Continue to Site

Welcome to MCAD Central

Join our MCAD Central community forums, the largest resource for MCAD (Mechanical Computer-Aided Design) professionals, including files, forums, jobs, articles, calendar, and more.

Register Log in

top down design...

design-engine

design-engine

New member
top down design....



I want to create a big list of all the different flavors of using top down
design with definitions. I will compile it. I thought this might be
a good forum to get them all out there. I was going to ask
you guys for a pros and cons list for each and try to compile all that into one
document to make available for everyone.



I know there are different techniques in each of these below too but Ill start
it off with this. All these techniqes require the 'advanced assembly' option:



assembly copy surfaces (surface copy from an assembly)

Inheritance

publish geom

copy geom

skeletons

merge

Master Merge

mapping



Pro/PROGRAM will let you create a sort of top down too using PTC LAYOUT ... says Grant Atkinson of PTCUSER Toronto





Edited by: design-engine
 
amateur comments:

assembly copy surfaces (surface copy from an assembly): Risky, but useful if managed well.

Inheritance: have not used much

publish geom: good way to control what is linked to the source details

copy geom: useful, but often abused when the feature is built on in a careless way, not always able to update if the original is no longer available

skeletons: have not used much

merge: useful if done at a stage when the regeneration can support changes

Master Merge: works great on housings or known features.Can crash horribly in later features if there is not enough flexibility in the original master part or if features are built without expecting room for changes in the overall shape

other tools for TDD:
layouts: useful but still a bit cumbersome in pro-e
shrinkwrap: good for showing areas that are off limits or relative positions

cheers,

M


Edited by: magneplanar
 
publish geom: is good for large groups of designers working together on a project. Also when there are multiple Coordinate systems available to choose from a Publish Geometry is good so others don't have to guess to choose. The designer just does a publish geom and that geometry is available w/o question.
merge or master merge: old way how designers created a group together back in 1993 to 1997. When Pro/E version 20 came out there was a new method using a skeleton feature. Many still do it with a master merge feature that combines various other parts to form design options. With this tool users could copy all the features in or completely reference features. PRO: easy to use. CON: Layers become a problem to manage because you get separate merge layer. circular references are easy to make by accident.


Edited by: design-engine
 
Great topic! This is something I am currently dealing with on a project.
Namely, external merge vs. external copy geom.

Merge: Great for pulling *all* of the master references into the part
(with or without the datum planes option). This is great when you are
developing the master and parts concurrently. As you add information to
the master, it automatically shows up in the children parts. Also, as you
delete or redefine features in the master, they update or delete in the
children parts. This is only a problem if the part was referencing the
feature. Otherwise, it's great since you aren't forced to redefine the
merge feature over and over again.

The big downside with merge is that the heavier your master gets, all of
that weight propagates into the children parts.The database gets huge.
So if you're using merge, you probably want to keep your master as light
as possible. Some people like heavier masters, so...

Also, Bart mentioned the layers issue. We have found that all of our
master's layers end up in the children parts. The problem is that
refinements to the master layers do not also propagate down to the
children parts. It'll just keep adding more and more layers to your
children, even if you just renamed a layer or consolidated multiple layers
into one. We had to open up the assembly and manually delete all of the
irrelevant layers and then save all the children parts back to the database.

Copy Geom: This is a great way to pull references from the master
into your part without bringing the weight and clutter of the entire
master. The problem is that it does not gracefully lose references when
they are deleted from the master. You have to manually go in and tell
copy geom to delete the reference. You also have to manually add new
references as they are added to the master. This can be a royal pain
when you are developing the parts concurrently with the master because
you are constantly having to redefine the copy geom.

So between these two, I'd say the best choice depends on:
1. How complicated or heavy is your master likely to get?
2. How far along is the master?

Also, there's no reason not to use a mixture of these two. Use merge on
the big parts that need a lot of references. Use copy geom on the small
localized parts that only need a few, key references.

Caveat: I'm still working in Wildfire 2.0, so maybe WF3 has made
improvements in these features.

I haven't worked in groups large enough to need the overhead of publish
geom. I also haven't used Inheritance or some of these other features.
Anyone want to post their thoughts on those? I'd love to learn more about
them!

Also, what's the difference between "merge", "master merge", and
skeletons? Actually, I know skeletons reside as the first feature in your
assembly, do not affect mass properties, and do not show up on BOMS
and such. But I guess I'm looking for some best practices with regard to
these different options.

Thanks!
A
 
merge vs copy geom:

*merge is more reliable when the iteration between start part and endo one is big - more on it if needed, copy geom can fail while regenrating models in sesion
*merge takes to much info than it is needed - all stuff:layers, unnecessary surfs, cosmetics and so on, in copy geom you can choose what you exactly want, it is more easy if only solid surf are copied

top down - well all goes right if project is well thought at the start. I see it in this way: top down is nice tool sharing changes among all files, but small changes like position, or simple dims. If general change is needed, top down often fails because the reference relationship between models can not handle it. No big changes are desired, allowed. I have seen this many times when start conditons where dramaticaly changed in half of the project life and everything had to be deone from start. So think wise, make it easier.

next thing considering top down -external refernce. It is nice and easy to build models on others. So, basicaly it works but in narrowed scope. When one change only dims - no problem, but when reference curve is changed by deleting some if its entities, oither sketches fail, because Pro\e can not handle. And yes, You can use then Edit>Replace in Sketcher to have everything still right but it is time consuming and frustrating. The Replace tool should be much more flexible and allow to replace not only one vs one entity.

And all of this - top down - has sense only with good pdm tool. I have bad taste of working on mapped drives when Pro\e with each change of base asm was loosing all extwrnal reference.

do it wise at start then it comes easy
 
Bart,


Coz I don't do much by way of plastic (cheap nasty material, should be banned from engineering applications
smiley36.gif
) design and most of my design is heavy machinary with lots and lots of asseblies and sub-assemblies, I tend to stick to skeletons.


Grantedthe skeleton geometry (planes and axes as well) can get pretty complex for say something like a compact wheel loader or mining machines, but with help of groups, layers and hiding I have always found I can keep it manageable.


One of the main advantages for me (and I have other threads regarding this) is that I can keep external references out of my models, which all the PDM's I have worked with love. It also makes it very simple to 'swap' ssub-assemblies in and out of assemblies (all based on default constraint). Again, it makes life quite simple when more than one engineer is working on the same top level assembly.


Kev
 
I saw 'mapping' in one of the menus and recall one of my past students talking about it a lot years ago. I'll hunt around today and locate that option.
 
Bart

Did you get your list compiled? and if so can you post it/or send me a copy?

Cheers

Ben
 
great initiative by Bart


please add the following headings also:


simplified rep


interchangeable assembly


make flexible, restructure(under edit menu)
 
Hello All,


I work on very large assemblies (2000+ components)and couldn't imagine doing this without the use of skeletons, publish geometry, simplified repsetc. We work from a master datum and create a top level skeleton, from which data is published down into the lower lever skeletons for the functional groups such as chassis and furniture teams to work on. We always use the publish geometry feature to control information as mentioned in previous posts copy geometry can easily be 'abused' and end in an almightly mess.


My experience of the inheretance feature is that it is useful for de- featuring parts for manufacture. Take a cast part for example in which the part is fully detailed in the 'parent' part file then the base features are inherated into the child part which then represents the cast part. I do not use the inheratence feature for typical top down exercise however I'm sure it has its applications.


I will post some more when I get time!!
 
I used the inheritance feature recently to make a vacuum formed clam shell. Turned out to be more work than necessary but I am still experimenting with various uses.
 
One use of inheritance features is the following:

Create a part (part a). Then create another part (part b) where you model all of the cutouts that need to go into "part a" as solid extrusions.

Now use the inheritance feature to merge "part b" into "part a" and select the remove material option.

Now you have a part that can be used on any number of parts and can get the cutouts you need without having to model the cutouts in each part.

We make configurable sheet metal enclosures and I have used this method to provide a library of standard cutouts. Now, when I have a new profile to make, I can simply model the new part, then inherit all of the cutouts I need into it.

Imagine 50 different profiles, that are interchanged in an assembly and no matter what profile is in session, the cutouts are there, all controlled from the one cutout part.

I don't know if this makes any sense, but it works.

Jim
 
Not too long ago I worked for a company that utilized Pro/PROGRAM, interchange, & skeleton modeling together. Everything is assembled to the skeleton and the program would flex the whole assembly from the top level down to the parts & drawings. For instance, one assembly for a tower would control all configurations from single/double door, lock or no lock, width, depth height, wood grain direction, etc. Pros: one model controls everything, swaps out parts with no dependant failures... Cons: one model controls everything, you have to pay close attention to ensure everything interfaces correctly (ie. no gaps or interferences) and you have to do an interference check from time to time.
 
I came up with this idea by fooling around in ACAD solids. I know...I know... they suck. But this before I was exposed to a true 3D modeler like Pro.

Basically in AutoCAD, you create two solids, intersect them, then subtract one from the other.

So, when I was told we had to come up with a way of having standard cutouts, this was the result. I am thinking it has slowed the regen times down a bit, but the benefits are immense.

The way spidernate described his previous job, this is kind of how we do it. But with many many input params and so many options, it was the only way to do it sanely. If I had to physically go into every part and create not just the features, but the inputs and pro/program to control all of these, I would never get done.

Doing it in an inheritance part seemed to make the most sense.

Jim
 
Jim is right... a lot of time put into a product up front, but the programs provided a big return on the later part of the product's life.
 
You must log in or register to reply here.

Sponsor

Articles From 3DCAD World

Back
Top