1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

20072 going off the rails again

Discussion in 'Using Alibre Design' started by JST, Jan 11, 2020.

  1. JST

    JST Alibre Super User


    David suggested that it needs a good starting point to know what to do. So....

    I first did an align of the OD of the half pipe to the centerline of the solid round, to get the half pipe into a good starting position.

    Then I deleted the align, and applied the tangent using the ID if the half pipe and the OD of the SR. Should have been the best of all possible start points, easy movement directly to tangent position, etc.

    The result was what you see, the tangent pulled the half pipe out at an angle, which does not seem to satisfy the requirements of "tangency".

    Do not think this is tangent.jpg

    Attached Files:

  2. DavidJ

    DavidJ Alibre Super User Staff Member

    I also suggested that you need to constrain things enough to limit the number of possible solutions. If there are many 'possible solutions', I'm not even sure that you'll get a valid one displayed as the solver is just confused.

    Why did you remove the initial align...?

    Attached Files:

    Last edited: Jan 13, 2020
  3. JST

    JST Alibre Super User

    The align would have prevented the tangent, as it aligns the centerlines of the objects, and the nearest size pipe is not a perfect fit on the SR. Tangent requires them to touch, or be in a position to and the align would hold them away from each other.

    A flat side can be tangent to a cylinder and yet be miles away. I do not think a curved surface can be in the same way, it can only slide along the axis, spin around it, and move around the axis as the contact line moves around the inner part. These are not even on a common axis.

    I have abandoned the use of "tangent" in this case, I need to get work done, and I think the illustration will be good enough on the drawing this way. I am not sure what special gyrations you used to get it right as you showed me. Mating base planes of the two parts?.

    As for the other solutions, I do not think either of the ending positions IS a correct solution. If surfaces of cylinder OD and tube ID are tangent, the very first consideration is that they should be so at all points along... That is what Tangent did the couple times it actually worked. I have NEVER had a tangent in Alibre do this stuff before, and I have used this same tangent at least a dozen times in this sort of construction before.

    Those may have been with the old ACIS, and were definitely with older versions of Alibre, without the new HOOPS (that has other known issues).
  4. DavidJ

    DavidJ Alibre Super User Staff Member

    'Special Gyrations' - show reference geometry of the half pipe, constrain one of the principle planes of half pipe to a principle plane of the tower leg. This limits the number of possible solutions for the tangent constraint. Choice of the planes depends on where you want the parts to touch.

    An alternative (as shown in your response from support) is to use align with an offset. That still needs an additional constraint to fix the solution, but at least shows a possible solution before finally being fixed.
  5. JST

    JST Alibre Super User

    I do not want to appear to disregard your advice, which is no doubt good advice. My biggest issue is that

    A) it seems that what worked before no longer does, at least not without extra stuff not previously required

    B) There is enough just plain "squirrely" stuff going on to cause a person to reasonably blame ALL of what is seen on the new version. It makes it more reasonable to discount "user error" issues.

    And, I HAVE been using the program for at least 8 years. This does not eliminate "brain farts", I am at least as susceptible as anyone else. But it does suggest that when I find things different, they may BE different.

    Have not seen those ideas from support, but DID try variations of what has been suggested and what you report they said. Most variations resulted in error messages, although the applied constraints should not have prevented the movements suggested. These were not "no solution " messages, but "overconstrained" messages, indicating a different level of issue.

    I settled on just an "align" before I saw your responses. The offset they apparently suggested I just discounted as not actually required, I can compensate.

    An orient does not prevent an align or a tangent, or should not, but it did. "Did sometimes", I should say. Same thing might work fine after a "save" operation.

    And, I am using workflows I have used many times, and now they do not work half or more of the time. This seems wrong. Was the prior workflow wrong? Did I just get lucky a dozen or more times?

    Mixed up with this is the problem of "unselectable surfaces", which is associated with "disappearing parts" The latest of that was a number of parts that would disappear when manually moved. The invisible ones would reappear for a moment when the next part was selected, bt only for a flash.

    Without being really rude here, I plainly do not have time for this nonsense. I am getting paid for what I do, and something very odd is making well-known and customary workflows flat-out not work correctly.

    I am having to develop work-arounds for things that should not be going on.

    It is preventing me from doing what I need to do, and hence preventing me from getting paid. I do not like that.

    Forgive be for being unappreciative of your response, but things I normally do and that worked in the past, are not working, and this is a "tad bit of a problem" here.

    Just to recap a few recent issues:

    Transparent parts with new graphics, have to use legacy.... one of many such that have been reported

    unselectable parts, requiring a store, or quit and restart program to get back.

    Disappearing parts when zoomed.... associated with disappearing parts in "most" but not all cases, and NOT requiring any adjustment of facets etc to fix, just a program restart or sometimes a save operation.

    An apparent change in the operation of tangent, or at least a vast increase in the criticality of initial position.

    Apparent creation of tangency that does not appear tangent.

    All this on what should be a more than adequate computer.

    I expect there are more. I have to quit for a meeting soon, so wanted to leave this response as hopefully a clarification of what I am seeing.

    Final one..... "unhandled exception" when closing Alibre.....??????
    Last edited: Jan 13, 2020
  6. Max

    Max Administrator Staff Member

    JST, the model you uploaded is correct - mathematical tangency was created. The problem is there are infinitely many ways the orientation you have can achieve tangency with no further constraints. You view tangency in your mind as "tangent where the solids also do not intersect", but that is not the mathematical definition which ACIS is using. Different types of combinations of geometry have more or less possible outcomes. A sphere-sphere tangency is very predictable on how it will behave - sometimes a cylindrical face - cylindrical face gives a solution that is correct, but that you might not expect, without further constraints.

    What you need to do is apply an orient to the 2 front faces, and then apply a tangent constraint.


    Cylinder-Cylinder tangency can be confusing because it applies a tangent constraint to the infinite mathematical surface, some of which you can see (aka - your model's cylindrical face). These 2 faces are tangent:


    How is that possible? Well, they have a .06 degree angle between them, and somewhere off in the ether these cylinders, or rather the infinite mathematical model of the faces, meet. For this type of constraint you often need to be a bit more bounded in how you ask the solver to function - this is achieved by adding more constraints - an orient in this case being optimal as you force the inter-angle value to be 0, which gives you tangency as you would expect.
    Last edited: Jan 13, 2020
    Jake_Steidy likes this.
  7. JST

    JST Alibre Super User

    Ah, yes, the off in infinity problem.

    Perhaps the simpler (for the user) approach would be to ask the user to select either a "tangent crossing" or a "tangent aligned". Those are likely the two main cases that would be used in CAD. it also avoids issues in cases where there is no convenient locating surface, and the inherent part geometry is not helpful due to odd angles, etc.

    the issue here is that

    1) applying the orient made the tangent refuse to work, with an over-constraint. No discernible reason for it, measurements said it was not so....

    2) The prior use of same tangent was much more predictable in the result.... I never had any issue with the results in prior ACIS. I presume that the new ACIS is more "rigorous" in its definitions. That may explain a number of things, but one could with they were less academic and more practical in defining it.

    3) when I applied an "orient" AFTER the tangent, that application snapped the tangent to a new place, after it had done what I wanted to begin with, and I wanted the half-pipe in a particular orientation around the SR..

    I am reasonably conversant with geometry, but did not foresee that condition, and surely did not foresee it being selected when the parts were already oriented as desired, and the simplest way to satisfy the "orient" was mere rotation.

    Others may be unfamiliar with the whole idea. Making it less academically correct may annoy the math folks, but it may be much more handy for the rest of us who just want to do work.
  8. Max

    Max Administrator Staff Member

    No argument there. I'm not sure ACIS offers much on this front, but perhaps there is something that could be done outside of them.
  9. JST

    JST Alibre Super User

    One other question:

    The behavior has apparently CHANGED. In several ways, things that used to work fine, no longer do, or seem to work slightly differently.

    The difference seems to be in details, in the area of the "tricks" needed to get good results. Obviously not in gross characteristics or operation

    Is this due to the new version of ACIS?
  10. Max

    Max Administrator Staff Member

    It's hard to make a blanket statement without seeing any issues visually. It is quite hard to follow verbal descriptions in a wall of text about 7 different things.

    I can say we haven't changed tangent constraint behavior. I might suggest getting a free screen recorder and using that to show the issues you are seeing.
  11. DavidJ

    DavidJ Alibre Super User Staff Member

    I have re-checked this in v2018.2 : The behaviour of tangent constraint with a solid rod and a 'half pipe' of slightly larger internal radius then the rod is identical to that reported here.
  12. GIOV

    GIOV Senior Member

    I agree with JST.
    I did exactly the same model in three AD different versions. The best for design: AD11.2, The best for achievement in performance without memory lak and messages issues specially shell feature: AD13 and the last one AD2019 I didn't achieve the final design due the several warning message, specially shell one.
    JST give very good advice to me in this forum so his statements must be considered seriously.

Share This Page