Wavelinking - The baby's going with the bathwater! 1

[JonSelby]

I'm trying to convince our CAD support guys of the benefits of Wavelinking but unfortunately they have had their fingers burnt by someones over exuberant use.

As an example; I recently inherited four parts all linked to a 'dumb' part, where each each 'real' part had 6 linked bodies, 9 linked datums and 23 linked sketches! It was quite frustrating to jump back and forth and took an age to update (even with Delayed Interpart Update).
Having remodelled this I now have one master from which with three other parts are wavelinked at a particular timestamp. These have just 1 linked body and 3 linked sketches and then maybe a dozen features unique to that part.

I'd be interested in hearing peoples view on the best way of managing the wavelinking workflow.
The main concern my CAD support guys have is that someone picking up a new model may not realise that changes they make are affecting other components.
They are trialing the Wavelinking Manager but are finding that it does not pick up all the links and that are present.

I may be wrong but I'm anticipating we may get told to desist from using this feature.
And I think interpart expressions and attributes would probably go with it (although I don't think people here have discovered those yet)!!

As a principal would one or other of my work-flows (see attached .PDF) be preferable?

Any advice, comments, concerns or flashes of inspiration would be greatfully received.

Regards,
Jon

Jon S

Medical Design Engineer - Glos. UK
NX 6.0.4.3 / TCE V10.0.3.8.6

 

[rg006]

Oh my friend....
Thats a much easier thing to do using UDF's (user defined features).
The way UDF's work:
You create a common feature (like a cylinder for sinking). When you start a new part, you just import these UDF features from a library of UDF's you have created. After you have imported everything that you call common, you then drawing in and create the specialty features. In practice, once you have spent enough time creating "alots" of UDF's, you pretty much can quickly put together a finished part in less than half the time.
UDF's consist of:
sketches/extrudes/etc
expressions/attributes/etc
When you create a UDF, the expressions will define the feature. So lets say you import a box UDF. You can specify that it should ask for three expressions (length, width, height). As for placement, the easiest way is to create a template part with customized datum and plane names. So when you import the first "body" UDF, it is set to automatically position face 1 to plane 1 at 10mm, and edge 1 to datum 1 at 3mm, and so on. UDF's are a very efficient way to quickly build models without starting from scratch. There are only two difficulties. First, is the time it takes to build your UDF library, since it's built from scratch by you. The second, is that it takes a UG license to use. "A company Im aware of" spent about 6-9 months with the UDF's before getting "everything" they commonly or remotely used. However, it now takes 1/4 the original time to create new parts. Hope this helps. If not, ask John.

 

[JohnRBaker]

I'm going to sit this one by the side of the road for at least a while yet and watch what comes by as I think the views of real users might be more productive at the moment, such as the comment about UDF's, which indeed has merit, but like people and companies, there are many different ones out there with different products and workflows and so NX offers many pathes to many solution, some better suited to some situations than others, so I'll just enjoy the parade for a while, OK?

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.

 

[ewh]

Are you mellowing out for retirement, John? ;-)

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[JohnRBaker]

Not for a while yet ;-)

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.

 

[hudson888]

Writing about this makes me feel old but the discussion around wave linking reminds me of ones I was having with COBOL programmers 30 years ago. The question becomes whether it is "maintainable" or not. I guess that's what best practice aims for.

The rub is that in the real world you might label your baggage or address an envelope in a certain way because you know that it won't get to where it is going unless you follow the recognised convention. This kind of common knowledge enters into the culture in the real world in entirely the same way as does using a CAD system in a virtual realm. If you want to wave link without labelling your features it might work for one, but it makes no sense for six instances within a single file.

Wave linking is extremely flexible and most regard that as a good thing. However if you want to save time in recognising and analysing the work of other individuals then working within set conventions is the best way. In developing those conventions I would concentrate on saying what is preferred rather that what is allowed. Being intolerant of difference is almost always unhelpful and usually presumes knowledge of different sets of problems that you can't guarantee that you possess until you're confronted with them. Better to say that you usually prefer such and such a method perhaps with the part as a component of the file it gets linked to, or using a top down strategy for tooling etc. What I would add to that is that users should take the trouble to label features especially where their reasons for a particular construction are less straightforward.

Create a few good examples and save then on the system for reference purposes. Through discussion and feedback among other users a happy medium will usually quickly emerge.

Beyond that I'm not for forcing you into anything because I don't know what is appropriate to your workflow. On the technical side wave linking within teamcentre is a little more involved that in native so have greater care not to use it with abandon there. And finally learn how to analyse wave links using the tools with NX, take a course if necessary, and make sure others in the office share that knowledge evenly. Assume only that one man's badly structured quick fix always becomes someone else's unwanted legacy data and I think you'll be well on the way to building an office culture around sensible use of wave.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[phillpd]

Just my 2 cents ...

We're actually migrating our UDFs (have used them for many years) towards WAVE links (making more extensive use of them now).

The reason being is that UDFs are embedded in the file that they have been added to. So, unless the UDF has infinite flexibility, if there's major change required (to it), you may have to rip it out of the file and add a new one. Can be painful and time consuming.

With WAVE links, you're always using a reference to the actual geometry, so you have infinite flexibility in what changes you can make to it.
Now, you do need to establish strict guidelines on their usage, as WAVE links in multiple-directions can be very difficult to trace.

 

[JonSelby]

I'm back in the office and a big thanks for the comments guys ...

@rg006
My understanding (or assumption/miss-assumption) of UDF's is that they would be most applicable where you are pumping out designs with lots of simliar features. So I imagine someone designing chairs and they might have a UDF of a chair wheel that they pop onto different designs and adjust the wheel size to suit the chair style.
Am I on the right lines here?

What I'm doing is completely unique from project to project. So at present I'm designing some 'shell' coverings where I need a left set and a right set with each set being split into an inner and an outer. So I really don't want to carry out all my modelling four times and risk something being dissimilar between the four.

@hudson888
I'm entirely in agreement - and very elegantly put!
I'm working in a team where most people take pride in producing good quality models. Even where I've picked up poor examples, the designer has always been in agreement that in hindsight it wasn't the best approach. I guess that's an effect of how designs evolve and part of recognising that we are all on a learning curve - just at different phases! (and with some people transitioning from Solidworks ... !!)

The labelling is also key to me. I've always labelled a datum and sketch which preceed a chunk of modelling so you can flick through and find a feaure with all elements grouped together. For me that's a legacy from the bushy history tree in I-deas.

So on WAVE ... If you WAVE from another part I think the traceability backwards to the master is fairly obvious. But the other way to me isn't so clear. This is where it might be nice to introduce a labelling best-practice.
Also I wonder whether the drawing (being the main sign-off) could be the place to drop a note to future designers to alert them that their changes may have effected other parts.

Cheers for your comments.
I shall mull them over and endeavour to bump into CAD support by the 'water cooler' and slip in my thoughts!!!