40 Replies Latest reply on Oct 18, 2011 8:08 AM by Bill McEachern

    welded aluminum walkway simulation

    Dave Scott

      I want to run a simulation on a 60 foot aluminum walkway where the beams are only welded on the side faces and not all around. I split the model up into "parts" thinking this would help I want to select these faces to define contact but its not letting me or i just dont know what i am doing. i cant seem to find the answers in any forums so any help would be great

      thanky you

      Dave

        • Re: welded aluminum walkway simulation
          Ryan Werner

          Hi Dave,

           

          From the picture it appears you are running this analysis using Beams.  I am not totally clear what side faces you are referring to.  Are you trying to constrain the ends of the model?  Anyway, if you are running this model as a Beam analysis than you will not be able to constrain any faces, you are only able to constrain nodes in a Beam analysis so the first step is to get your nodes set up the way you want them.  Then apply you constraints to them.

           

          Ryan W.

            • Re: welded aluminum walkway simulation
              Dave Scott

              Hi Ryan

              not sure what i am doing then. do i treat each body as a truss?

              i go thru the simulation advisor and get to the part of selecting faces to weld and i get the run around of questions and always end up at the start. It seams to default the walkway as a beam? the faces i want to will are the inner and outer faces of the walkway post which are 3x3 sq tube. not all around

              Dave

                • Re: welded aluminum walkway simulation
                  Ryan Werner

                  Dave,

                   

                  First off, I should have asked if you are using a full version of Simulation or just Express.  I have never used the advisor before so I am not sure what you are referring to exactly when you say you are selecting faces to weld.

                   

                  If you built this structure as a weldment and used structural members than by default SW will treat all the bodies as Beams.  For your purpose this does makes sense plus it will solve much quicker.  You could treat all the bodies as solids and then you could choose faces to constrain.  What are you trying to determine with this analysis?  That might help pinpoint the issue you are having.

                   

                  Ryan W.

                    • Re: welded aluminum walkway simulation
                      Dave Scott

                      I am using a full version of simulation, i did build it as a weldment using structural members.

                      I am trying to detremine a factor of safety since people will be walking down this and it will be free spanning 60 feet.

                      im joining all the 3x3x1/8 sq together with welds and i am not welding it all the way around at every joint only 2 welds per joint, opposite faces. im trying to attach a picture but i am not winning that one now either

                      Dave

                        • Re: welded aluminum walkway simulation
                          Ryan Werner

                          Well I can maybe save you some time there then because a Beam analysis will not take into account how the ends are welded together.  It just assumes they come together at a node.  You can refine the connection at the node some but you cannot do any kind of weld analysis at the node.  Beams is a very simplistic way to view a model and is more for gettingquick results over large truss like structures.  If you want detailed weld analysis I know there is a feature in 2011 (and I assume 2012) that allows you to define a weld between two bodies and then do an analysis strictly on that connection, however, the pieces must be defined as shells I believe.  I have not been able to use this feature since I am still stuck using 2010 thanks to corporate bureaucracy.  Plus I am not sure how well it would work in a larger structure such as yours.  If you want to look more closely at connections in particular you can treat all bodies as solids and add features to mimic your welds but that can be pretty time consuming.  You can also choose one connection and model it separately and then run an analysis on it by itself using forces determined from an analysis of the structure as a whole.  Hope some of this helps.

                           

                          Ryan W.

                          • Re: welded aluminum walkway simulation
                            Bill McEachern

                            Hey Dave,

                            Ryan is on the ball. You use the beam analysis to get the loads at the joint. You should also know that SWX beam analysis is not their strongest suit by a long shot so ensure that you believe your analysis results. Many people have found many issues with this aspect of the code. Since it sounds like people could get hurt or killed be sure the analysis is good enough and you understand what's up. With that said, find the joint with the worst loading - these will be resolvable on beam center lines then make a free body diagram of the joint and figure out the loads in the welds and use a manual method to assess the welded joints. The welded joint function Ryan has mentioned has also had its share of issues - like bad results at non normal angles. No idea if it has been addressed yet or not. You could do a solid analysis on hte joint but again be judicous in asssessing the results though this aspect of the code tends to work pretty well. These analysis need to be done with some care in set up and interpretation.

                             

                            No idea what you mean by "only on the side faces" but in general you want the welds loaded in shear and not in tension with throat depths and lengths sufficient to get the job done, Given this is aluminum - not sure which one but in the welded state it may have substantially reduced strengths in the as welded condition. And fatigue of welded Al joints is also something to pay attention to. You are going to want to aim for stress in the welds in the very low 1000's of psi or less depending on what the material is, how long it is supposed to last, what the load spectrum is and other things. You might want to consider a marine alloy in the 5000 series for corrsion and weldability . Good luck.

                    • Re: welded aluminum walkway simulation
                      John Stoltzfus

                      Good afternoon Dave,

                       

                      I have no experience in Walk Way design or simulation, however one thing popped out at me;  Welded Aluminum??  I would think you have chosen Aluminum for weight, but do you realize how far the expansion and contraction is going to be with a temperature change of 80 degrees.  Can't you use a clamp design rather than welding, aluminum welds tend to fracture at the weld root.  With all the possible linear movement, even as a manufacturer I would shy away from this project, unless it is designed in a way that movement won't put undue stress in the assembly.

                       

                      Just my two cents,

                       

                      John

                      • Re: welded aluminum walkway simulation
                        Billy Wight

                        Unless you have a supercomputer, solid elements are not the way to go here.  Even with 2nd order tetrahedral elements (SW default), I would want at least 3 across the thickness of all components in order to trust the results.  With something that large and more than likely with small wall thicknesses (can't tell from the pdf), that's a lot of elements.  If beams are not good enough (and they probalby aren't for the level of detail you want), you should use shell elements.  Be sure to take into account the as-welded strength (T0) as others have mentioned, and design for fatigue, not yield, if this is to be used regularly.  Good luck!

                          • Re: welded aluminum walkway simulation
                            Bill McEachern

                            With all due respect, to Billy Wright,  the 3 elements thru the "thickness" needs to be viewed thru the lens of practicality. For global bending of a beam three elements thru the beam X-section is sufficient and would likely occur in the coarseist of meshes. 3 elemetns thru the wall is likely insufficient in the complex state of stress at a joint in the above truss but is surely better than less than 3 elements.

                             

                            Further, on the above truss and given that it is made from Al I would assume the design approach is to put the angled members into compression and thus welding on the cut faces would put the joint loads in shear and bearing - a decidedly good way to go in my humble opinion. The mesh problems are likely due to the the fact that the members have flat angled cuts that leave a nice notch for a fillet weld in the direction of the longeron. To get a mesh, you need to model this area with the weld material in the notch. So make a cut on the angled member along the direction of the longeron so that it is coped to the longeron such that contact is only made where the weld material is applied (leave a very small gap on the top/bottom surface of the longeron - but don't leave a small face that will require a very small element size to resolve). However, in the global analysis, putting a ton of elements at any or all of these joints will be a ton of work and of dubious value. Once the analysis is complete, use the forces optiion and get the loads transmitted across the faces that are on the angled brace in contact with the longeron and design the weld accordingly with conventional methods. You only need to find the highest one and assess it. You could essentially do the same from a beam element analysis (my earlier suggestion). The forces computed in the solid element analysis should be checked to ensure they do not vary a lot with mesh density - in my expereince it matters, believe it or not.  If you want to examine the detailed state of stress in a joint, just take a single joint and mesh it to death and apply the loads using a suitable free body diagram - or leave the rest as a beam element model - this is easy but does require some knowledge of how to pull this off. Too much to describe here and sometimes it doesn't really work as the beams don not have the right stiffness at the beam to solid connection - well at least I have expereinced this with the code.  But be advised that the occurance of stress singularities may arise in a linear analysis at re-entrant corners - nodes of infinite stress the finer the mesh goes. One approach maybe to examine the elemental stresses and or average the nodal stresses in the neighborhood. Or you could do a NL analysis and let the sigularities bleed out to a finite value. This is avoided if the loads approach and conventional weld design is used as mentioned above.

                             

                            A reasonable intro book on FEA is Practical Stress Analysis with Finite Elements 2nd ed. by Bryan J. MacDonald - while it is far from a comprehensive treatment for the practicing engineer it does cover the basics pretty well and is code independent. It also has the attribute of being quite inexpensive.

                              • Re: welded aluminum walkway simulation
                                Billy Wight

                                Bill,

                                     Notice I said "at least" 3 elements across the thickness.  You will always want to ensure you have solution convergence by refining mesh sizing in areas of high stress (as you mention in your post).  I regularly see thin plate like structures incorrectly solid meshed in SolidWorks end up with only 1 element across the thickness, and usually with terrible aspect ratios, especially when the structures are very large (as in this case) and the user has relatively low amounts of computing power (RAM).  Scary stuff!  These types of simulations are just begging for shells or, as you mention, beams with detailed 3D analysis in areas of concern.  The problem with making FEA so accesable is that it makes it very easy to get impressive looking contour plots with meaningless results.

                                  • Re: welded aluminum walkway simulation
                                    Bill McEachern

                                    Hi Billy,

                                    I noticed all that was said. I was just pointing out that for global beam bending stress and gross deflections you need 3 elements thru the X-section, not thru the wall thickness. If you are after something more local then yeah 3 thru the thickness is more appropriate. Even with moderately bad aspect ratio's the bending stress with 3 thru the X-section will still resolve fairly well. Depending on what you are after and what the model is solids in SWX sim can be very efficient for beam type analysis as Ryan has pointed out. You just need to be very "heads up" on what you are looking at.

                                     

                                    I do not believe there is any problem whatsoever with making FEA mroe accessible. As a fairly experienced guy at this sort of stuff the avoidance of drudgery is a very welcome developement in FEA, at least by me. The more established players, like ANSYS, ABAQUS, MSC and others should do a lot more to allieve their users of it. As opposed to saying the stuff is hard  and so are our interfaces........The thought that making it easy is the same as making it dangerous is just out and out daft. I get a bit ticked (sorry) at any implication that they are in any way equivalent. The thought that any engineer worth his salt is just going to start "believing" some pretty picture produced by some tool he is relatively new to using implies that they don't understand their professional responsibilites. I find that idea rather disrespectful to the entrie profession and it reeks of eliteism in the skill set. Sure there is a whole new set of questions to consider when you are using this stuff but that doesn't mean you should not give it a whirl. You can't learn how to ski powder unless you get in it. Same is true for a lot of things. I encourage anybody to explore the tool and ask the questions and figure stuff out. Hence my participation in these discussions. You can get a lot by reading books on this subject but clarity will not emerge till you actually start doing it in my humble opinion.

                                     

                                    Sorry really didn't intend to rant.....but from a philisophical perspective any analysis is better than no analysis - just respect what you don't know. And, the more you know, the more you appreicate what you don't know as your exposure to the adjacent possible goes up proportional to the cube, maybe the fourth power of what you do know.

                                    Well, enough of that.

                                      • Re: welded aluminum walkway simulation
                                        Dave Scott

                                        thanks for the input, I am learning to ski poweder as a matter of fact.

                                        I guess it all comes with experience figuring all this stuff out, but how does one know he has actually done it right other than the fact we see the results we want. I would never build anything like this without an engineers approval. i am merely using it as a tool and i am enjoying using it. i thank you all again for your epertise even though some of it goes right over my head

                                        Dave

                                        • Re: welded aluminum walkway simulation
                                          Billy Wight

                                          You are correct Bill, and I agree about the accessability, my post was poorly worded.  I think FEA does need to me more accessable in general, especially with the more advanced packages.  That said, however, I have found that if a user is working with an advanced package, they more than likely have a good understanding of the theroy and general good modeling practices (yes, it's a stereotype).  I have seen a lot of bad analysis performed in SolidWorks and similar packages because they are so accessable, which allows users, who do not have enough experience, to easily run a bad analysis and believe the results are good.  Sure, this isn't the fault of the program so much, it is the fault of the user, but I see it all the time.  It isn't always an engineer doing this; it can be a draftsman, or general designer that is using FEA.  It's perfectly OK for them to dabble with it and learn, and I encourage that, however, basing design decisions on the results without fully understanding them or how they came to be is asking for trouble.  I see this all the time. 

                                           

                                          I would really like to see SolidWork have messages or similar that help to give good suggestions for modeling and keep people on the right track, but I doubt we'll ever see that due to legal implications...  There are some features that have no business being there too; look at the "Mesh and Run" option.  This defaults to whatever mesh settings SW thinks your computer is capable of (more often than not, bad settings) and produces a result!  The mesh is one of the most important aspects driving the solution, bad mesh, bad solution.  When you allow a user to mesh and run, you remove the opportunity to review the mesh, check for bad elements, etc. and get them a result.  This was a poor call on SolidWork's part, all to save a mouse click.  SolidWorks is by far not the worst package from this respect, though, I've seen packages where you can't ever view the mesh, you have to trust that is was done right! 

                                           

                                          You say "from a philisophical perspective any analysis is better than no analysis - just respect what you don't know", which I completely agree with provided that the user respects what they don't know.  I see so many examples where the user did not respect what they didn't know, many times because they didn't know what they didn't know.  This is where I see room for improvement in increasing the accessability of the simulation software. 

                                           

                                          Dave, please don't be detered by my posts, you obviously are respecting what you don't know, and are here willing to learn, which is fantastic.  We all started out at some point with no idea of what we're doing, and you're on the right track.

                                            • Re: welded aluminum walkway simulation
                                              Bill McEachern

                                              Nicely put Billy.

                                               

                                              People often confuse the responsibility of the individual with the responsibilities of the organization. While an individual should strive for an analysis suitable to get the job done it is the organizations ultimate responsibility to ensure that any analysis is used appropraitely. If you are a one man shop you typically know this with a lot more clarity that those embedded in organizations.

                                               

                                              The key question is what are you willing to risk or are risking on this analysis? If we take Dave here as an example, he already knows what works from experience. He is trying to figure out how the other guys are getting away with less. He isn't risking anything until they adopt a new approach with less and that isn't going to happen till the organization is convinced they can adopt "less". By the way Dave, if the other guys are just welding on the side and you are welding all the way round - just build a joint model and test both ways with some representative loads - you will likely find the extra welding is not required but don't take my word for it and the modeling subtelties will come into play no doubt.

                                               

                                              As for the products....I don't think it reasonable to expect a software provider to police and guide a user and point out what they don't know to ensure a suitable analysis.....SWX tries and it drives me crazy as they put it in the way more often than not - the analysis advisor.......who would rely on that bit of engineering prowess? It would be nice if they did tell you how all the stuff worked. I will start ranting on this topic so time to go...