1 2 3 First Previous 104 Replies Latest reply on Jun 9, 2018 4:48 PM by Matt Lombard Branched to a new discussion.

    modeling methods

    Matt Lombard

      Its easy to get caught in a trap and model yourself into a corner in SW, especially if you buy into all of the methods they demonstrate and teach for making links between files. If you're a super expert, you can sometimes get yourself out of some of these tangles without having to redo everything, but there is a lot of model rebuilding going on.

       

      I'm curious about how many of you have an actual structured method for building models, creating links between documents, in-context methods, and parent/child relations in general. I've been reading and writing about John Stoltzfus's SSP, about Gebhard's RMS, and other methods like what Pro/E people used to teach in the early 90s, and Horizontal modeling, skeleton and master modeling.

       

      Do you follow some process that is formally documented, or follow it informally? Do you just rely on your wits to get out of any bind you create? Do you destroy all of the intelligence in your model at the end of a project (break all links)? Do you have formally structured best practice documents that you follow and are enforced, or are you less rigid about it? If you have success or failure with your method, please describe.

        • 1. Re: modeling methods
          Steve Calvert

          I'll jump to the end here.  Yes, I break references when the design is close to the end and it's still mainly because of PDM and not knowing how it may affect us.

           

          There's nobody here including myself that looks at the "modeling" BUT we do all have design reviews and the CAD people will ask questions about how and why.

           

          I myself have been using more in context modeling mainly because of the latest fuel dispenser was done that way and it's just easier in the long run.  I'd love to really get my mind around master modeling one day so I can use that to teach as well.

           

          Steve C

          • 2. Re: modeling methods
            Alex Lachance

            Hey Matt,

             

             

            We currently use the skeleton/master modeling technique and it is not documented right now, though we are working on documenting it per products we develop. I understand very well how SolidWorks works and it's logic so I rarely run into a problem while working. I don't break links if I have them, I lock/unlock them depending on my needs. It makes it easier to modify afterwards. If I do break the link for some reason, I'll remate it accordingly so that I can modify it quickly without having to switch back and forth between assemblies to understand it's intent.

             

            As I said, we aren't documented right now but it is in the works. Me and another guy have been working this way for 2 years and we're now working on documenting it to hand it out inside the company.

             

            Some good practices to have:

            • If you don't have to do an external reference, don't do one. In the example I will post below, there is no external references but the center axle could have been one. I didn't put an external reference because it doesn't really require one. I do have to remember to change the dimension if I do make some modifications on it tough.
            • I've noticed that assemblies that aren't ''fixed'' will always rebuild themselves, even if they are entirely mated plane to plane or origin to origin. I started fixing my assemblies/skeleton parts after mating them by the origin and noticed a big improvement in rebuild times.
            • Find a good way to identify your skeletons. We started adding suffixes at the end of file names to identify our skeleton parts.
            • Do not use configurations if you are going to insert a part inside a part. They are very buggy.

             

             

             

            Example:

            • 3. Re: modeling methods
              Dwight Livingston

              Matt

               

              We have a modeling standard with various rules about making references. Generally they are discouraged unless the need is overwhelming. One must get permission from the project lead to create an SSP. Here is a key paragraph for using master model technique . . .

               

              • A reference to a master model shall not be broken, be ignored, or refer to a part model that is not in the SAP production database. If a valid and active master model reference cannot be maintained through the life of a derived part, either that part shall be rebuilt from scratch or the master model references shall be removed and the part model repaired.

               

              We have the same paragraph for SSP. Our design environment, with team member turnover and long product lives, makes it hard to keep design intent alive and relevant. Too much entropy.

               

              Dwight

              • 4. Re: modeling methods
                Dan Golthing

                One thing is for sure...everybody has their own way of skinning a cat.

                 

                I see how some people model and am appalled at how much lack of forethought there is in their design intent.

                 

                On the other hand, unless you're doing a similar task over and over, the design approach is always going to vary.

                 

                I did one part with around 300 features in the tree once.  The CAD department was complaining as of the complexity and lack of design intent.  I told them they were welcome to remodel the part.  I had spent months designing one of the most complicated parts I had ever modeled going though multiple iterations to arrive at a final solution, and for what was to be a prototype.  Was I going to spend what would have probably been a week cleaning it up and risking the part blowing up?  No.

                 

                If the part was to move to production, then yes, having someone remodel the part would be a good idea.  But this part was crazy, with imported spreadsheets for defining involute curves and so on.

                 

                Simpler is better.  Can anyone after you use the part and make changes?  Or are there such advanced techniques used that you and only two other people on the planet can understand how to change it?

                 

                Symmetry!  Start with the fricken origin at the middle of the part!!  How is that not obvious?  do not start a base sketch with the rectangle corner at the origin.  This is amateur hour stuff.

                 

                Break all links!!   Yes there are probably exceptions to the rule, but external references are a recipe for disaster.  Everybody in an organization must understand those links and file management becomes paramount.  If you can break the links then you remove the possibility of something blowing up.

                 

                Fully-defined sketches.  Again, this amateur hour stuff.  If your sketches show blue, then a looser are you. 

                 

                Finish the design!  Make sure the model is finished if there is a chance anyone else is going to have to work with it.  that means that there's no errors, broken mates, etc.  Purge junk sketches, configurations, etc.  clean up your mess already.

                 

                The worst messes I get myself into are really related to major weaknesses in SolidWorks.  For instance, doing advanced weldments, people will know that if you go up in the tree and try to add or remove features from a weldment feature, there's a good chance it's going to blow up most of the rest of the tree.  There's often not much you can do about it.  SolidWorks hasn't built in the necessary intelligence yet to dangle the problems, it just simply craps the bed when it comes to this.  No amount of preparation can avoid this flaw.

                 

                Same thing in many cases when you want to delete a feature higher up in the tree and it ends up deleting the children.  Sometimes it allows you to dangle the references but sometimes that option is NOT available and it's a lot of work to manually break all the children links to the parent feature that you want to delete. 

                 

                I'm not sure trying to say there is any overreaching method that is superior to another is possible.  I would think that it's all application specific.  Every industry is a bit different, and everybody's design flow varies.  Telling someone who is incredibly creative as a mechanical designer that they have to follow a set of rules as they design is just a way of stifling their creativity. 

                • 5. Re: modeling methods
                  Alex Lachance

                  Dan Golthing a écrit:

                   

                   

                  Simpler is better. Can anyone after you use the part and make changes? Or are there such advanced techniques used that you and only two other people on the planet can understand how to change it?

                   

                  Symmetry! Start with the fricken origin at the middle of the part!! How is that not obvious? do not start a base sketch with the rectangle corner at the origin. This is amateur hour stuff.

                   

                   

                  Fully-defined sketches. Again, this amateur hour stuff. If your sketches show blue, then a looser are you.

                   

                  Finish the design! Make sure the model is finished if there is a chance anyone else is going to have to work with it. that means that there's no errors, broken mates, etc. Purge junk sketches, configurations, etc. clean up your mess already.

                   

                  These are some good tips that I haven't mentionned!

                  Dan Golthing a écrit:

                   

                  Same thing in many cases when you want to delete a feature higher up in the tree and it ends up deleting the children. Sometimes it allows you to dangle the references but sometimes that option is NOT available and it's a lot of work to manually break all the children links to the parent feature that you want to delete.

                  Hey Dan,

                   

                  is this box grayed out for you sometimes? I usually use this setting to fix what you speak of but it's sometimes an annoyance because if you delete a part and forget to retick the box then the part will be deleted but the mates will remain.

                  • 6. Re: modeling methods
                    Chris Saller

                    We don't break links to the 'master' (and no, it's not racist; unless you want it to be).

                    I don't. Keep it for a different discussion.

                    Anyway, I keep everything intact. All model are thought ahead for CAM milling, origins in the middle of the part, etc.

                    Having the origin in the middle of the part makes life easier for assy alignment and CAM programming.

                    All materials are assigned for future FEA, mass, etc.

                    All files are checked into the PDM vault, all users are trained to use the PDM vault.

                    All templates/settings are shared from the same server folder.

                    Parts are modeled to nominal, or mean, whichever nomenclature you use.

                    Fully define the sketches!

                    • 7. Re: modeling methods
                      Vladimir Urazhdin
                      Do you destroy all of the intelligence in your model at the end of a project (break all links)?

                      I'm just wondering - what it could be the reason to destroy links and/or intelligence when all done?

                      To make sure nobody (customer included) will be able to use it as master model for repetitive design?

                      • 8. Re: modeling methods
                        Matt Lombard

                        Hmm interesting. So you don't reuse projects when they are done?

                        • 9. Re: modeling methods
                          Matt Lombard

                          Well, that's the question. Some people believe there is some advantage. I think if you have taken the time to build the intelligence, design intent, or links, whatever you call it, it should remain. Reusable data is valuable. People have reasons for destroying the links, which usually have to do with fear of something changing.

                          • 10. Re: modeling methods
                            Vladimir Urazhdin

                            In my company the activity such destroying links is the punishable "crime"

                            • 11. Re: modeling methods
                              Dan Golthing

                              There are often times where that isn't enough to save the children...SAVE THE CHILDREN FOR HEAVEN'S SAKE!!!

                              • 12. Re: modeling methods
                                Steve Calvert

                                "Projects" no, but we reuse as much as possible.  PDM is great...

                                 

                                Steve C

                                • 13. Re: modeling methods
                                  Dan Golthing

                                  Linking one part to another can be a two-edged sword.  You almost always have to develop in context, but that context doesn't have to remain once the part is finished.

                                   

                                  If you want to add the out-of-context features back in down the road, it's usually not difficult. 

                                   

                                  But if your 10,000 part assembly blows up because one of 20,000 out of context relationships has somehow changed...now that's a problem.  And when you're dealing with assemblies in the thousands to tens of thousands of parts, all having external references, nobody can keep it all at the forefront of their mind as they go in and start changing things.

                                   

                                  Eliminating potential cherry bombs in advance is typically a good idea. 

                                  • 14. Re: modeling methods
                                    Vladimir Urazhdin

                                    I'm not a fan of context linking technique.

                                    Indeed - context links are hard to recognize and usually well hidden. Easy to get screwed.

                                    We link via Global Variables and Skeleton sketches - pretty much transparent way with comments available.

                                    1 2 3 First Previous