Needed Devlopments in Constraints

[JST]

Old fox...

That is all very well, but unfortunately is not how it works.

Measuring does not work for aligns, see above. Alibre uses undisplayed numbers for determining the alignment. That is the basic point of this argument. Two distances that both measure the same as far as anything you can SEE in Alibre, may STILL not align, the two distances are NOT the same as far as Alibre is concerned.

Your use of the offset is unclear.

If your offset is to be used for an alignment (you did not seem to do this in the example) then that is a "cheat", which does not actually make a constraint to the part in two dimensions.

 

[oldfox]

Alibre uses undisplayed numbers for determining the alignment. That is the basic point of this argument.

OK. Then that is my misunderstanding. I thought the idea was to get 2 parts constrained so they were "locked" to each other. That is all
my example was showing.

Measuring does not work for aligns, see above.

I didn't "measure" I used the Alibre "Measurement Tool" to return the dimension and then I just copied that. (for later pasting)
Then I pasted that measurement onto the part to which I wanted to connect to. All of the holes aligned without further ado.
(still confused)

The midpoint of the offset line, in this example, is where the hole will be punched for that final align. The distance of the offset is
the copied number from the "Measurement Tool", which is the diagonal length.

If your offset is to be used for an alignment (you did not seem to do this in the example) then that is a "cheat", which does not actually make a constraint to the part in two dimensions.

That is exactly what I did. I offset the bottom hole in the diagonal by the distance (pasted) from the dimension that the Measurement
Tool returned. (copied)

Both parts, the 90 degree corner piece and the straight diagonal piece, are constrained to each other. What more are you looking for?

 

[Lew_Merrick]

Oldfox -- I guess that I do not see an "issue here. At the Assembly level, Constructing and Locating the 3rd (or would you consider it the 2nd) attach hole is a non-event. When you place the "2nd Diagonal," you (A) Align the (if you will) Lower Left Attach Hole with the "L-Base Attach Hole," Place a Reference CL Axis in the "Lower Vertical Attach Hole," and Align a Reference CL Plane in the 2ns Diagonal to establish Constraint-based positioning. Then Project to Sketch the the "Lower Vertical Attach Hole" to the Diagonal and Extrude Cut Through. ???

 

[oldfox]

Oldfox -- I guess that I do not see an "issue here. At the Assembly level, Constructing and Locating the 3rd (or would you consider it the 2nd) attach hole is a non-event. When you place the "2nd Diagonal," you (A) Align the (if you will) Lower Left Attach Hole with the "L-Base Attach Hole," Place a Reference CL Axis in the "Lower Vertical Attach Hole," and Align a Reference CL Plane in the 2ns Diagonal to establish Constraint-based positioning. Then Project to Sketch the the "Lower Vertical Attach Hole" to the Diagonal and Extrude Cut Through. ???

As far as I'm concerned, there is no issue here. Certainly not on my part. The way I did the assembly seems to be a lot easier and a lot
less work than doing it your way.

 

[Lew_Merrick]

As far as I'm concerned, there is no issue here. Certainly not on my part. The way I did the assembly seems to be a lot easier and a lot less work than doing it your way.

Not knowing what you had as "input data" and how you had to develop it, my way avoids (for the most part) those issues.

 

[oldfox]

how you had to develop it, my way avoids (for the most part) those issues.

As far as development, I just pulled some arbitrary numbers out of the ether and went from there until I had 2 parts to
constrain together. They both constrained fine, NO ISSUES. So what issues are you avoiding.

 

[JST]

You DID misunderstand again.

I donlt know why this is so hard, but I will try once more to explain it.

When I said "measure", I mean what you did. That displays 6 places, BUT alibre uses many more to determine the distance for evaluating an align.

So your 6 places CAN BE OK in SOME cases, but in others, it is not, and Alibre will not evaluate the two visually identical numbers as the same.

Yes, the "project to sketch" CAN place a hole that will align.

BUT, take the case of the 1/32" grid parts, which WILL NOT be made to some precise number at 19 places. You will, if you model the part as it will ACTUALLY BE MADE, find there is no way to align.

Your two holes are, for instance, 12' 5.934563044" actual distance apart, just due to the way the existing parts fit together. The shop will make that to the next even increment of 1/32, NOT to the actual distance. It will be made to 12' 5 15/16"

Since that is what they will actually do, you make the model at that distance. The other distance you leave, because that is what the assembly is, you do not get to change that, it comes from the parts and the geometry.

Now, with the 'absolutely exact only" alignments, you can only use a "cheat" to get an alignment, even though you think that it would fit. (or can spend an hour which you do not have, calculating to check on all the parts involved).

CHEAT #1.... you do what Lew proposes, and just project the hole.

That will work, but is not centered on where the shop will locate the hole. You know there is an error. You are accepting a model that does not represent the part as it will be, JUST FOR THESAKE OF GETTING A CONSTRAINT. Precisely what Lew and Bigseb complained about, a fix which is ONLY "cosmetic".... lipstick on a pig, which does nothing but "look nice".

CHEAT #2..... You make the part to the actual shop size. It will not align so you use an align with the side of the part, plus an offset, or some other such method to make the part sit where it LOOKS LIKE IT IS ALIGNED (but actually is not).

That will work much of the time, but does not constrain the parts in two dimensions as a hole align does, instead, it allows one part to shift lengthwise relative to the others. If the assembly is later modified, that so-called "constraint" may fall apart and cause trouble. This also is just lipstick on a pig which is simply to "look nice".

The IDEAL "NO CHEATING" METHOD:
If there was a "tolerant alignment", one could set the allowed misalignment to the difference between the hole radius and the bolt radius (which is the normal allowance for steelwork) and then Alibre would ONLY scream about misalignment if the error of position exceeded the allowed tolerance. Anything smaller would allow the assembly, and is perfectly acceptable.

Note that this is NOT A CHEAT. It is based on the ACTUAL TOLERANCE. You get actual information, verification that the parts really WILL fit together as-designed, with the tolerance automatically considered.

So, instead of "cheating" as the other methods actually do, the "tolerant alignment" provides an assurance that there really is no problem, as well as making an actual alignment, a constraint that will not fall apart.

Not something that you would use on every part, but a very valuable tool for the cases where you need it.

 

[bigseb]

Where do I start...

When I said "measure", I mean what you did. That displays 6 places, BUT alibre uses many more to determine the distance for evaluating an align.

So your 6 places CAN BE OK in SOME cases, but in others, it is not, and Alibre will not evaluate the two visually identical numbers as the same.

I have already addressed how to do this. If you persist in doing it in a way that doesn't give the required result... well....

Your two holes are, for instance, 12' 5.934563044" actual distance apart, just due to the way the existing parts fit together. The shop will make that to the next even increment of 1/32, NOT to the actual distance. It will be made to 12' 5 15/16"

If that were my business I would find a new fabrication shop asap. First time ever that I have heard of this.

Since that is what they will actually do, you make the model at that distance.

Again... unbelievable

CHEAT #1.... you do what Lew proposes, and just project the hole.

That will work, but is not centered on where the shop will locate the hole. You know there is an error. You are accepting a model that does not represent the part as it will be, JUST FOR THESAKE OF GETTING A CONSTRAINT. Precisely what Lew and Bigseb complained about, a fix which is ONLY "cosmetic".... lipstick on a pig, which does nothing but "look nice".

Your model doesn't represent the part because your fabrication shop is just plain sloppy. Designs are precise for a reason. This is not about cosmetics, its about engineering integrity. Your finished product does not need to be 'precision engineering' for it to be properly planned. Going by your description it sounds incredibly slapdash.

CHEAT #2..... You make the part to the actual shop size. It will not align so you use an align with the side of the part, plus an offset, or some other such method to make the part sit where it LOOKS LIKE IT IS ALIGNED (but actually is not).

And it never will be, nor does it matter apparently. You keep banging on about constraints and alignments but why even bother? Again, going by your description of the process you use it just doesn't matter.

The IDEAL "NO CHEATING" METHOD:
If there was a "tolerant alignment", one could set the allowed misalignment to the difference between the hole radius and the bolt radius (which is the normal allowance for steelwork) and then Alibre would ONLY scream about misalignment if the error of position exceeded the allowed tolerance. Anything smaller would allow the assembly, and is perfectly acceptable.

To my eyes this is the ultimate cheat. You want to purposely design something incorrectly because of sloppy downline processes. And this Alibre should address? As I said before, Alibre has far more important issues to address than this.

Note that this is NOT A CHEAT. It is based on the ACTUAL TOLERANCE. You get actual information, verification that the parts really WILL fit together as-designed, with the tolerance automatically considered.

So, instead of "cheating" as the other methods actually do, the "tolerant alignment" provides an assurance that there really is no problem, as well as making an actual alignment, a constraint that will not fall apart.

My advice: design it properly, tolerance it properly and the fabrication shop must work to spec. Also this is the first time ever in my life that I have heard the designer say' "its designed within tolerance." Designers design bang on and the manufacturer gets allowances within which to work.

 

[jfleming]

If that were my business I would find a new fabrication shop asap. First time ever I have heard that I have heard of this.

I've been following along silently... What JST describes (1/32" or even 1/16" tape measurements) is in fact how beams/channels/etc. are manufactured in the steel construction world. It is not uncommon or unheard of and is a very common process.

I fully agree with JST and his issue here. I used to draw up steel members at a previous job using AutoCAD and am very fortunate that in my new role, I don't have to deal with what JST is experiencing. I've had to do a few things like what he has described, at length here, only to be just as frustrated with the tool (Alibre/Geomagic).

 

[oldfox]

I didn't use any *displayed* numbers in that number. Nor did I *input* any numbers. This referring to the diagonal distance. Just COPY
and PASTE.
Alibre takes what ever number for the COPY and uses that same number for the PASTE, no matter how many places it is.
If everything is designed dead on, everyone is on the same page and the shop folks can use any size drills they want to. Not your
problem. Or at least it shouldn't be.

CHEAT #1.... you do what Lew proposes, and just project the hole.

That will work, but is not centered on where the shop will locate the hole. You know there is an error. You are accepting a model that does not represent the part as it will be, JUST FOR THESAKE OF GETTING A CONSTRAINT. Precisely what Lew and Bigseb complained about, a fix which is ONLY "cosmetic".... lipstick on a pig, which does nothing but "look nice".

So if the shop isn't drilling on the center of the hole, how do you know in advance which direction they are going to miss it and by how much?

Now, with the 'absolutely exact only" alignments, you can only use a "cheat" to get an alignment, even though you think that it would fit. (or can spend an hour which you do not have, calculating to check on all the parts involved).

I didn't use any cheat, unless you consider copying and pasting a "cheat".

CHEAT #2..... You make the part to the actual shop size. It will not align so you use an align with the side of the part, plus an offset, or some other such method to make the part sit where it LOOKS LIKE IT IS ALIGNED (but actually is not).

If the actual shop size is *accurate* then it will align if it is cut accurate. (within tolerance)

The IDEAL "NO CHEATING" METHOD:
If there was a "tolerant alignment", one could set the allowed misalignment to the difference between the hole radius and the bolt radius (which is the normal allowance for steelwork) and then Alibre would ONLY scream about misalignment if the error of position exceeded the allowed tolerance. Anything smaller would allow the assembly, and is perfectly acceptable.

I have no problem with this. Only any shoddy workmanship after the part is designed.
If your client will accept an allowed misalignment then go for it.

The IDEAL "NO CHEATING" METHOD:
If there was a "tolerant alignment", one could set the allowed misalignment to the difference between the hole radius and the bolt radius (which is the normal allowance for steelwork) and then Alibre would ONLY scream about misalignment if the error of position exceeded the allowed tolerance. Anything smaller would allow the assembly, and is perfectly acceptable.

Note that this is NOT A CHEAT. It is based on the ACTUAL TOLERANCE. You get actual information, verification that the parts really WILL fit together as-designed, with the tolerance automatically considered.

So, instead of "cheating" as the other methods actually do, the "tolerant alignment" provides an assurance that there really is no problem, as well as making an actual alignment, a constraint that will not fall apart.

If I were the PM or the shop boss and something didn't fit and I was losing time and money, hogging out or fileing holes, I would have the draftsman AND the shop worker, both, in front of my desk to find out what the problem was. Would you want that?What would
your defense be?

My advice: design it properly, tolerance it properly and the fabrication shop must work to spec. Also this is the first time ever in my life that I have heard the designer say' "its designed within tolerance." Designers design bang on and the manufacturer gets allowances within which to work.

+1000

CHEAT #1.... you do what Lew proposes, and just project the hole.

That will work, but is not centered on where the shop will locate the hole. You know there is an error.

OOPS. Look at the assembly again. 2 very simple sketches and 6 constraints and NO cheats. I still say that is easier than Lew's way.

easier = faster = more profit

 

[bigseb]

I've been following along silently... What JST describes (1/32" or even 1/16" tape measurements) is in fact how beams/channels/etc. are manufactured in the steel construction world. It is not uncommon or unheard of and is a very common process.

I fully agree with JST and his issue here. I used to draw up steel members at a previous job using AutoCAD and am very fortunate that in my new role, I don't have to deal with what JST is experiencing. I've had to do a few things like what he has described, at length here, only to be just as frustrated with the tool (Alibre/Geomagic).

If your smallest increment is 1/32" (about 0.8mm) then rounding is not the problem. The chance that holes will line up at such spacing is slim (read: very very slim). So use a different constraint and make sure the holes are big enough that the overlap allows the bolt to go through. Because this is purely cosmetic.

 

[simonb65]

If your smallest increment is 1/32" (about 0.8mm) then rounding is not the problem. The chance that holes will line up at such spacing is slim (read: very very slim). So use a different constraint and make sure the holes are big enough that the overlap allows the bolt to go through. Because this is purely cosmetic.

+100

 

[JST]

You guys are TOTALLY UNBELIEVEABLE.

You have never been NEAR a steel fab joint in your life, apparently. (aside from jfleming. He "gets it") Thank you, Mr Fleming.

The engineer stamped prints specify that all holes SHALL BE 1/16" oversize from the bolt. SHALL BE..... not "if you prefer you can make them anything you like.....".

Would it really have made a difference if the number were a smaller tolerance? Absolutely not, but if you insist on your opinions, which are not on topic, go for it.

"Design it properly"... which is of course exactly what is being done, within the parameters specified by the customer.....

"Designed within tolerance"..... YES, DESIGNED IN ALIBRE "within tolerance"...... the parts as designed line up within the allowed tolerance. I assumed that was so obvious nobody could misunderstand. But you did.

Moderator PLEASE CLOSE THIS THREAD.

 

[bigseb]

*moderator hat on*

Dude, this is Lew's thread. You can't just have it closed....

 

[Thompson]

If your smallest increment is 1/32" (about 0.8mm) then rounding is not the problem. The chance that holes will line up at such spacing is slim (read: very very slim). So use a different constraint and make sure the holes are big enough that the overlap allows the bolt to go through. Because this is purely cosmetic.

Um, isn't that what the request was - having a "different constraint"? It's hard to figure out where to start about this. I'll keep it short - my pointy-haired boss (thinks hypotenuse is a large African animal) is lurking around my office door with a stack of paperwork, giving me the evil eye.
If I specified a hole position to within plus or minus 1/32" in a 40 foot beam just to make my model work, I'd get laughed out of the office, then fired. That totally unnecessary precision is very expensive.

If I took a 'bought in' or in-stock beam and redesigned it just to make the model hang together where there would be no necessity to do so in the real world; indeed it might be dangerous - I'd be in trouble.

Sure, there are work arounds - I use them occasionally in SW. But time is money, and the fewer gyrations one has to go through to get the drawing package done, the better. JST and jfleming tell the truth. These may be problems _you_ don't have, but it's a big beat-to-fit, paint-to-match world out there. The easily it is to model it in Alibre, the better they will do.

 

[bigseb]

That's all fair enough but since it going to off anyway and accuracy clearly isn't an issue then simply use the existing constraints. You guys want to 'align' two holes that don't actually align. Why not jsut use the angle constraint. Same result. Instead you want a whole new align constriant (that doesn't align).

Have you seen the 'suggest a feature page? Tons of stuff that AD needs a lot sooner than this 'cheat' constraint.

 

[JST]

DUDE:
The moderator already threatened to close it.

I am encouraging him to close it. The level of non-comprehension is such that it is starting to seem to be a deliberate personal attack.

That's all fair enough but since it going to off anyway and accuracy clearly isn't an issue then simply use the existing constraints. You guys want to 'align' two holes that don't actually align. Why not jsut use the angle constraint. Same result. Instead you want a whole new align constriant (that doesn't align).

Have you seen the 'suggest a feature page? Tons of stuff that AD needs a lot sooner than this 'cheat' constraint.

OK ... let's run through the facts again.....

All CAD is a lie and a deception. That is inevitable in the fact that it assumes every part is precise. Everything is made precisely to some size, and if a part varies by one digit in the nth decimal place from that, it is impossible to align in Alibre. That is of course a lie told by Alibre, it is asserting a gross error when the issue is of a magnitude that is inconsequential in anyone's world.

But there does need to be a decision made as to what the error amount is that is considered a problem. That decision IS made by most CAD systems as "precision or nothing". It simplifies the calculations, so it is a practical solution. And that usually is fine.

Parts are not precise, and in particular, structural steel is not precise. it's hard to be precise over a 40 foot beam anyway. They use a 1/32 grid because that is the finest division the tape can be read to. Some use 1/16", go read jfleming's post.

First. Steelwork is not done to please your prejudices. GET USED TO THAT. YES, it is done to what any machinist would consider loose tolerances. GET USED TO THAT TOO. If you do not want to, just go and tell the fab shop how stupid they are. I suggest you do it before your vacation so you have time to recover.....

This constraint is the OPPOSITE of a cheat. You need to wrap your head around that fact and stop insisting on it being a cheat.

This proposal is a "SPECIFIED TOLERANCE constraint". That actually has a purpose. You put in the actual tolerance of the alignment, and if the result is within that tolerance, it will align and constrain, JUST LIKE THE REAL WORLD PART DOES.

IF it is NOT within tolerance, the constraint will still fail, and by failing, it tells you that you have a problem.

Please explain how that is a cheat. It obviously is not one, so you are now just being insulting by insisting it is one. If anything, it AVOIDS the need to cheat.

Mind, right now Alibre tells a lie by refusing to align and constrain parts that are well within their alignment tolerance. The inability to constrain is built into the need for absolute exactness. Alibre is simply lying, and turning the issue over to you, but offering no way to do anything about the problem that is not in itself a CHEAT.

Everything you folks have suggested IS in fact an out and out CHEAT, just a cosmetic cover-up of a non-alignment that you decide to ignore. You have no idea what it is, you have no idea if it is in tolerance. Your CHEATING method is exactly what you blame this idea for being.

Of course Alibre actually actively PREVENTS alignments and constraints, by using numbers that you cannot see or enter to evaluate alignment, but that is a separate problem.

 

[Thompson]

That's all fair enough but since it going to off anyway and accuracy clearly isn't an issue then simply use the existing constraints. You guys want to 'align' two holes that don't actually align. Why not jsut use the angle constraint. Same result. Instead you want a whole new align constriant (that doesn't align).

Have you seen the 'suggest a feature page? Tons of stuff that AD needs a lot sooner than this 'cheat' constraint.

"accuracy clearly isn't an issue" Well, yes - in this particular case, that's basically true.

"then simply use the existing constraints" Well, OK - It's been pointed out by both you and me that workarounds are often available. But those workarounds are often unfortunately complicated or fragile. A 'one click' (or so) solution would be desirable.

"You guys want to 'align' two holes that don't actually align." This may be the crux of our communication problem. You seem to be misunderstanding the scenario. No one wants to align two holes that won't align. Faced with the problem of two holes that cannot align, it would be helpful to have a mate that would link those holes in such a way that the model would not fly apart when poked. See previous comment about workarounds.

"Why not jsut use the angle constraint." Why, yes, that would possibly be a good workaround. Kind of confuses the design intent, but desperate times sometimes require desperate measures.

"Instead you want a whole new align constriant (that doesn't align)" YES!!! We have a winner! A constraint just like real life. Personally, I think it would be nice, but I'm not sure it wouldn't cause more problems than it solved and I'm OK with using the occasional workaround, provided it's not excessively complicated - I just hate wasting the time screwing around with things like that.

"Have you seen the 'suggest a feature page?" Yes, I have. For several years, now. I think I'll wait and see how the next couple of updates go before suggesting anything. I don't want to rush in and seem ungrateful.

"Tons of stuff that AD needs a lot sooner than this 'cheat' constraint." Truer words have never been spoken. I like Alibre, but it make me sigh a lot.

 

[bigseb]

DUDE:
The moderator already threatened to close it. I am encouraging him to close it. The level of non-comprehension is such that it is starting to seem to be a deliberate personal attack.

Just to clarify:
- This forum has three moderators: myself, DavidJ and Swertel.
- Another moderator threatened to close it because of certain comments that were made, not because you requested it.

 

[DavidJ]

Calm down guys - JST suggested a new type of constraint ('approach as close as possible to aligned, and tell me what the mismatch is, or tell me if it is within certain bounds' - my paraphrasing).

Some see that as a useful concept, others don't believe it is needed. Fine - you can all have your opinions.

By all means offer alternative workflows - constructively please. Also please accept that what is 'correct' from your perspective may be a real PITA for another. Repeatedly and aggressively re-stating your opinion/position generally doesn't move the discussion forward.

This forum is NOT where any decisions will be made on new features - so 'winning the argument' should not be the goal.

I refer all to Max's recent post here - https://alibreforum.com/forum/index.php?threads/forum-moderation.20220/

I'd suggest that there has been enough discussion (or pointless argument) about JST's feature suggestion. Let's move on.