find their (Kurzgesagt) work fascinating. Their video on “How to Make a Kurzgesagt Video in 1200 Hours or more” for example, had made me pause and I couldn’t help but wonder how in the word is the CEE and Mr. Editor even begin to manage to produce such top-notch videos in 7days x 24hrs = 168 hrs or less?😳😳🤔 I have said it before and I will say it again: If you (CEE) were ever in doubt, know that you are doing exceptionally well with your videos. I am and always have been in awe 👏💯👏. Keep up the good work sir!
thnx a lot Engr. DK ^_^ Happy you liked the reference. Editor told me he is watching videos about how to improve the presentation of CEE videos to be a little bit more "action packed" and less monotonic. He is still learning so I do not expect anything soon. Kurzgesagt is amaaaaazing
Great video! Your pedagogical skills are on another level. Due to this, I benefit greatly from your videos, despite the fact that I have already taken challenging courses(in my opinion) in the university on this topic. I know you have been an TA in some classes, however, I've had several professors that I have had a really hard time learning from, so all the credit to you, it is clear that you have a talent for teaching. This, combined with your background (having an academic background from PhD work in addition to experience as a working engineer(if I have caught that right)) truly makes this an exceptional channel for engineers of every level. Furthermore, I am always curious of what I am missing out on when considering other software. Currently, I have used a lot of time trying to really master Robot and its features. I find that the most valuable skills and information are those that can be applied across a wide range of software. As it seems that you have experience within different types of software, in addition to the fact that you are creating your own FE-program, it would be great to get your opinion on how Robot performs compared to other. What are the ups and downs with Robot? Actually, I just recently found out that Robot offers volumetric elements(i.e. still not anywhere near of "mastering Robot" xD), and was really surprised. And I started to wonder what distinguishes Robot from software programs like Abaqus and Ansys(which I view as the cornerstones of the FEM, but have very little experience with). Do you have some thoughts about this? Thank you for your videos!
Thnx a lot ^_^, I do appreciate your comment and feedback. About some general thoughts on Autodesk Robot, I would split the comment into two parts: Part 1: the comparison of robot as a structural analysis software in comparison to regular structural analysis softwares. Here I am talking about ETABS, SAP, STAAD, Prokon, AxisVM and Dlubal (there is more, but those are what came into my mind). Here Robot has multiple advantages: first of all, it is really easy to use and the controls are kind of trivial. Even when you mess something up (as is always the case in my videos -_-), Robot is there to the rescue and provides you with multiple ways to "save the day". Furthermore, its ability to mesh is second to none. Also, the seamless switch between analysis, design, drawing, calculation notes is quite outstanding. Furthermore, BIM seems to be getting increasingly interesting to the general Engineering profession. Here, Autodesk Revit is a resonating software. Now what does robot do with revit? a lot. Since Robot is an Autodesk product, there is a Revit/Robot integration, which is helpful in case you are working in a design firm that does implement Revit for a multi-disciplinary team. Now about the comparison of Robot with fully fledged research softwares such as AnSYS, Abaqus, LS-Dyna and the liking, here we are talking about two totally different softwares. Now note that AnSYS can do ANYTHING robot can do in the FEM department, but does not have a lot to offer in practical outputs such as drawings, calculation notes and so on. AnSYS has a much much much much deeper FEM basis, allowing users to dive deep into non-linearity, dynamics, buckling problems, contact problems in ways Robot (and all application-based softwares) would never do. Since you seem to be interested in AnSYS and Abaqus, I might detect some sort of "positive interest" or fascination in those cool results those research softwares provide. Maybe you are interested in a research career? if this is the case, then Autodesk robot would not be the best option for doing research. I will touch on the volumetric elements in future videos, but they are still a little bit limited when compared to research-level softwares. Finally, Indeed, I was a teaching assistant in 2010, then a lecturer in 2012, then since 2016, an assistant professor
Thank you for your response! Ofcourse, im sorry about missing out on the lecturer and Assistant professor part. I do think you have mentioned this before, so my bad for messing up. I do indeed have a interest for the more «advanced» softwares. In my master thesis i used this program called DIANA, which was great for representing different nonlinear properties of concrete. Atm. I have started working as an engineer, and do like this as well. However, your videos keeps poking into fields that makes me want to do more research again😄 Anyway, i really appriciate your videos and that you take the time to reply the comments.
Hi, good video. At around 23mins 20 secs in the video you speak about high concentrated stressed areas, burning areas. Is there a rule of thumb of how much you move away from these highly concentrated areas when interpreting moments and shears? Even when applying reduced forces over columns and walls? As this is always very difficult to interpret, and even when interpreting concentrated reinforcement areas over the heads of columns? Should it be taken at column face perhaps rather than centre of column node? Also, a video of how to interpret such results would be good, as there can be a huge difference between moments and and shears, and reinforcement areas required directly over column heads and those a short distance away, just a suggestion….keep up the go work!👍
Hi there, first of all, a huge thnx to your comment. My own little rule of thumb is to move half the width of the column away from its center. But I will be talking about that in the next video, coz I found that it might be too long for this video to be included. So, your suggestion is highly regarded and I will take it into account the next video (or maybe the one after it). However, I do not just want to shift the answer for a next video, you can read a little bit about impossible convergence in A first Course in the Finite Element Method (Daryl L Logan) Concentrated or Point Loads and Infinite Stress page 360 Anyways, thank you very much for your comment. Consider supporting the channel by suggesting it to your colleagues. It helps immensely. Regards, CEE
Dear CEE, let me tell you that your videos are getting better and better! A Few things i want to disscuss: - in collegue i was told to use in slab models FE size about 0,5-1 m. Do you agree? - also, to use emitters or “radial” FE in corners and singularities such hollows or high loads. - whats wrapling in meshing? As always thank you very much!
Hi there dear Engr. Candea, thnx a lot for your comment. I do actually feel larger responsibility with the increasing number of subscribers. I am actually learning how to more effectively video edit my videos and make them more interesting. Once I am at a certain level, then I will try it out. Now, to your questions: about mesh size, here is an article www.masterseries.com/blog/fe-mesh-size But to summarize, span / 10 or 1000mm, so in case of normal spans between 3 to 8 meters, you would end up with 0.3-0.8 so your college approximation is quite on spot. And I agree with emitters, but to a certain extend, you saw what happened if you go too far. Finally, about mesh wrapling, I am not 100% sure about it to be honest. Do you mean "warping"? Warping in that context is how much a FEM element departs from being a plane. Now in normal FEM, the warping should be zero because we are assuming that our elements are plane elements. This is EXTREMELY important because we do have an assumption about "plane stress" or "plane strain", with plane being the operative word. Now there could be some interesting research in 3D warped elements that do not follow a plane, but in the regular sense, a FEM will break it into triangles that are always planes (a triangle is guaranteed to be planar). Hope I was able to answer your inquiry. Regards, CEE
@@CivilEngineeringEssentials indeed they are! Crystal clear i must say thank you for everything again CEE. Hoping to watch the FEM advanced options from RSAP. Regards from spain
Hi dear CEE, theres an option in SAP2000 "automatic frame mesh" where you can set the minimum number of segments in bars when analyzing the structure, they act like nodes but they arent, so you can provide more realistic modal analysis results as accurate frequencies and so on... is there something like that in robot aswell? or the only thing we can do is to generate nodes without creating new bars with the division tool for all the bars before running calculation? As always thank you so much! Saludos
Hi there Engr. Candea, yep, you can create nodes without actually breaking the bar. (It is gonna be 1 bar, but multiple nodes on it) I think it was in (Edit -> Divide) Generate Nodes without dividing However, I have to mention something from the "insides of Robot". If you actually go to: Right Click -> Display -> Members -> Number of Calculation Elements then you will see that robot "behind the scenes" generated FEM elements. It is a loooooooong programming gibberish, but here is the quicker version: 1) You have "geometric elements" those are the things that you define, such as "members and lines" with their regular number 2) You have "FEM elements" that you do not define, but are generated for you during the "model generation stage". Those hidden elements can only be shown given the display trick The only reason I know this, is because I program my own FEM software ^_^ Still, not to derail your question, yes, via the divide command, you can add nodes in the mesh of a linear element without breaking it into multiple pieces "you actually do, but the interface hides those" Regards, CEE
Hello CEE,
It is finally time indeed 😊. For me personally, and especially in situations when I continuously struggle to understand a particular subject, I get to a point where I “make peace” with that fact and “move-on”, however such things are almost always guaranteed to find their way back one way or the other and I often find myself having to apply or to use the same particular subject(s) over and over again irrespective of the supposedly “peace” that I had thought to have made with the matter.
One of those subjects (for me) is the “internal workings” of the Finite Element Method. I would like CEE to know that all of your teachings, responses, and comments regarding the FE method so far in both past videos, including the current introductory video on meshing are a huge (gigantic) help for someone like me to begin to appreciate the “internal workings” and the sheer breadth, and possible limitations of the FE method. Thank you for your patience and for finding ways to try and simplify such involved issues in order to allow for essential appreciation of the main concepts, plus thank you also for the bonus provision of the key reference materials and links on the FE method.
I am definitely never forgetting that a “rectangle” … (oh no, scratch that 😊) …I mean a “square” or “a square_ish” element shape is the ideal FE shape for best results with the FE method. Granted that triangular element shapes are also used in the FE method.
13:03 I find your comparison of the subtle differences in the results from the use of quad v/s triangular shaped finite elements to be quite enlightening; also, how one “gets the best of both worlds” with quad when one considers a quad as a shape made of two (or double) triangles.
19:55 Thank you for the practical introduction to available “adaptive mesh refinement” options in RSAP. Being “the average user 😊” myself, I can’t help to laugh at the undisputable fact which CEE have pointed out that “…the average user goes to job preferences and just select the mesh fineness and goes on with life….”; Yes, this is true, but at the same time, it also does really feel pretty good to me (one of the average users) to know the basics of how to use mesh refinement or adjustment options in RSAP. I think such refinements can actually become handy from time to time; uses at other possible irregular points of interest in a structure being analysed come to mind.
20:52 The stress (and moments) on a point load on an area: I find this concept to have been very well-illustrated and captured here by the CEE, thank you! It is interesting indeed and it does give food for thoughts about what the CEE have refer to here as a “…. numerical problem in the FE….”. 23:09 Thank you Mr. Editor for the note. 23:14 What an interesting (and yet spot on) choice of words “……how to detect if one is going into a numerical singularity…. with ones FE mesh….”, these are the kinds of questions and answers that (for me) makes the CEE channel unique. To the CEE team: kindly know that these kinds of Q&A come highly appreciated and are treasured.
25:30 Thank you for touching upon this issue. To add to this discussion, the differences are even more “visible” if one were to design the slab reinforcements based on the moments results from an unsymmetrical mesh on a symmetrical plate or structure. The resulting required reinforcement regions in this case are not symmetrical as one would ordinarily anticipate. I am satisfied with the current explanation offered by the CEE in this meshing video. My addition to the discussion is intended if it could also help others.
26:15 & 27:45 I guess there is “homework” here to read about shear locking, shallow angles, and NURBS.
28:35 If I recall correctly, the meshing (or non-meshing by RSAP in this case) around the point of penetration of a shell with a beam element is something that was also explored in the silos video series while it was being intended to generating specific located points in a shell to attach the “silos column support bracket”. Congratulations that your software STRADO actually does that. This is an important feature and it does have its uses in the FE method.
29:36 This is hilarious and educative at the same tine: …...if you don’t have things connected to a mesh at the node(s), then “…they don’t see, they don’t feel, they don’t care….” about each other 😂🤣😂. 29:57 and 30:42 Thanks for showing some of the common things to avoid with FE modelling.
32:30 I welcome the opportunity to explore further meshing options in future video(s). And within the context and requirements of the practical problem(s) being solved at that time. For example, CEE have identified certain mesh elements at 31:50 as “bad elements”, now if there are advanced meshing options to “correct or to adjust such elements” then this is where those options should be explored (in my opinion). The same principles could be considered to introduce and to explore further advanced meshing options on video or problem basis (or even an approach which looks at a collection of common problems and the associated advanced meshing options that can be implemented in order to deal with those problems could be considered?)
By the way, can the “bad elements” mentioned in 31:50 be corrected by using some of the “adaptive mesh refinement” options discussed here at 19:55?
More “homework” 😊 Delaunay triangulation and Voronoi diagram? On second thoughts, I think we definitely need the planned deep dive video by the CEE 😁.
For me, this (today’s) meshing video is one of those “not your everyday kind of lessons”, It was a good introduction and an absolutely necessary one in the field of the FE method. Thank you so much for the video and keep up the good work.
I am looking forward to the next CEE videos
Regards, DK
Ps…
Tangent: at 02:39 I am curious about the presence of the Fx, Fy, reactions values (greater than zeros) on each of the individual support nodes. Considering that the structure and the mesh are (for all intents and purposes) perfectly symmetrical, and that the only load applied to the structure is a vertical load along the global Z-axis direction. Why are there non-zero reaction values in the X and Y (Fx and Fy) directions at all? What calculations or scenarios is RSAP considering which results in these reaction values being non-zeros? Apologies if this question seems a repetition of a question, I asked some time ago to which the CEE noted that materials have “Poisson’s ratio”, [are the Fx, Fy reactions due to the deformation in the material in a direction (X and Y) perpendicular to the direction (Global Z) of the applied load or force?], If this is related, can CEE expand a bit on this in the current context?
Hello Engr. DK,
the topic of meshing is a really deep one. To be honest, the video actually scratches the surface, and we are lucky that RSA has such good mesh generation capabilities.
It pleases me to know that you do actually want to understand some of the inner workings of the FEM. Please note that I did provide a rather simplistic explanation with some "key words" that would help interested viewers like you to research on their own. You know, typical CEE style.
For some reason I messed up and kepy saying rectangles (LOL). The editor actually hated me for this and just gave up, I guess: "achievement unlocked: make the editor quit"?
Also, the editor in 13:03 actually had a great addition, as he usually do. Look, I know the editor does sometimes roast me, but he is actually a cool dude who knows what to say when.
When I was talking about mesh refinement options, I actually had in my mind ppl like you. Now truth to be said, I do not expect this part to appeal to a wide audience, but is actually a kind of "testing the waters" of a "Deep ocean".
On "burning moments", I have been asked to give more clarification and recommendations about that, so I am searching for a reference I can cite before I just give an opinion. Otherwise, I will only have my own opinion to give. (I usually like to provide references, as in a discussion, you cannot just appeal to a youtube channel, but you could appeal to a book).
Your feedback on those internal monologues and question I do have is highly appreciated. I do take an "academic" approach in my videos, but also coupled with my experience as a site engineer and design engineer in my past. Actually the length of videos is usually longer, because I usually speak my mind. Now sometimes, Mr. Editor cuts stuff because my thoughts just decent into chaos and gibberish. This is basically the creme-de-la-creme.
I agree 100% with your observation about reinforcement and mesh symmetry.
About "not seeing each other ^_^", I proving this with number and results.. I should have somehow opened or shown something to prove that those two things are not connected and do not share anything.
I want to thank you with regard to your feedback about deeper meshing topics. I agree. I think I should try "catch" the best chance to include it in a practical and helpful way, rather than forcing the issue, because it is kind of deeper than what was presented in this video.
To answer the question: Bad elements can be fixedd by both, either adaptive mesh refinement or mesh size. Of course the warning here is that mesh size affects the entire structure, which would increase significantly the calculation effort, whereas adaptive mesh refinement affects the position you are refining.
I love delaunay triangulation, I know it because I struggled with it while programming STRADO's mesher. Now STRADO's mesher is NO WAY near RSA's, but I had to really work hard on the triangulation to make STRADO do anything.
Oh, about the X and Y reactions, it is because you have indeterminate structures... The easiest way for me to explain this in written form is to remind you of what happens when you have a frame with as follows:
Bay: 5meters
Height: 3meters
Gravity Loads
Pin supported (2 pins, 1 for each column)
So you have 3 elements, 2 columns and 1 beam.
Let's imagine that the supports are not pins, but 1 pin and 1 roller. The frame under the gravity loads would "open up". i.e., the column with a roller would have the tendency to open up, as the beam wants to bend in a "smily face" pattern.
Now if you have 2 pins, the pin will restrict this movement, and generate Fx, which causes the other pin to generate an opposite Fx.
Now this concept can be extended to 3D structures.
Please let me know if this explanation made sense, coz I know it is kind of odd to explain visual stuff using text. I tried to be as vivid as possible.
Regards,
CEE
@@CivilEngineeringEssentials
Thank you for the feedback, it is very much appreciated. I agree with you 100% about the editor; that dude does know his stuff and he is quick-witted. I cannot think of anyone more suited for the task, we are all very fortunate to have him.
Yes, I can happily confirm that the explanation about the reactions Fx and Fy is as vivid as it was intended and it does make good sense. Thank you for this.
See you in part 4 of the slab series.
Kind regards, DK
There was a nice discussion with Engr. Candea about mesh sizes.
Here is an out-take of my comment to him:
about mesh size, here is an article www.masterseries.com/blog/fe-mesh-size
But to summarize, span / 10 or 1000mm, so in case of normal spans between 3 to 8 meters, you would end up with 0.3-0.8
so your college approximation is quite on spot.
@@CivilEngineeringEssentials
What a beautiful 💐reference to complement the discussions and the corresponding content of part 3 of the slab series video (thanks to you both, Jesus Candea and the CEE).
I especially love the resonating tone of this article and its practical guidance for the designer or the user to always aim to maintain an acceptable level of “balance” based on the designer or the user’s own informed judgment and also based on the specific requirements of the model that is being solved or that is being considered.
I think that the author(s) have done well in trying to define words such as “too big” and “too small” into relatable terms for the purpose of FE mesh sizing. The suggested convergence testing for both global analysis and for localised areas of high-stress concentrations coupled with localised refinement of the mesh, merges well with similar concepts that were discussed by the CEE in today’s main video including localised mesh refinement options such as “adaptive mesh refinement” techniques, and the theoretical infinite stresses at points of singularity.
This article is an ideal summary in a “Kurzgesagt” fashion on the introduction and basics of the FE mesh sizing, accuracy, required computational time and processing power, convergence testing to help the designer or the user to make informed judgment and decisions on the final adopted size of mesh for their model, or for their project. Of course, there are multitude of other factors such as geometry, constraints, loading, etc. which would require more in-depth investigations. But this article right here, is without doubt or questions, a very good starting point.
By the way, I started watching some of the videos from the Kurzgesagt Channel and what I can say is ‘man, oh man!'🥰; there are some interesting points of views to life, to science, and to every other thing(s) out there. I can only now begin to grasp why you would find their (Kurzgesagt) work fascinating.
Their video on “How to Make a Kurzgesagt Video in 1200 Hours or more” for example, had made me pause and I couldn’t help but wonder how in the word is the CEE and Mr. Editor even begin to manage to produce such top-notch videos in 7days x 24hrs = 168 hrs or less?😳😳🤔
I have said it before and I will say it again: If you (CEE) were ever in doubt, know that you are doing exceptionally well with your videos. I am and always have been in awe 👏💯👏.
Keep up the good work sir!
thnx a lot Engr. DK ^_^
Happy you liked the reference. Editor told me he is watching videos about how to improve the presentation of CEE videos to be a little bit more "action packed" and less monotonic. He is still learning so I do not expect anything soon.
Kurzgesagt is amaaaaazing
Great video! Your pedagogical skills are on another level. Due to this, I benefit greatly from your videos, despite the fact that I have already taken challenging courses(in my opinion) in the university on this topic. I know you have been an TA in some classes, however, I've had several professors that I have had a really hard time learning from, so all the credit to you, it is clear that you have a talent for teaching. This, combined with your background (having an academic background from PhD work in addition to experience as a working engineer(if I have caught that right)) truly makes this an exceptional channel for engineers of every level.
Furthermore, I am always curious of what I am missing out on when considering other software. Currently, I have used a lot of time trying to really master Robot and its features.
I find that the most valuable skills and information are those that can be applied across a wide range of software. As it seems that you have experience within different types of software, in addition to the fact that you are creating your own FE-program, it would be great to get your opinion on how Robot performs compared to other. What are the ups and downs with Robot? Actually, I just recently found out that Robot offers volumetric elements(i.e. still not anywhere near of "mastering Robot" xD), and was really surprised. And I started to wonder what distinguishes Robot from software programs like Abaqus and Ansys(which I view as the cornerstones of the FEM, but have very little experience with). Do you have some thoughts about this?
Thank you for your videos!
Thnx a lot ^_^, I do appreciate your comment and feedback.
About some general thoughts on Autodesk Robot, I would split the comment into two parts:
Part 1: the comparison of robot as a structural analysis software in comparison to regular structural analysis softwares. Here I am talking about ETABS, SAP, STAAD, Prokon, AxisVM and Dlubal (there is more, but those are what came into my mind).
Here Robot has multiple advantages: first of all, it is really easy to use and the controls are kind of trivial. Even when you mess something up (as is always the case in my videos -_-), Robot is there to the rescue and provides you with multiple ways to "save the day".
Furthermore, its ability to mesh is second to none.
Also, the seamless switch between analysis, design, drawing, calculation notes is quite outstanding.
Furthermore, BIM seems to be getting increasingly interesting to the general Engineering profession. Here, Autodesk Revit is a resonating software. Now what does robot do with revit? a lot. Since Robot is an Autodesk product, there is a Revit/Robot integration, which is helpful in case you are working in a design firm that does implement Revit for a multi-disciplinary team.
Now about the comparison of Robot with fully fledged research softwares such as AnSYS, Abaqus, LS-Dyna and the liking, here we are talking about two totally different softwares. Now note that AnSYS can do ANYTHING robot can do in the FEM department, but does not have a lot to offer in practical outputs such as drawings, calculation notes and so on.
AnSYS has a much much much much deeper FEM basis, allowing users to dive deep into non-linearity, dynamics, buckling problems, contact problems in ways Robot (and all application-based softwares) would never do.
Since you seem to be interested in AnSYS and Abaqus, I might detect some sort of "positive interest" or fascination in those cool results those research softwares provide. Maybe you are interested in a research career? if this is the case, then Autodesk robot would not be the best option for doing research.
I will touch on the volumetric elements in future videos, but they are still a little bit limited when compared to research-level softwares.
Finally,
Indeed, I was a teaching assistant in 2010, then a lecturer in 2012, then since 2016, an assistant professor
Thank you for your response! Ofcourse, im sorry about missing out on the lecturer and Assistant professor part. I do think you have mentioned this before, so my bad for messing up.
I do indeed have a interest for the more «advanced» softwares. In my master thesis i used this program called DIANA, which was great for representing different nonlinear properties of concrete. Atm. I have started working as an engineer, and do like this as well. However, your videos keeps poking into fields that makes me want to do more research again😄 Anyway, i really appriciate your videos and that you take the time to reply the comments.
Most welcome 👍
Stay tuned for more content
Great video! waiting for the next one! GG
thnx for the comment,
Indeed, GG no re ^_^
Hi, good video. At around 23mins 20 secs in the video you speak about high concentrated stressed areas, burning areas. Is there a rule of thumb of how much you move away from these highly concentrated areas when interpreting moments and shears? Even when applying reduced forces over columns and walls? As this is always very difficult to interpret, and even when interpreting concentrated reinforcement areas over the heads of columns? Should it be taken at column face perhaps rather than centre of column node? Also, a video of how to interpret such results would be good, as there can be a huge difference between moments and and shears, and reinforcement areas required directly over column heads and those a short distance away, just a suggestion….keep up the go work!👍
Hi there, first of all, a huge thnx to your comment.
My own little rule of thumb is to move half the width of the column away from its center. But I will be talking about that in the next video, coz I found that it might be too long for this video to be included.
So, your suggestion is highly regarded and I will take it into account the next video (or maybe the one after it).
However, I do not just want to shift the answer for a next video, you can read a little bit about impossible convergence in
A first Course in the Finite Element Method (Daryl L Logan)
Concentrated or Point Loads and Infinite Stress page 360
Anyways, thank you very much for your comment. Consider supporting the channel by suggesting it to your colleagues. It helps immensely.
Regards,
CEE
Dear CEE, let me tell you that your videos are getting better and better!
A Few things i want to disscuss:
- in collegue i was told to use in slab models FE size about 0,5-1 m. Do you agree?
- also, to use emitters or “radial” FE in corners and singularities such hollows or high loads.
- whats wrapling in meshing?
As always thank you very much!
Hi there dear Engr. Candea,
thnx a lot for your comment. I do actually feel larger responsibility with the increasing number of subscribers. I am actually learning how to more effectively video edit my videos and make them more interesting. Once I am at a certain level, then I will try it out.
Now, to your questions:
about mesh size, here is an article www.masterseries.com/blog/fe-mesh-size
But to summarize, span / 10 or 1000mm, so in case of normal spans between 3 to 8 meters, you would end up with 0.3-0.8
so your college approximation is quite on spot.
And I agree with emitters, but to a certain extend, you saw what happened if you go too far.
Finally, about mesh wrapling, I am not 100% sure about it to be honest. Do you mean "warping"? Warping in that context is how much a FEM element departs from being a plane. Now in normal FEM, the warping should be zero because we are assuming that our elements are plane elements. This is EXTREMELY important because we do have an assumption about "plane stress" or "plane strain", with plane being the operative word.
Now there could be some interesting research in 3D warped elements that do not follow a plane, but in the regular sense, a FEM will break it into triangles that are always planes (a triangle is guaranteed to be planar).
Hope I was able to answer your inquiry.
Regards,
CEE
@@CivilEngineeringEssentials indeed they are! Crystal clear i must say thank you for everything again CEE. Hoping to watch the FEM advanced options from RSAP. Regards from spain
Hola Amigo
No hablas espanol (thats all i know)😁
Stay tuned
Hello brother I really enjoy your video
Most welcome. Stay tuned for more content. Also, if possible, suggest this channel to your friends.
Hi dear CEE, theres an option in SAP2000 "automatic frame mesh" where you can set the minimum number of segments in bars when analyzing the structure, they act like nodes but they arent, so you can provide more realistic modal analysis results as accurate frequencies and so on... is there something like that in robot aswell? or the only thing we can do is to generate nodes without creating new bars with the division tool for all the bars before running calculation? As always thank you so much! Saludos
Hi there Engr. Candea,
yep, you can create nodes without actually breaking the bar. (It is gonna be 1 bar, but multiple nodes on it)
I think it was in (Edit -> Divide) Generate Nodes without dividing
However, I have to mention something from the "insides of Robot". If you actually go to:
Right Click -> Display -> Members -> Number of Calculation Elements
then you will see that robot "behind the scenes" generated FEM elements.
It is a loooooooong programming gibberish, but here is the quicker version:
1) You have "geometric elements" those are the things that you define, such as "members and lines" with their regular number
2) You have "FEM elements" that you do not define, but are generated for you during the "model generation stage". Those hidden elements can only be shown given the display trick
The only reason I know this, is because I program my own FEM software ^_^
Still, not to derail your question,
yes, via the divide command, you can add nodes in the mesh of a linear element without breaking it into multiple pieces "you actually do, but the interface hides those"
Regards,
CEE
@@CivilEngineeringEssentials awesome i will check those! Thank you very much