Skeletons in Top Down Design 2

[markborges]

Hi All,

Been reading thru some of the posts on TD Design and have a few questions.

Is there an advantage to ECGing component geometry into the skeleton and then ECGing the same geometry to the subassemby for component design. With a large top level assemby, I could see how this would be an advantagte to open the top level up once, grab everything that I need and then place it in the skeleton. (Of course only once is in theory.) Or is it just as easy to ECG the needed geometry from the Top Level directly into each subassembly. (We have subassemblies installed in to the top level via a main vehicle coordinate system.)

Another question. I have inherited this model and the top level skeleton, actually all skeletons, were not created with the skelton module; ie., they are just parts with sketches in them. Is there a good reason to convert these to "real" ProE skeletons. I have noticed that the sketched parts come into my BOM as needed parts; but can these be deleted out of the BOM table? If they can just be deleted then it is not big deal.

Thanks,:-D

Mark

 

[JohnAndrews]

Mark,

I think you need to explain your first qustion more, i.e what is it that you are designing, what is typically in your skeletons, what you are using External Copy Geom with, etc.

The main advantage to using skeletons rather than conventional parts is that they are automatically excluded from the BOM, if you use standard parts as "skeletons" they can be excluded from the BOM using repeat region filters.

 

[daveykbelgium]

Hi Mark

John is to the point with his question about the purpose of your skeletons.

ECG like most Pro/E functionality has its vices and virtues.

Don’t use ECG to define part geometry if you want to dimension the part like a good boy, i.e. using “show dimension’ and not created dimensions. In this situation you would need to create a “detailing assembly” and model the part by activating it in the assembly, and then creating your part using references to the skeleton. Detail the part using the detailing assembly, and show the dimensions of the skeleton for detailing your part. There are ways to work with mid-bound tolerances too, because skeletons can’t handle these. We have used this method for many years, and it is very well proven.

The true virtue of ECG is that you can donate certain features of your sub-assemblies skeletons to higher level assemblies – in-fact right up to your highest level assembly. In such cases, the ECG only shows representation, and very importantly, not geometry that requires to dimensions. I reiterate that ECG allows you to be selective about what features of the skeleton are donated to the higher level assemblies. It is a scaleable solution too.

Your point about sucking information for ECG from a top-level assembly is not possible on really big projects as the assembly is usually unmanageable, even using simplified reps. It is sort of reverse engineering anyway.

I suggest you start by defining skeletons for sub-assemblies that contribute to higher level sub-modules. All the sub-modules come together in your top level assembly. Like this, skeleton can detail parts, and donate to higher-level assemblies through ECG. Say if ten sub-assemblies donate information to a sub-module through publish geometry and ECGs, you can then simplify the sub-module using a new publish geometry/ECG again – so, scalable.

ECG can misbehave. If the skeleton that the ECG was generated from is not in session, when you redefine the position of the ECG it won’t move. You will need to create a reference assembly that contains all the skeletons – called a retrieval assembly – and load this at the beginning of your working day.

TDD is a very detailed subject, and a post here is still only a very superficial overview. I’m defining documents on the subject covering the complete development cycle, and they are large. Remember, most important is what deliverable you require from each stage of your design and development process, and this will define whether ECG is a candidate.

Every company has a unique product and unique production technologies and possibly unique manufacturing process, so there is no global solution to TDD. Develop you ideas on mini-projects and prove everything – otherwise you can get egg on your face!

Finally, like me, a lot of the guys here create skeletons that define an assembly, so more than one part. I should add the objective is only to include fundamental sub-assembly geometry. Whether or not we generate parts including datum curves, or “true” Pro/E skeletons isn’t the point here. We create publish geometry features and then use these in ECG in new parts, thereby creating individual skeleton parts - pseudo skeletons if you like. These are only used in the definition of mechanisms, and nothing else. Parts are detailed as I stated a few paragraphs above. These new pseudo skeletons are very light, and this justifies the methodology.

What I call “true” Pro/E skeletons do have the advantage of being filtered form BOMs in Pro/E and Intralink ).

To convert a “normal” part with curves to a true Pro/E skeleton, you can be sneaky. In any assembly, use the insert/component/create/skeleton model. Define the name, but use the “copy from” box to use your present part as the template. That might save you some time?


By the way, I thought you automotive guys had very strict ways of working. I read a few methodologies which had me chuckle. Is it really true that most big parts, i.e. a bumper/fender, even though they’re a million miles away from the “master co-ordinate system” of the car, still need a co-ordinate system to allow them to assemble onto that point, making the bounding box of the model or assembly simply enormous?

Any more questions, please feel free to drop me a line. We haven’t covered layouts yet either. Oh, and one big skeleton would mean all designers would be fighting to modify one file. Keeping the one skeleton to one sub-assembly rule makes file handling and data-base issues far easier to handle.

I’m rushing a bit as it’s late, so hopefully I explained everything clearly enough?

Have a nice weekend mate.

Dave

 

[markborges]

Thanks Guys,

We are designing the frame and suspension system for a military truck.

Individual coordinate systems are located in the top level skeleton and each sub assembly are installed to these coordinate systems. Some subassemblies have vehicle coordinates defined. These subassembies are, of course assembled to the skeletons, main coordinate system. I have found that a vehicle coordinate system in the steering linkage to be very useful for my ECGs. Ie., I can easly bring tie rods into the cab mounting subassembly, because they share the same vehicle coordinates.

So that was the concept that I was first writing about. In practice I have found it to be quite easy, effective and fast.

What is the advantage to published geom? Can surfaces and whole parts be published?

As I understand it layouts are created with draft entities; and contain parameters that can be used in skeletons etc. What are notebooks.

Thanks for your replies. I too have to go.

Mark

 

[jeff4136]

Dave,
Very interesting run down on your top down techniques.

>... away from the “master co-ordinate system” ...
>... making the bounding box of the model or assembly
> simply enormous?"

FWIW if the first feature is a csys; parts can be modeled in body coords far from that csys and assembled at default location without expanding the part or assy bounding box.

 

[Guest0527211403]

Dave has offered some excellent advice.

In response to your question, a Publish Geometry simplifies the External Copy Geom process by allowing you to control what is passed on through an ECG from within the donor part itself. A Publish Geometry feature is like a "Package" that you make in the skeleton model for use downstream. You can include whatever you want (surfaces, curves, datums, etc) to be passed down. When you are defining you ECG, you can opt to just choose a publish geometry feature, and then set CSYS references.

Speacking of CSYS, while it is advantageous to have a vehicle coordinate system for the reason Jeff said, Dave is also right in the fact that it is irritating to work in models with a csys off in space since spinning the part around the part's center becomes practically useless.

One way around this is to define coordinate systems in your skeleton model, and assemble your components to these coordinate systems. In past discussions this has been reffered to as "hanging" the part off of the skeleton. This allows you to a) define your skeleton models with the appropriate degrees of freedom for motion analysis, and/or b) define your parts so that the "front" view is always the true front, allowing you to implement drawing template standards. Ususally it is best to assemble to the same coordinate system in the skeleton as the one you used to define the publish geometry.

Good luck!

 

[jeff4136]

Sorry to keep wandering off topic. The concise descriptions have been an illuminating glimpse of things I've not been exposed to (don't have AAX).

> ... a csys off in space since
> spinning the part around the
> part's center becomes practically
> useless.


I don't believe a csys will influence Pro/E's interpretation of model size or affect model (spin) center. There's something else; datum point, plane, geometry, that's contributing.

My template files have, starting from Empty; CSYS0, CSYS1. All subsequent features (including planes) are dependant on CSYS1. By manipulating it I can translate and rotate geometry anywhere in space relative CSYS0 while maintaining a "normal" bounding box and spin center. Sometimes handy.

 

[Heckler]

A little late but I ran across this article today.

Top-down Goes Bottom-up
- by Jim Buchanan


You’re a manufacturing engineer, and you get the phone call you’ve been dreading. It’s from the shop floor, where the team is setting up the work cells for your new product.

“We need to weld the arms together, but there’s not enough room for the welding robot to squeeze in there.”
Oops. That’s your greatest fear — a new, expensive ECO. Now the product will have to be modified so the robot can fit. This means it will have to go back to the designer, then back to you, then back into manufacturing.
Changing the way you work. You’re probably already familiar with Pro/ENGINEER Advanced Assembly Extension (AAX). But did you know it helps turn top-down designs in Pro/ENGINEER into bottom-up assembly instructions for manufacturing? This means a lot for manufacturing engineers because part of your job is to write the detailed process sheets that are used to guide the shop-floor assembly of your company’s products.
But to do this effectively, you need to be able to convert your designer’s assembly model into a “bottom-up” template, one that shows how the smallest parts fit together to form the overall assembly structure.
For example, with a lawnmower, you might break the assembly into groups for, say, the engine, the transmission, frame, controls, and so on. Then you’d look at each of these small assemblies, and break them down further. Then you’d produce the necessary bill-of-materials documentation to guide the construction of the assembly, as well as any necessary fixtures, such as jigs or cranes, that might be needed for the construction.
Handling pressure. But regardless of the product you’re making, whether it’s a lawnmower or jet engine, you’re typically under pressure to do your job quickly, and to avoid making mistakes that might result in downstream ECOs. Adding to that pressure, you usually have to wait until the assembly model is finished by the designer before you can pull it apart to create your shop floor guides. And only when you’re finished can the actual work cells be organized.

But process planning in AAX changes that. The reason: With AAX, you don’t have to wait for the assembly model to be completed before you can get started — you can have your own process model, and do your process planning in parallel with the assembly model design.
Because Pro/ENGINEER is always associative, any time there’s a change to the assembly design model, your manufacturing assembly will be updated too. So while you might have to do some tweaking as the assembly design is being completed, you’ll know that you’ll have 80 percent-or-so of your own process work done when the assembly design is released to manufacturing.

Process planning made simple. Process planning is one of several major tools within AAX — including, among others, top-down design and configuration tools — that all share the same purpose: to bring simplicity to the inherently complex endeavor of assembly modeling. Process planning in AAX does its part by giving the manufacturing engineer a flexible tool for assembly planning.

For instance, process planning in AAX lets engineers generate associative step-by-step 3D and 2D process models with views that show the construction sequences in detail, as well as associative bills-of-material for each stage of the construction process. Also, the process models are color-coded, making it clear which parts are being assembled — and which have already been assembled — at every step along the way.

Associativity with Pro/ENGINEER means that changes in the design model are reflected in the process models. But as a safeguard, changes in process models are not reflected back to the design model.

Faster prototypes mean fewer ECOs. In the case of one PTC customer — a fuel cell maker — the average number of engineering change orders dropped from 40 to just two when AAX was used.

As a design engineer, you’ll be thankful for the process planning capabilities in AAX the minute a manufacturing engineer comes to you — while you’re still designing your assembly model — and points out that some part of your design may be very difficult or impossible to assemble. A fix at this stage will be easier than later on, and you’ll avoid the cost of a more expensive downstream repair.

Overall this should give manufacturing engineers something to smile about. Process planning in AAX puts engineering teams one step closer to making the concept of design-for-manufacturing a reality.

http://www.tristar.com/news/articles/0306-2.asp

Best Regards,

Heckler
Sr. Mechanical Engineer
SW2005 SP 5.0 & Pro/E 2001
Dell Precision 370
P4 3.6 GHz, 1GB RAM
XP Pro SP2.0
NVIDIA Quadro FX 1400
o
_`\(,_
(_)/ (_)

Never argue with an idiot. They'll bring you down to their level and beat you with experience every time.

 

[markborges]

Thanks Heckler for reviving this thread--I am still piecing it together. As far as assigning coordinates, some subassemblies have vehicle coordinates and some have local coordinates that are "hung" on to the skeleton. If I need to ECG something out of a subassembly, I have been creating a vehicle coordinates and using it as my common coordinate reference. Some subasseblies have been brought in from Solid Works w/o a coordinate system; so, I have added a vehicle coordinates system to those models. So we have a hodge-podge and the hod-pod is moving toward using a common vehicle coordinate system. I think we are on the right track; all of our MD subassemblies are stand alone; ie., the moveable crane is mounted to a base and the base can be assigned a vehicle coordinate system. I like the system that Jeff mentioned. I have seen it before; but now I can see its merits.

I have found publishing geometry to be very effective means to share geometry. The geometry only has to be published once and it can be used multiple times for multiple instances. I am currently trying Dave's "scaling" suggestion to create a light weight top level assembly from ECG features. Instead of droping all the way down to the piece parts, I have created shrinkwraps at the first or second subassembly level. I had to use shrinkwraps, because published geometry would not let me work with the entire subassembly. I could only work with one part. So I had to create a shrinkwrap, which enabled me to select and publish individual quilts. I found this to work; however, I would like to speed the process up a bit. I have tried the option to select "Copy Geom" and I am able to select the entire shrinkwrap and publish it; however, it does not ECG into my top level assembly. The ECG appears to work, I select my published geometry; but it just dosn't show up on the screen. I have lots of funky layers; but even with them all on nothing.

Is the right way to go about building a light weight assembly. I know about simpified reps and I use them where appropriate; but here I am working on the top level assembly and want to see as much detail as we can. (Some of this is for pretty pictures.) For relatively small subassemblies; ie., 15 or less members, I have found shrinkwrapping, publishing, and ECGing up to be very effective; but I am struggling with some of the large subassemblies. Is it possible to ECG up published geometry from the lower levels? I mean levels lower then the next? Or is it possible to "bring up" all published geometry that is burried in the subassembly.

I have tried to bring in a complete shrinkwrap of the subassembly; but, even at level 10 I am missing fairly important surfaces. I like the idea of flowing geometry up, because, I know exactly what I am getting--I selected it. In addition, I have noticed that the my subassembly shrinkwraps are frought with geom check errors; ie., too small of surfaces. I suppose I could just opt out the small surfaces. Just havent tried it yet.

Have a good day,

Mark

 

[daveykbelgium]

Hello again!
Wow, this post is getting quite lengthy isn't it? Sorry I haven't replied sooner, but I've been too busy to look at the forums lately.

I can't spend a lot of time now, bit have a few comments to make.

I wanted to say that I only use ECG to make skeletons. A publish geometry feature from a sub-assembly skeleton only donates to an ECG feature one level higher, i.e. to an ECG in the sub-module skeleton. A publish geometry feature from a sub-module skeleton, only donates to an ECG feature one level higher, i.e. to an ECG in the top-level assembly.

The only other time I presently use ECG is to extract selective geometry out of a sub-assembly skeleton to create a “pseudo” skeleton that is used inside a mechanism assembly skeleton. However, don’t get confused and think that the “pseudo” skeleton is used to define a part because this is DEFINITELY NOT the case.

From what I have read, there is now “true” skeleton assembly functionality in Wildfire3, so I’m keen to see what this really is.

Skeletons serve a number of purposes. They show simplified overviews, whether this be at sub-assembly, sub-module or top-level assembly level.

The scalable solution, i.e. using publish geometry features of the most fundamental geometry from a sub-assembly skeleton in the next level, keep the sub-module’s skeleton clean.

All skeletons can be shared by designers with their mechanical engineers, and/or engineers of other disciplines, i.e. process and even software engineers who need distances for setting parameters inside their code.

Skeletons also create the frameworks on which to hang parts, low-level sub-assemblies, sub-assemblies and sub-modules together. In order to so this, you create co-ordinate systems on the skeletons. With layouts you can exploit auto-assemble functionality too.


I NEVER EVER, EVER use copy geometry to contribute “alien” features towards the definition of new or higher-level parts. Relationships become so complex that you can kill yourself. Nested ECG in models from a design consultancy has given our engineers the worst nightmare I have ever seen with Pro/ENGINEER. As a hardened Pro/ENGINEER user you often think that you are man enough to jump into any re-define scenario, but this is one of the few times I can say we have been reduced to the feeling of a Pro/E virgin and want to start again. This is a continued engineering activity on released parts, and the implications on a full set of covers are truly enormous!

If you want to do things properly, there will always be a way to use your sub-assembly skeleton to define the geometry you are trying to create. Remember, every product needs a deliverable, and dimensions in deliverables must be shown and not created! ECG means this isn’t possible, so don’t waste your time thinking about it.

Don’t get me wrong though, as I said before, ECG has it vices and virtues, and used at the appropriate time, ECG is the best tool for the job.

So, coming back to you Mark, you say that you’re using my scaling methodology to create a lightweight top-level assembly. This isn’t my intention.

Our designers have the freedom to work in two ways. They can use a “working assembly”, or a “light-weight assembly”. Every designer has their own unique “working assembly”, and initially, this only contains the top-level assembly of the product and some datum curves and planes used for the creation of zones.

If a designer is asked to start a new design activity, he will create a new assembly in his “working assembly”, thus his “fast-moving and conceptual“ work will not be viewable by other designers. Only when his design is released and included into the top-level assembly by the system engineer, will it be available to others.

“Working assemblies” were developed so that designers could each create their own unique simplified reps of the top-level assembly. This avoids the problems that many designers are fighting to save the top-level assembly back into the database.

However, some designers don’t like to make simplified reps from very large model trees, thus a light-weight method is also available.

The “light-weight” assembly is a more free-form way of working. Here designers add a skeleton into an empty assembly. They then hang the appropriate assemblies onto the skeleton. This can be performed at sub-assembly, sub-module or top-level assembly level. Layouts give the bonus of auto-assemble functionality, so the system is very fast.

Also, once you have assembled your skeleton into the “light-weight assembly”, you do have the ability to hang a simplified rep of a new assembly, i.e. not the complete assembly. If you have an empty simplified rep within the assembly you are hanging onto the skeleton, this can be assembled far faster then using the master rep. After this the substitute command is available to switch simplified reps.

Perhaps you might try one of these techniques?


My previous point regarding coordinate systems seems to have created more of a stir than anything else – at least I have one supporter in justkeepgiviner! Perhaps we are just old school though? I haven’t tried the co-ordinate system thing recently, and perhaps these are just bad memories from older revisions of Pro/E where there was an effect.

What I would say that I’m an advocate of always using the 0,0,0 point of a part or an assembly as the true manufacturing datum or the true assembly position to the next level of assembly, thus I’d never use co-ordinate systems in the way that you guys do.

Last year I worked in an organisation using Pro/E but we had technology partners using other CAD systems, and using inherited files isn’t nice and always compromised. The biggest worry is always assembling their work correctly relative to your own, so you need to have agreements and methodologies to suit. The other common problem was screw threads, and you need to know how your partners model them – we had a few embarrassing moments!

If you have any other questions, then please let me know. I do actually have formalised notes on my methodologies and I’m currently reviewing them, so perhaps another set of eyes is a good thing? The more people who try to shoot holes in it, the better the final solution?

One final question - are you using a database? ECG could do you in after release for continued engineering activities and/or new products, if you don’t respect it. My methodologies are safe and won’t get you into heaps of problems later, but this is also dependent on how you develop your products later.

My final and perhaps “profound” comment – use ECG with caution and respect, or it might come and bite you on the ass!

Cheers

Dave

 

[markborges]

Thanks Dave for the clarification on your light weight assemblies. I have aborted my mission to bring all geometry up to the top level w/ ECGs. It does open quickly; however, if the geometry changes, which it is, I have found that I have to "open up" each ECG to get it to update with the new geometry which takes time thus defeating the purpose. And not to mention, sometimes, entirely new components are added to the assemblies. we are in a state of flux with many components being changed. "Pretty pictures" just take time.

I would like to read your "formalized notes". How can I get a copy of them?

Mark

 

[daveykbelgium]

are you on msn or skype?

how do people usually send stuff here?

 

[Guest0527211403]

A while back Heckler recommended Mooload.com . It's a good service to post a file. I do believe they expire if no one downloads them for a month or so, however.

 

[markborges]

Thanks Dave,

If it is less then 50 megs, you can post to my private ftp address..

ftp://FTP.AROS.NET/pub/users/mborges/

Mark