Constrain solving - as complicated as that

[Cris-Ideas]

Hello All,

I have started this thread to bring up and hopefully explain problems related to constrain solver and gain knowledge on how exactly constrain solver handles the assembly.

This is because I am experiencing many problems related to assembly solving that should not be expected.

Originally I was discussing this issues in the thread related to problems with flexible assemblies but I decided to split this topics and this thread is intended to focus on constrain solver it self.

I would like to encourage especially people from Autodesk who have the knowledge about constrain solver it self to participate and help us understand better how this works to make our lives easier.

I will be posting videos and data sets, so anyone could try the same what I do and see if is getting the same outcome.

I also would like to ask you to do the same, and if possible use videos with comment rather than long posts, as this is more easy to follow and understand the intent.

I will be also giving each example a name, so it was easy to track replays. So when posting please make it obvious what you are referring to. 

If you are interested in problems related to flexible assemblies please visit this thread Flexibility not working properly in inventor - BUG that has been there for ever

Cris.

[Cris-Ideas]

@johnsonshiue

Hi Johnson,

I believe this may have been on.

BUT

I am not sure if this is the same issue and the one you refer, that has already been logged.

So difference is that in this case I was applying FIRST constrain for the bolt-washers-nut set in context of this stationary plates. So THERE IS NOTHING by this time that restricts position of the set in context of top level assembly. Therefore ANY CONSTRAIN should be possible to apply.

Also if I switch to undirected axial mate it applies constrain as before.

Also I should not be forced to go through application setting and switching them on or off depending on what bugs they cause.

Also I am not sure at all if this application option is to blame. I have recently documented another example showing how the same assembly is "properly" solved or not depending on order of constrain creation.

@SteveMDennis

Hi Steven,

I need to make this remark here.

It was BAD choice to set default setting for new axial mate to something other than undirected if you did not allow default to be set by the user in the same time. I have just found this my self.

I think that when functional, I may like the new function, but initially default should be undirected.

I am sorry to say that but you need to considered users as lazy, not liking to be forced to change people, (As I consider my self).

Average user is not Inventor geek, and is not looking what's new just to have fun trying it. Primary interest of average user is to get job done with a minimum time and effort. 

If he gets "new relies" of the software it is only profitable if it is easier to get this done. If user (I) need to struggle more, and my favourite workflows are no longer as easy as they ware I am unhappy.

And unhappy user is something you do not want.

I am not saying you should not change software. But definitely you should do this more wisely.

@Xun.Zhang

 Hi Xun,

Answering your post from Whats NEW 2019 - NEW BUGS - Helping improve Inventor

---

@Xun.Zhang wrote:

...

It seems a face mate or flush was already exist, how about switch to Aligned option?

In the video, it seems no preview when select opposed which means the constraint chain solve is incorrect, have to switch to alternate solution which is similar with Insert or face mate/flush.

The better idea here is more like auto-predict and sort out a proper resolution for the constraint creation, such as in your case, it's better to pre-select Aligned instead of opposed, does this make sense?

..

---

Well,

unfortunately No.

There is no mate, there is in fact NO constrain between any element of the bolt set and elements fixed it context of top level assembly.

So your suggestions do not make it work.

Design intent is simple: I want to place a bot in the hole. And to be able to flip sides should I desire.

Please see previous posts retarding this assembly, starting in the begging of this thread.

Cris.

[SteveMDennis]

---

@Cris-Ideas wrote:

@SteveMDennis

Hi Steven,

I need to make this remark here.

It was BAD choice to set default setting for new axial mate to something other than undirected if you did not allow default to be set by the user in the same time. I have just found this my self.

I think that when functional, I may like the new function, but initially default should be undirected.

I am sorry to say that but you need to considered users as lazy, not liking to be forced to change people, (As I consider my self).

Average user is not Inventor geek, and is not looking what's new just to have fun trying it. Primary interest of average user is to get job done with a minimum time and effort. 

If he gets "new relies" of the software it is only profitable if it is easier to get this done. If user (I) need to struggle more, and my favourite workflows are no longer as easy as they ware I am unhappy.

And unhappy user is something you do not want.

I am not saying you should not change software. But definitely you should do this more wisely.

---

@Cris-Ideas Believe it or not we debate what should be the default for every change we make quite a lot! Some very heated arguments happen in our office when this topic comes up. MOST of the time I fall on the "leave the default to be legacy behavior". Sometimes I lose the argument, sometimes I win. We cannot and will not make a blanket policy. Each and every new functionality will have the pros and cons evaluated as best we can internally and by asking users in the beta forum, surveys, etc.

In this case I happen to agree with the change because for most users they SHOULD want to have a direction used. Undirected when not fully constrained by other constraints is a recipe for disaster, and that is the main reason we made the decision we made.

From this thread and others we do have some who disagree and it sounds like you now fall into this group. If you look back though we have several users who agree with our decision. We will not always get it right for all of our many users who use different modeling techniques. That is just a fact of life for such a diverse user base that crosses industries, training, and backgrounds...

[Frederick_Law]

May be allow user to change the default?

For new users they use default and develop their work flow from there.

For old users, like me (since R1), would like to stay on the old path until I feel like trying new things.

A setting to change it could help.

I do know thats lots of programming to get that done.

[Cris-Ideas]

@johnsonshiue

---

@johnsonshiue wrote:

Hi Cris,

You got me on this one. I don't think I have an answer to that. The constraint solver is proprietary. I am not sure if we could make such information public. I will work with the project team but I don't think I would get an answer soon.

Many thanks!

---

Hi Johnson,

Could you please briefly indicate which question or post you are referring to when you do.

Also could you comment on post #76?

Cris.

[jeremy_wasserstrass]

As I am still using 2018 my question is how does this work if I place the face/face mate or flush first? Do I then have to go and clear an error if I forget to apply the correct direction or is the program smart enough to make the correct decision on it's own? I happen to fall in with the group who thinks undirected should be default. If I wanted that kind of functionality before I would switch and use a Joint.

[Cris-Ideas]

Hi, I think we should leave DOFs for a moment, until @johnsonshiue will be back with some answers.

In the meantime w can have a little side discussion regarding axial mate, which I hope will not take over whole thread.

But in the subject of constrains.

As promised something entirely different:

#196 bug.

It turns out order creation of constrains DOES MATTER.

It is strange as constrain solver is claimed to be not following any hierarchy of constrains, also no creation order.

I have few examples that seem to indicate otherwise.

This is valid in 2017 and 2019. Applications options and binoculars OFF, so no influence. Just constrain solver in play.

data set for download: https://autode.sk/2Lq34Bc

Cris.

[Xun.Zhang]

---

@Cris-Ideas wrote:

Well,

unfortunately No.

There is no mate, there is in fact NO constrain between any element of the bolt set and elements fixed it context of top level assembly.

So your suggestions do not make it work.

Design intent is simple: I want to place a bot in the hole. And to be able to flip sides should I desire.

Please see previous posts retarding this assembly, starting in the begging of this thread.

Cris. 

---

I don't believe so without data. It's better to show me the sample data in your video.

[Xun.Zhang]

---

 #196 bug.

It turns out order creation of constrains DOES MATTER.

It is strange as constrain solver is claimed to be not following any hierarchy of constrains, also no creation order.

I have few examples that seem to indicate otherwise.

This is valid in 2017 and 2019. Applications options and binoculars OFF, so no influence. Just constrain solver in play. 

data set for download: https://autode.sk/2Lq34Bc

Cris.

---

Hello Cris,

How about Accept the confliction in the warning dialog and then click Rebuild all button to get the constrain solve up to date?

It seems to me more like the axis mate was temporary want to keep the design intention for you if you make the axis mate first, so the confliction warning dialog pops up for your attention.  Otherwise, if face mate first, then, directionless axis mate will be ignored and no kindly warning dialog is needed.

It can be better to have another warning message like do you want to flip the axis mate instead of confliction.

Hope it helps!

[Cris-Ideas]

@Xun.Zhang

Hi,

So problem with solution you propose is that that in complexed cases it leads to the situation nothing is working. Like in the example you saw in this thread Fix it challenge

And in general in this thread and in this case in particular it is not about how to get inventor to solve this, as there are at least few ways how this can be achieved. But here it is used as a very simple example showing how inventor is not capable of solving simple case when we know that solution is obvious.

Now I need to make a general comment on this "flipping" axial mate constrains behaviour, so many are complaining.

So "flipping" in essence is not happening at all because constrain is undirected so it does not apply any restriction on relative orientation of two axes participating in the constrain. So if constrain does not require any particular orientation of axes (except of them being collinear) than both possible orientation alignments (aliened or opposed) are fully legitimate for the solve of this constrain.

So it does not flip. It is only solved so to meet requirements of other constrains that may impose one of two possible alignments of axial mate, or it is just randomly assigned one of two possible solutions in case there is nothing that would indicate one of them as "the right one".

So called "flipping" only refers to user experience of the constrain behaviour in cases when user assumes undirected axial mate imposes also aligned or opposed axes orientation. But this is only caused by lack of understanding what are actual conditions imposed by this type of constrain. And that's only users fault if this so called "flipping" causes any problems in the assembly.

Undirected axial mate DOES NOT flip as both solutions are fully legitimate. If it "flips" for you this means you do not have constrains in place that should fix required orientation of your component.

Cris.

[Cris-Ideas]

@johnsonshiue

Hi Johnson,

Is there any news on what exactly DOF in inventor is?

Cris.

[Xun.Zhang]

Hi @Cris-Ideas,

Thanks for the information.

Will discuss with constraint solver team for sure.

Thanks!

[Cris-Ideas]

---

@Xun.Zhang wrote:

---

---

I don't believe so without data. It's better to show me the sample data in your video.

---

as always data set for download: https://autode.sk/2wJJlYy

Cris.

[Cris-Ideas]

@johnsonshiue

Hi Johnson,

Should we expect any answer regarding "What DOFs in inventor are?"

This is quite important question to answer, as DOFs are used in various places and seem have significant importance  in the software.

Cris.

[johnsonshiue]

Hi Cris,

I have contacted the constraint solver team but I have not got a reply yet. I believe the way Inventor or the constraint solver define degree of freedom should be the same as common understanding. If the DOF analysis reports more or less DOF than it should be, it will be a bug.

Many thanks!

[jtylerbc]

---

@johnsonshiue wrote:

If the DOF analysis reports more or less DOF than it should be, it will be a bug.

---

I can point out a situation where this happens, although it doesn't relate directly to Flexible subassemblies.  The DOF Analysis completely ignores the Symmetry constraint, and always has.  Any degrees of freedom removed by Symmetry constraints will not be accounted for in the analysis. 

It's pretty easy to verify this.  Build an assembly with any two parts, with the intent of centering it around the origin plane.  Add all constraints except for the Symmetry constraint.  Check DOF Analysis, and note that it shows 1 degree of freedom.

Add the Symmetry constraint, then check the DOF Analysis again.  It will still show 1 degree of freedom, even though none of the parts are free to move.  This is true up to at least version 2018 - I have not checked in 2019 to see if it has changed there.

[Cris-Ideas]

---

@johnsonshiue wrote:..,
I believe the way Inventor or the constraint solver define degree of freedom should be the same as common understanding.

..

---

Hi Johnson,

If this would be the case I would not ask this question.

It is not working as that in inventor. Just get back to your answer saying (not a direct quote):  "nut is moving with a bolt so it does not have DOFs".

Also look in to results on DOFs analysis in any assembly that has moving components that are constrained in sets (nuts and washers for example). Most components in such group would have indicated 0 (zero) DOFs when it is clear they can move so in real life they HAVE DOFs.

Also look in to numerous examples of DOFs analysis indicating ridiculous number of DOFs (6, so all available) for components that are obviously restricted by constrains and for example can only translate along some axis.

So in short: It is not the case so inventor sees DOFs as we would in real life.

So question still stays.

"What are DOFs exactly for Inventor?"

Cris.

[Cris-Ideas]

I would like to moved DOFs discussion to this thread

DOFs Analysis Problems / BUGS - What is a DOF in invnetor?

So we could continue on constrain solving in this thread, while waiting for answer regarding DOFs.

Please subscribe to linked thread if you wish to follow or participate in DOFs related discussion.

Cris.

[Cris-Ideas]

So lets get back to constrain solving problems.

this is

# 093

strange and hard to explain problem with applying simple face mate to an element that has no constrains so far.

video is from 2016 but this still works behaves the same in 2017.

data set for download: https://autode.sk/2pqYDxE

Cris.

[Cris-Ideas]

And another one.

# 208 - this solution does not have any sense or logic. At least for me, and for how one would logically understand redundancy of constrains.

data set for download: https://autode.sk/2QYsEBn

Cris.

[Cris-Ideas]

So finally I have a minute to update this thread.

This case proves that: "Order of constrains and components does matter- therefore solver is directional, rather than un-directional as it is stated by Autodesk"

model for download

Cris.