When to choose Part Design WB or Part WB?

[chrisb]

I mostly use Part Design WB because I guess I think more like it does.

I think it's as easy as that. And as complicated.

Where you finally end is a way of personal preferences. I see many power users here preferring PartDesign, but by far not all. I wouldn't say that the german forum is dominated by Part users, but they have a powerful position.

For the very very first steps I see advantages of Part workbench. It is a very intuitive approach to create primitives, join them, cut them. But after going through the basics things seem to become complex. The tree is deeply nested, the sequence of steps is not as easy to follow as with PartDesign.

PartDesign seems a good start. If you understand Sketch, Pad, and Pocket, you are almost there. Revolve, Loft and Sweep can be seen as special pads, their subtractive counterparts as special pockets. The tree in PartDesign remains much simpler, because the boolean opertion is always included as the feature adds or subtracts by itself.

Alas, PartDesign does not yet offer all the things that Part workbench has, such as faces, slicing, mirroring whole objects, etc. and the pattern functions are not on par yet with their Draft counterparts.

That's where a combination of both worlds is required.

Going the way proposed by vocx seems to be of rather theoretical interest. Besides the fact that it doesn't work reliably if you start a body with a boolean, I see here only very rare cases where it is sensible to use a PartDesign boolean. It seems to me that they combine all disadvantages from both worlds: The tree becomes as unreadable as in Part workbench, if not worse; and editing things inside of such a boolean is almost impossible for a short term user without a heap of explanations.

I for myself use PartDesign. If needed I use Part and Draft on top of PartDesign bodies. If I have to continue working on those results I think twice about using it as a BaseFeature in PartDesign vs. continuing in Part workbench.

Basically I recommend:
- Play around with Part workbench
- Start with the PartDesign Tutorials from the wiki, and make some serious models
- After feeling comfortable with PartDesign, go back to Part workbench and learn in more detail about its capabilities
- Decide what suits best your personal preferences.

[fcaduser]

I recurrently see the statement that Part WB is a somewhat antiquated tool for doing models. I only use this one and I really like the fact the boolean structure is strikingly exposed. Whichever Part or PartDesign one eventuality choses, the fundamental point is robustness of your model, and I think it has a *lot* more to do with the user skill and his garnered good practices. That said, I strive not to be stubborn, I will make a global reappraisal of PartDesign WB.

[drmacro]

Miscellaneous musings about various posts in this thread.

This is exactly the sort of discussions I wanted when I posted this thread. I'm relatively agnostic about Part versus Part Design, but, way back in the 1970's I began my journey in the CAD world with a modeling software that allowed the user to draw a profile and extrude it into space. (In fact, I was a developer, instructor, and consultant on that product for around 20 years.) Indeed, this is how drafting was taught for the better part of last century...draw a profile, project it into a right angle (or other angled aux) view, repeat to show all sides necessary for manufacturing, and dimension accordingly. This, to me, always seemed a more intuitive way to create models that having to imagine what boxes, spheres, etc. added and subtracted would make the desired "thing". Though, some level of CSG modeling was typically used to shape extruded models as well.

For me, much of what Part WB offers is the boxes and spheres approach, with a lot of other functionality. It is unfortunate, that some of that functionality has not yet transferred over to Part Design WB. And. of course, there are plenty of reasons for that.

As for the Assembly4 points. Again, I'm sure there are coding issues that make the container process necessary. But, it certainly clouds the issue of what models are usable and how to pre-process them for use, etc. As evidenced by the discussion of App_part, Std_part, etc. For me, with almost all models done in Part Design WB, it means using Assembly4 requires extra steps for every part in the assembly. Where, with A2Plus for example, I just grab the part, specify where it goes with constraints to other parts and done. (Note: I have spent minimal time with Assembly4, so this may be naive...)

[chrisb]

I don't think that at the current state of the assembly situation should drive the selection of how to model. It should be vice versa. A good assembly module should of course work with Part and PartDesign without much hassle.

To come back to topic: Teaching newbies should not be driven by the choice of an assembly workbench.

[RatonLaveur]

I thank the contributors to this discussion so far and see that preference and FreeCAD's own history plays a significant role in the choice.

I see as well that there are tools in Part WB (or even Draft) that are not yet fully implemented in PartDesign and that clouds the issue. It is true that I use about 80% PartDesign to produce industrial parts and the rest with Part. So it seems that one obstacle that this discussion reveals is incomplete integration of "legacy" tools (draft arrays / part tools) into PartDesin.

I also tend to agree with Chris that the assy module should not dictate the modelling type. Then again this is a difficult assertion since the assy modules in question were made freely by some individual based on their needs/wishes. My opinion, right or wrong, is that any assy should be compatible with PartDesign first and Part second, and that's only because industrial designers and mechanical engineers would work with PartDesign first and foremost. Overall the main issue here seems to be that the the Part container is not well integrated into the PartDesign workflow with regards to assy modules.

By this i mean that currently, it's really up to me whether i use part containers whereas bodies are necessary. I feel like the "forced" workflow for PartDesign should be overhauled. Part is a powerful container and should be perhaps featured more prominently in PartDesign, thus helping the assy modules with LCS and intercompatibility.

[drmacro]

I'd have to agree with chrisb, ideally the assembly module choice shouldn't determine the workbench (i.e. either Part or Part Design).
(aside, is there a way to tag someone, like prefixing their handle with an "@" ?)

Generally, I think the Part Design can do a large percentage of what is needed to design mechanical parts and should be encouraged as the beginners place to start and get comfortable.

As they become familiar with FC then trying to grok where and when to use other workbenches becomes easy to absorb.

(This touches on another thread; the discussion of what to expose to the new user on the first launch of FC.)

[chrisb]

The @drmarco does not work here. We use a quote with the contents "ping" instead.

Code: Select all

[quote=drmacroTheQuoted user_id=2867]
ping
[/quote]

Most important is the user_id value, which can derive from another post, or from the peoples menu. The text above yields the quote below.

ping

[vocx]

I am actually quite curious, i always feel like the Part way of modelling with primitives is a "lesser" version of the true parametric modelling of PartDesign.

"Parametric" in a strict sense means that the object can have its properties changed, and the object will change its geometry. It doesn't mean that it's very user friendly. So, Part Workbench is parametric, but most of the operations with it are rather simple. Meanwhile, PartDesign adds a bit more "logic" and user interfaces to make the creation of objects more friendly.

I feel that primitives modelling is a good entry point for complete beginners to start thinking in volumes but has no real use in engineering, and is too limited for design modelling (curves) or mesh modelling (blender...etc).

If I'm not mistaken this is exactly how it was done in the past for all engineering work. There weren't a lot of pretty interfaces. You had to enter a bunch of commands to create and modify solid objects until you got the desired shape.

As such i am always surprised at the maintained prevalence of Part WB in FreeCAD given that PartDesign has been around and powerful enough since 0.17 (PartDesignNext) to rival expensive pro software.
...

The Part Workbench is where the code of the OpenCASCADE Technology (OCCT) geometric kernel is first implemented. Therefore, Part basically demonstrates how to use the most basic functions of OCCT. As long as every function is used in some way in another workbench you don't really need to have Part Workbench any more. However, a lot of people just like using the Part Workbench to do modelling. It's what they are used to and they like it that way. So, removing this workbench is a bit extreme, and more importantly, as you know, some things can only be done with Part, like Part Helix.

[C_h_o_p_i_n]

I think it's depending on how you ususally design the things your going to build.

Do you usaually combine semi-finished materials using "holes" & "screws", welding etc.
Do you usually mill a block of material until you get the thing you need.
Or do you usually use additive methods like 3d-Printing.

Also - it depends on the way you are used to think - are you more used to derive every measurement/dimension (e.g. where a hole has its place) relative to a vertex or do you rather use absolute/relative dimensions to reference a common origin.

insead of using additive/subtractive features in PD I would have found it more intuitive to use parametrically constrained boolean functions w/ sketches - to add or remove material from/to a solid.

just my 2 cents

[fcaduser]

A point in this debate is paramount : does one of the two tools can do something the other can't (and vice-versa) ? I'm not talking about their respective workflows and user-friendliness, but really about their capability to generate a complex shape with as little limitation as possible. At this stage I haven't detected something I could do with PartDesign whereas I can't do with Part.

@ vocx
Geometric primitives are essential features ! It's a must have.