A puzzle for the "exact design" believers (evidence for toleranced alignment)

[JST]

Here we have a tower extension. This is steel construction, so everything MUST be on 1/32" grid at the finest. No dimension may be anything other than an exact 1/32" increment. (because you want EXACT DESIGN)

All the bolting plates must be identical and symmetrical. All the (steel angle) diagonals must be identical.

The diagonals at the ends go from topmost hole of top plate to topmost of next plate on adjacent pipe. The middle ones go between the closer holes, and the bottom is a mirror of the top. There is a set of three steel angles that go around the top, sharing holes with the diagonals.

The pipes must be exactly 4' 0" on centers.

Your task is to find a set of dimensions for plates and diagonal braces that will end up with all dimensions on exact 1/32" increments, AND have all holes line up for exact alignments that Alibre will accept.

No complaining about the requirements, that is none of your concern or business.

All alignments MUST be acceptable alignment within Alibre, no fudging the alignment with a dimension to the side of the part or other non-EXACT design.

I'm betting that you will not be able to do it within the requirements, and that you will have to CHEAT an alignment.

Of course, with the toleranced alignment I proposed, you would be able to do everything and know that all lengths were good within the normal 1/32" all-around bolt-hole tolerance

(Note that the top girt measures 41.375000 for both pieces and yet it still does not allow alignment. This is a fault in Alibre.)

 

[oldfox]

How about some pictures for us non-ironworkers?

 

[bigseb]

Where do I start?

- Sketches, feature, constraints NOT renamed
- Only default planes used, not user defined ones.
- No material density

 

[bigseb]

Just kidding. You're alright, JST. No matter what they say.

 

[bigseb]

 

[bigseb]

To your post: you're right. Within that specific increment of 1/32 it'll never constrain porperly. That's how triangles work. My suggestion is to constrain the crossbar using one align and one angle instead of two aligns.

 

[JST]

So you are telling me to cheat? to make "FAKE CONSTRAINTS"? Dude, we have been through this, and you guys made it really ugly. Let's not, let's look st this rationally.

You folks (plural) asked me to cheat the model, and then you turned around and called my proposed solution to that problem a cheat that had no place in a serious CAD program. Which one is the real "cheat", the "visually OK model", or the one with a toleranced constraint?

So, here is a real-world case, taken from an actual job specification set. As you see and ADMIT, it cannot work precisely with the required dimensions. A couple of us told you that last time.

We explained that the layouts are done to a coarse grid, due to errors of reading a tape. And a tape etc is what is used. Even the CNC hole punching machines do not have a lot better tolerance than what we say. My guys accept 1/32". The other fellow mentioned 1/16" minimum increment, which is even worse.

That is just the real world with steel fabrication. Holes for 1/2" bolts are punched or drilled to 9/16", and other sizes similarly, for the specific purpose of letting somewhat imprecise dimensions line up. Buildings and structures are just not made to millionths of an inch.... nor thousandths of an inch either.

What I/we want is for Alibre to provide the tool that is needed to make it work without "cheating". We want to be able, in at least selected cases, to INPUT THE TOLERANCE and, if the lineup is within that tolerance, to allow it to constrain, so that models do not "blow up" when changes are made , etc.

As in this case... Without that type constraint, I had to cheat those parts into what is just a "visual alignment", so really if anything gets changed, that may all be out the window. The cheat is valid, because the error is less than the tolerance, but they all must be checked again manually if any change is made.

Not so with the toleranced constraints. If the result is still within the tolerance set, all is well, JUST AS IT WOULD BE WITH THE ACTUAL STRUCTURE. I would not have to go and verify each one manually, one by one.

Conversely, if a change forces the lineup to be outside the specified tolerance, it would generate an overconstrained error just as any constraint does now if the constraint cannot be made without violating a different constraint.

Naturally, it is up to the user to set the tolerance to something meaningful and useful. I see no problem with that, it is no more demanding that asking that constraints be made legitimately in models now.

This would make Alibre really SHINE for setting up steel fab... I do not know if Autodesk "Revit" allows that., but I suspect it may, because it is used so commonly for construction type work.

They brag about how it takes the design to steel fab.... sounds as if they might, I went looking for that info, but I did not find it yet. It may not be possible to discover without using the program.

*****************************************************************

As for the "general bitching" about named this and that, etc..... these models are specifically one-time use, for fab drawings that will be used once and filed away. The actual drawings are reviewed by a PE, and if stamped, then what these models have or do not have, is totally irrelevant to anyone.

It would be a really POOR USE OF TIME to treat these models as if they were for some product that will be updated and referred to many times by different people over the next several years.

I get 2 or 3 days to make sense out of rough and sometimes wrong engineer's drawings and create valid fab drawings. Academically correct CAD best practice can go hang. I have done many very complex structures, and when the parts were made to print, they fit and built the structure.

 

[RCH_Projects]

Years earlier on Alibre, I was a novice with a somewhat unique assembly. It was, however, designed for motion.

As I recall I could click drag parts for motion (and/or use the API with Excel). But at some points in motion the whole thing would lock up and constraints jump to red all over. In Excel I even created a jiggle routine to nudge the parts around which actually restored the constraints and motion (until the next lockup). When I finally started thinking it wasn't newbie errors I asked support about it.

I seem to recall the issue cleared up in a patch release (although it recurred in another later release and was patched again as I recall.)

My Point?

It is not as simple for yourself. But could you try to hit the "regenerate" assembly multiple times". It was "one" of the "jiggle" tricks I coded into excel but I don't think the other tricks would apply for your case.

I don't know if it remotely applies to your needs! I never knew what Alibre did (if anything) to tighten or loosen in the "patches" that helped my situation. I presume that if there is an obscure parameter they could/did adjust, that "too much" could have been worse than "too little".

Please try the "jiggle/regen". It seems unlikely but if it accomplishes anything (I coded about 3 tries) a quick Wizoscript can invoke it. NOT the solution anyone wants, but I target solutions closest at hand first.

Send the model to Alibre with a link to this conversation. Maybe they can do (if they ever did) an adjustment that helps. Can they be criticized if they make it work and the standard definition is that it cannot be done by "proper" convention?

I'm about to regenerate much of that old work in 2018; I guess I better get to it and see if "I" need jiggling again for that matter.

(Muttering to self - always issues, issues,... issues...)

Always best wishes and regards,

Roger

 

[JST]

The "jiggle" is not applicable. Parts are not the same dimension, even though in some cases you have no way to know that.

They CAN be criticised for allowing two parts with identical displayed dimensions to be treated as being different lengths, which they probably ARE down in the 12th decimal place. You have no tool to fix that problem, you are left S.O.L.

The case of steelwork is different, that is a needed enhancement, which, as mentioned, competitors have made at least partial fixes for (SWX slot alignment, etc).

 

[bigseb]

I still think expecting Alibre to create whole new constraint is a bit much, considering what else still needs to be fixed. If, as you say, it really doesn't matter then why even bother? Align one end and just line up the other by eye. Its not like parametrics matter here.

 

[JST]

I still think expecting Alibre to create whole new constraint is a bit much, considering what else still needs to be fixed. If, as you say, it really doesn't matter then why even bother? Align one end and just line up the other by eye. Its not like parametrics matter here.

I do NOT say it does not matter. Parametrics DO matter, if I catch your drift.

It makes the job of doing steel structure design more difficult, and OTHER COMPETITIVE CAD HAS IT, or something very much like it. SWX has a similar constraint, I am told (maybe by you?). Steel is not the only thing it is useful for.

Do I say drop everything and do it now? no. But it should be on the list. There are many other things that the capability for such a constraint can open up.

Suggesting anything out of the ordinary here seems to get a response on the order of "Ah tol' Wilbur, an ah tol' Or-ville, ah tol' them boys thet thang will nevvah fly."

Now, allowing two parts to NOT constrain as equal dimensions, when both DISPLAY as identical lengths at maximum resolution....? YES, THAT needs to be fixed pronto. It's an error in [programming, without a doubt.

 

[DavidJ]

I thought we learnt in the previous thread on this subject that you guys simply aren't going to agree. The points have been made on both 'sides' - surely we don't need to repeat it all again here.

 

[JST]

I thought we learnt in the previous thread on this subject that you guys simply aren't going to agree. The points have been made on both 'sides' - surely we don't need to repeat it all again here.

The point was that here I have a tangible example, one that came up in current client work. Thankfully, one or two others have stayed out of this.

I think Bigseb does understand the issue, his disagreement is that there are bigger problems ahead of this, a point which I am in full agreement about.

But here is an opportunity, and a concrete example of a case where things just are never going to work well without the suggested solution, or something much like it.

 

[DavidJ]

We understood last time what you were asking for, and for what type of situation.

The thread seems to have re-started the 'debate' (which doesn't need to be had).

 

[Markaj]

In reality you would put bigger holes in to give a bit of slop.
Maybe some kind of restraint for keeping a small hole within the circumference of a big hole?
I spend a lot of my working day putting holes in things to get them to line up - a bit of 'slop' in constraints would be a useful feature where 'thereabouts' is whats needed. Obviously it depends what line of work your in....!

 

[RCH_Projects]

FWIW

The following (used as a sub-assembly) is an example of a "designed" constraint that appears to achieve required goals. It is not to the specifications spoken of, I would leave that to the refinements of actual users. I used hollow balls for example. The number of balls as well.

With one end inserted into each of the holes (mated to the surface most distant from each other - offset by a specified distance varied by material thickness) of mated materials it (can) allow travel within any given circumference.

Just an observation on my part. If used it can be "hidden" for aesthetic purposes.


If anyone makes use of the approach please do a tutorial.

Questions?, observations? - a new thread please. I am very abbreviated in this "approach" because it could cover a lot of territory and the final user can learn something from their own implementation anyway.

BallsOut.AD_PKG
Video_2018-04-21_185721.zip

 

[RCH_Projects]

BTW - great application for "Global parameter" sub-assembly!

Thickness, Circumference(?), Tolerance = ball size - +20 points to best solution(s).

Disguised as a bolt head or rivets?

Let the contest begin, and show results in the gallery, how about that?
I'm gone!

 

[bigseb]

So you are telling me to cheat? to make "FAKE CONSTRAINTS"? Dude, we have been through this, and you guys made it really ugly. Let's not, let's look st this rationally.

You folks (plural) asked me to cheat the model, and then you turned around and called my proposed solution to that problem a cheat that had no place in a serious CAD program. Which one is the real "cheat", the "visually OK model", or the one with a toleranced constraint?

It isn't a cheat though. Angular constraints are perfectly credible constraints. And you yourself said that its just for initial alignment and that it'll never be used again. That's why I said parametrics don't matter here.

But David is right, this has been discussed to death. Perhaps down the road, when AD is working smoothly this can be revisited.

 

[RCH_Projects]

Aww, I was going to win a (my own) conceptual design contest.
<delete>

 

[JST]

We understood last time what you were asking for, and for what type of situation.

.

Obviously NOT the case... or there would be no debate.
I am not going to debate the point. If someone does not have the understanding to "get it" there is nothing I/we can do to help them.

I am providing an example simply to illustrate the reason for having the feature (which OTHER programs have, at least in part). What people think about it is their problem.

But, although I did renew recently, I may change that decision next time, to assist in getting a program that actually does what needs to be done.

From what I have been told, SWX has a feature that will function like it, and Revit also has something similar. Just was talking to an architect friend today about that.