Fillet Failure 2019

[wsimonton]

I have a Part which has been stable for 5 or six years or more which with 2019 has "lost" or "failed" a fillet which had been stable. The location is indicated on the jpg included here along with the attached Package. 2019 has also exhibited some trouble loading this file when part of a Package. The Boolean Add Part included appears to cause some issues. If F405R Engine Frame will not load, load and save the F405R Boolean and then load the F405R Engine Frame. That usually works.

Any way to get the fillets Fillet <168> and Fillet <169> or any combination of fillets would be appreciated. The Edge which needs Filleted is indicated as the Blue Edge Line in the jpg.

What is really puzzling is that F406L which is a Mirror Image of F405R works. I guess I can load and save F406L save as F405R and eliminate the Mirrow and one other change - an area of wear on F406L not on F405R but I would like to figure out WHY.

 

[HaroldL]

Yeah, your model does have some issues. It crashed my AD three times trying to open it. I think you should send it off to Support to see what they can do with it.

As for the missing fillet - do you have Tangency turned on? Sometimes it helps to turn tangency off and select all the affected edges individually. But that was what I found sometimes in AD2018 it may not be the same in AD2019.

For creating the package file, you might try rolling the parts back to the start (move the Dog Bone to the top of Design Explorer) then package them up. It will create a smaller file to package and when someone else opens it there may be a better chance of it actually opening. Then the part can be rebuilt step by step and hopefully find the errant feature.

 

[Max]

Your suppressed fillet that is failing is .4". You have another fillet already at .5". If your fillet were to succeed, you would have a .4" meeting a .5" tangentially, which doesn't make sense to the software.

When I changed it to .5" it works, making sure to select all the edges.

 

[wsimonton]

The 0.40 was the result of me playing around with the radius of the fillet to determine if a smaller fillet would work. Failed to return it to 0.5 before I posted.. Sometimes Alibre will not fillet but getting it to fillet at a smaller radius one can work by increments back up to the desired radius. That has been true in the past. Do not know if that still works.

 

[simonb65]

The 0.40 was the result of me playing around with the radius of the fillet to determine if a smaller fillet would work. Failed to return it to 0.5 before I posted.. Sometimes Alibre will not fillet but getting it to fillet at a smaller radius one can work by increments back up to the desired radius. That has been true in the past. Do not know if that still works.

It's a pity the fillet radius selection doesn't allow you to select the adjacent radius in order to get round the trial and error or measure first approach!

i.e. select the existing edge (RED) as the 'radius' and apply to the edge (BLUE)...



@Max ... any chance of this enhancement??

 

[wsimonton]

It does to some extent using Variable Radius Option. See attached

 

[Max]

@Max ... any chance of this enhancement??

Perhaps. Can you tell me how often you run into this specific scenario? Let's say that takes a week of development - lots of other things we can do in a week. Personally I never run into this - maybe once in 10 years. If this is a common thing for everyone else, maybe it makes sense, but my gut tells me perhaps it wont get used often.

 

[HaroldL]

Max,
The few times that I've had to match a fillet to an existing rounded edge I just select the existing radiused face and read the value in the status bar at the bottom of the workspace. It's easy enough to key that value in for the fillet. If I ever had a problem with a fillet not propagating I've just select more edges in sequence until I see the preview along the entire edge I want filleted.


wsimonton, I finally got the part to open and was able to just add a couple more edges to Fillet <168> and update it. I then deleted Fillet <169>.
Also, I don't mean to be critical but I strongly suggest that you fully constrain all your sketches. It will be real sad day if you lose a part if an unconstrained sketch got fouled up after all the work you put into this project.

 

[wsimonton]

Although I have used 2D CADD for 40 years, I only started using Alibre and 3D CADD in 2009. I was learning on the fly.

I am in the process of going back and trying to back fill all the info on each part as I can. I was working just ahead of the assembly process and trying to get all the info CORRECT before each part became inaccessible as the locomotive was re-assembled. With about 995 Unique Parts (some duplicate Part Names in separate subdirectories) and 10,061 parts so far that is going to take me awhile. I have been working on the Climax Pattern Castings to start with, but that can be a bit tedious so I tend to do a small batch at a time.

I have reported the issues with F405R and F406L which include the problem with the Boolean Add causing Alibre to crash (except on my primary Desktop computer on which they were created) and a 3D Parsing Error which can occur. I believe they are working on those issues as I write this. Some referrals to System Analysis (probably wrong term) I have been advised have been made but I do not know the issues Alibre has internally referred for troubleshooting. They have reported to me some odd (their word) behavior by Alibre but have not elaborated.

Since I want to preserve the info contained in the CADD model and I am planning to distribute it to a number of preservation organizations and if I cannot provide a Package of 1551 of all the data that will open correctly on other Users systems all the info will ultimately be lost again.

Up until Alibre 2019 I have been able to share the 1551.AD_ASM with another User so hopefully Staff will resolve the issue(s).

 

[Max]

The few times that I've had to match a fillet to an existing rounded edge I just select the existing radiused face and read the value in the status bar at the bottom of the workspace. It's easy enough to key that value in for the fillet. If I ever had a problem with a fillet not propagating I've just select more edges in sequence until I see the preview along the entire edge I want filleted.

I see, your request isn't "match a fillet in this geometry case" but "populate the fillet radius box by clicking on geometry and just matching the radius". I could see that being useful.

 

[simonb65]

Perhaps. Can you tell me how often you run into this specific scenario? Let's say that takes a week of development - lots of other things we can do in a week. Personally I never run into this - maybe once in 10 years. If this is a common thing for everyone else, maybe it makes sense, but my gut tells me perhaps it wont get used often.

I've done this about half a dozen times since I got 2019. I guess it depends on your workflow, but I fillet as I build the model and not at the end (unless I have to mirror .. but thats anothe fillet issue!). A week!!!! The data is already available when select an edge (you display it in the status bar), you just need to put that in the edit box for radius! Probably 30 mins work, if that. Estimates of a week for something so trivial is a little worrying @Max, if not alarmist! Your talking to an experienced developer here.

 

[Max]

I've done this about half a dozen times since I got 2019. I guess it depends on your workflow, but I fillet as I build the model and not at the end (unless I have to mirror .. but thats anothe fillet issue!). A week!!!! The data is already available when select an edge (you display it in the status bar), you just need to put that in the edit box for radius! Probably 30 mins work, if that. Estimates of a week for something so trivial is a little worrying @Max, if not alarmist! Your talking to an experienced developer here.

Perhaps - but consider that we implement this only for fillets. Why not chamfers? Why not everywhere? All the sudden, the paradigm of using a geometric selection as a simple numeric input becomes product wide. Why can't I select an edge as an input for an extrude cut depth? Contraint offset? Why did we only do it for fillets? Well, we wouldn't - then we have to make it applicable to every dialog - a class change but one that needs to be tested - everywhere. Shouldn't it be associative too? That's different than what we do today, for this type of input.

These inputs have effects in the Equation Editor as parameters that get generated when making a new feature. How are those affected? They have to be read-only as they are from geometry and not user-inputs. They have to have a new naming scheme in the equation editor - it's a new type of associativity.

What about when there are multiple conflicting geometric properties? You want to pull the radius from something in the rotary pattern dialog, but that same geometry could be used to define the desired axis of rotation. How do we differentiate between the possible intentions? Does it affect the default selection order of various dialogs? It's all possible, but as an experienced developer you must know that getting 90% of the way there takes 10% of the time - the last 10% - what makes something usable, easily understood, etc. is usually the time sink. This is actually probably a 2-3 week project to get developed and tested.

You aren't wrong that it might be trivial to do, if we were focused only on copying some selection into the fillet dialog. But for consistency we can't have such a narrow implementation. And it has to be associative, or at least should be. That's complex.

"there is no low hanging fruit" - our dev team

 

[JST]

Can't suggest how many times it comes up, but it is super good for matching things that come in as STEP files, where you have no idea what the designer did unless you measure. And that may not be good enough.

The issue here is that if you simply manually copy the value out of a measurement window, you get 6 digit resolution. If you copy it as Alibre measures it internally, you have access to the entire Alibre internal resolution.

That can make a difference, apparently, since I often run into things such as a distance between identical holes, that measure the same to 6 digits, but which stubbornly refuse to align. Evidently there is a difference down in the "noise" by a few more decimal places, and Alibre does not accept it a equal.

Given the somewhat fragile operation of fillets to begin with *, that tiny error might easily lead to an inability to get them to match which is just as much of a "stopper" as the obvious .4" vs .5" issue you mentioned.

* I rarely get a part "fully filleted" without an error that forces me to change the order of operations in some arbitrary manner to get all the fillets to complete.

 

[simonb65]

You aren't wrong that it might be trivial to do, if we were focused only on copying some selection into the fillet dialog. But for consistency we can't have such a narrow implementation. And it has to be associative, or at least should be. That's complex.

Fair enough! My initial focus was quite narrow and specific to that single use case, but the points you raise are extremely valid in the wide context, some of your thought flow in your post would lead to a very powerful set of tools (especially from an ease of use perspective). Maybe some of those ideas could be considered for a future overall product enhancement when the the other medium and higher hanging fruit has been picked. It's all good discussion.

 

[Max]

Yeah, the more I think about this, the more I agree it would be a very powerful tool like you and JST suggest.

 

[simonb65]

Can't suggest how many times it comes up, but it is super good for matching things that come in as STEP files, where you have no idea what the designer did unless you measure. And that may not be good enough.

@Max, this is the usual case when I come across the need to match the existing target.