Extruding over features to eliminate 8

[rhmeng]

I was wondering if anyone sees any issues with extruding over features to eliminate them. There comes times when you are deep in a model, and you are unable to delete a certain feature (nor modify it) because it is the parent feature for several other things that you have in your model. Thus, instead of deleting the feature and building everything up again, I was using extrudes to do things such as eliminate fillets, as an example. Sometime during an extrude there is some overlap where the extrude is going through an existing piece of the model. So far I have not seen any issues with this, but I dont know if this can cause errors later when running FEA and meshing, or calculating mass, etc. I know I can test this but I thought it would be a good discussion. Obviously this is bad practice but sometimes speed trumps everything.

 

[TheTick]

It's called "hacksmanship". It's what you do when you really should be doing something else for a living.

 

[rhmeng]

i didnt ask what it was called, i asked if it was ok. too late to go back to college.

 

[JMirisola]

The biggest issue I see with it is for whoever works on the model after you. Having been that person, I can say that speed should never trump good modeling practices as said speed can, and probably will, cause slowdowns in the future. To answer your question about it affecting FEA or mass, no, it shouldn't cause any problems. But, it's still the wrong way to go about it.

Jeff Mirisola, CSWE
My Blog

 

[CorBlimeyLimey]

Obviously this is bad practice but sometimes speed trumps everything

.

It's good that you recognize this as bad practice, but stability should always trump speed.
Like Jeff, having had to work on such bad models I can attest to how frustrating and time wasting it can be.

Having said that, you should not experience problems with FEA or mass calcs.

 

[SBaugh]

One big problem I see besides bad practice, is the drawings that reference a feature that is swallowed up, will either be lost or might stay around and not dangle (seen it happen). Then the drawing is completely wrong. Its horrible practice and can lead to bad or incorrect drawings which usually cause your design to be incorrect and wastes the companies money. Its better to get out of the habit now, before it gets ingrained into doing it all the time.

Scott Baugh, CSWP
Gryphon Environmental

http://www.2gryphon.com

"If it's not broke, Don't fix it!"

faq731-376

 

[rhmeng]

great thanks for all of the input guys. I should have clarified a little further as to why I am doing this. I am predominantly extruding over features in models that I am using for FEA, where I am simplifying the model for meshing reasons/mating etc. Thus the models will never be passed off to some other poor soul or used for 2D drawings. Good point about drawing dimensions referencing swallowed features, didnt think about that and I might go try it out...

 

[TheTick]

I busted your chops a bit because I have made several months' mortgages in billable time undoing this kind of work.

That said, in your case it may be a good thing. Simplifying models for FEA--removing gaps, getting rid of small features--is a good thing. Just don't do it on the model used for design. Save a copy!

 

[CorBlimeyLimey]

rhmeng,
You could also try using the Defeature tool...

http://help.solidworks.com/2014/english/solidworks/sldworks/c_defeature_tool.htm

... and then saving the result as a dumb solid.

 

[Jboggs]

I am known in my department as the one that is the most anal about doing things right, and after reading the above I say that if the only thing you are doing is saving yourself some time, and the model WILL NOT be used for some other purpose later or by some other designer, then I say carry on. But I also say that with this VERY STRONG suggestion: Find out exactly what is causing the problem. This is not good or professional practice at all. It can be extremely frustrating to trace down the actual causes of situations like this, but it is also very valuable in that every single time I have had to things like this down I have learned something I used later to prevent it. Learn the details of Parent/Child relationships. Things like sketch relations, dimension endpoints, sketch planes, extrusion parameters are usually the culprits, and they can all be avoided if you know how.

 

[KevinDeSmet]

A nice tool for defeaturing manually in a relatively clean way is the "Delete Face" command, in particular with the option Delete and Patch. This will remove any face(s) you select and attempt to heal the surrounding geometry in one go. As for it being a bad practice...well...things aren't always the way we like them do be.

 

[davidinindy]

As everyone else has pointed out... Doing it for a simpler FEA model is fine (on a copy)
Doing it on a production model used in drawings is bad...
It obviously, will affect mass properties if you're filling in chamfers and holes...

A former coworker told us how a firm in India took his model, and had to move some holes. Instead of redefining the feature, they extruded a new feature to fill the holes, and created new features where the holes needed to be.

Move bodies is another command I hate dealing with...

Trying to figure out why I can't redefine a feature, only to discover a move body command will move it back no matter where I want it is a big time waster... I've learned to look for them first thing, now, but it's still hard to modify a model with move/delete bodies in the tree.

David
Connect with me on LinkedIn.

http://lnkd.in/fY7-QK

Quote: "If it ain't broke, I must not've fixed it good enough"

 

[Updraft]

I am 100% in agreement about avoiding this practice as it is a bad technique. I love configurations and use them extensively, but not for things like this. There is, however, a way to have your cake and eat it too. As previously suggested, saving the part as a separate file (Part-FEA.sldprt) breaks the connection. True, this will keep the original undisturbed and give you free reign to make the changes to better accommodate the FEA, however, by the OP's definition these changes are deeply embedded in the fabric of the file and making some changes for the benefit of the FEA might very well change the part in unintended ways. The new part can easily become something different from the original in ways more than just the FEA simplification.

I'd like to suggest an alternative approach. Make a new part file from a blank file and Insert->Part as the first feature, selecting the original part file. This will give dynamic linkage to the original part so that if it changes this new file will see those changes. However, in this new file the part appears as a dumb solid. Here you can make (and name) new features to deal with the simplifications you desire. You will not be able to suppress features in the original file nor change dimensions from here, but you will be able to add cuts and fills which will very likely be easier to control and understand.

We use Insert->Part a lot just for these very benefits. It has become an extremely valuable and reliable technique for us.

I hope this helps.

- - -Updraft

 

[cadagnostic]

@Rhmeng, you were right, this made a pretty good discussion!

@Updraft, excellent "cake and eat it too" idea! Thanks for sharing.

11859 Pecos Street
Suite 310

 

[GimpLizard]

Such is the world of feature-based, parametric modeling. As a Solidworks user since 2003 I have come to realize a couple of things.

First; Way too many engineers don't know diddly about CAD modeling. Because they are constantly creating features for no other purpose than to "remove" a previous feature. As a result, their history trees become excessively long and complicated. Trying to interpret their "design intent" is a lesson in futility.

And the second thing I've learned; The software should really be called Solidwrecks. Because, sooner or later, your model is going to crash.

"Go direct. Do not pass Go. Save yourself 200 hundred hours."

 

[bunnifer451]

The whole "good/bad" thing brings to mind a quote... "Only the Sith believe in absolutes." Yes, we would all like to be able to do everything perfectly all of the time, and I don't know about the rest of you, but many times, I have had to deal with others making unrealistic promises, be they engineers, designers, marketing or management. Inevitably, I end up being the one who has to make it happen. On some occasions, I have made similar promises (mea culpa), usually without enjoying a complete understanding of the scope of the project. So, yes, I sometimes make modeling choices I typically would not, and have done exactly what you describe. What I do in these situations is to be completely transparent with my clients, explaining what exactly is required, and yes, I can do it quickly to meet the time constraints, but time will need to be spent after if they want me to correct the "bad" modeling (there's that word again...). I either get a Yea or Nea here and proceed accordingly. If the mistake was mine, I end up eating the time/cost to correct everything. The important issue is to ensure that those responsible are completely informed.

Updraft describes the best workflow for those situations where one does not need to make any changes to the base model. If the model needs to be rebuilt, it needs to be rebuilt, even if you are not the one who began it.

My $.02...

 

[Clauson]

Users of feature- and history-based CAD systems don't always realize that what we see on the screen doen't necessarily reflect what is in the database. We don't "eliminate" a feature by extruding over it or uniting it with something else. It's still there in the database, where it can possibly cause unpredictable issues down the line. Expedience doesn't always pay off.

 

[rstupplebeen]

For messaging a CAD model for FEA I use 'split' to utilize symmetry planes and 'delete face' to get rid of holes and such. I usually label these as FEA.whatever so it's clear its my FEA features.

Updraft's suggestion I will consider in the future though.

Thank you.

Rob Stupplebeen
OptimalDevice.com
My Personal WP

 

[gfbotha]

To correctly model different production stages involving some very specific manufacturing steps, where say both stages do have its corresponding configured drawing, I figured 'extruding over' as a proper way to do it - but, that is only after having started a new part linked via Insert > Part Copy (also see Updraft's post). An example is, modelling a thin walled canister where the open end is later swaged/crimped into a groove in an end closure. The final part (with the sub-surface kink in) drives the required blank canister delivered by the part supplier. One needs to fill up the kink with material, and cut away material at the ID.

As already mentioned by some guys; I do not see (neither ever found) a probem doing it for FEA purposes. But, I then do it as a last resort - if an upstream feature cannot simply be suppressed. Also, sometimes one only want to remove part of an upstream feature..

Regards

 

[rhmeng]

I busted your chops a bit because I have made several months' mortgages in billable time undoing this kind of work.

No worries, I was in on a weekend and not to stoked about it at the time..

You could also try using the Defeature tool...

http://help.solidworks.com/2014/english/solidworks...

... and then saving the result as a dumb solid.

Thanks for this input, I tried this out and it is a nifty option that I never knew about. How is this different than just saving something as a STEP file and then importing it?

It seems to me the consensus is that extruding over features is looked down upon, obviously bad practice, and is "ok" but not ideal way of modifying models to make them compatible for FEA. I, like many of those that posted here, have also been down the long and treacherous road of trying to correct/change a model that has been passed on to me that was headed toward the Solidwrecks crash model [GimpLizard, 2015]. As several of you have stated , Updrafts post on inserting models into a new part is clutch. I tried it out and it works flawlessly, now I have a featureless model that updates to a master copy in my main assembly. There were alot of "shoulds" above if anyone comes across this in the future.. We have done several simulations (shock, vibe, frequency etc.) and have had no issues with any of our models that have the sketchy "extrude over" modeling practices. Thanks for all the input.