Welcome to the Onshape forum! Ask questions and join in the discussions about everything Onshape.
First time visiting? Here are some places to start:- Looking for a certain topic? Check out the categories filter or use Search (upper right).
- Need support? Ask a question to our Community Support category.
- Please submit support tickets for bugs but you can request improvements in the Product Feedback category.
- Be respectful, on topic and if you see a problem, Flag it.
If you would like to contact our Community Manager personally, feel free to send a private message or an email.
Can robust complex surfaces be modelled.
euan_dykes
Member Posts: 25 PRO
I was having a discussion with an industrial designer about advance surfacing, and if he believed it would be possible to produce robust designs that allow us to have variables that control final outputs. E.g. We can have configuration tables driving all the sub versions of a master design. First this concept seemed like a blind spot in his thinking, he uses Solidworks and the seldom configuration tables. To change a design he takes a copy of a file and modifies it. All us Onshape users know this is madness because what happens if the master design is changed? All that manual work has to be repeated over again just to stay in step with the versions.
His believe is that complex curvature can't be fully definable. That at some point a designer's eye has to make the call on what looks best. Which I do agree with.
I'm looking for success stores around complex surfacing that I can relate to packaging design. Think laundry detergent bottles and their complex surfacing. Any links, CAD forums etc most welcome. I'm not trying to improve the status quo, I'm trying to leapfrog how we do design.
Best Answer
-
MichaelPascoe
Member Posts: 2,616 PRO
Yes, but it depends on how robust you make your parametric model. If the entire model is made up of manual move faces and things that aren't stable, then no, you can't make it robust. If you build your model from the ground up with robust surfacing in mind, then absolutely.
One way to ensure downstream features don't break is to use attributes as well as query variables. This way no matter what shape your part takes, your downstream features are using attributes or query variables to find their targets instead of relying on the designer to make new manual inputs.
Do you have an example part your trying to do this with?
Learn more about the Gospel of Christ ( Here )
CADSharp - We make custom features and integrated Onshape apps! Learn How to FeatureScript Here 🔴0
Answers
Yes, but it depends on how robust you make your parametric model. If the entire model is made up of manual move faces and things that aren't stable, then no, you can't make it robust. If you build your model from the ground up with robust surfacing in mind, then absolutely.
One way to ensure downstream features don't break is to use attributes as well as query variables. This way no matter what shape your part takes, your downstream features are using attributes or query variables to find their targets instead of relying on the designer to make new manual inputs.
Do you have an example part your trying to do this with?
Learn more about the Gospel of Christ ( Here )
CADSharp - We make custom features and integrated Onshape apps! Learn How to FeatureScript Here 🔴
It really depends on how much a set of designs relates to each other, and how much work one wants to do with configuring things. A classic example would be a set of kitchen knives. They have similar features, but very different proportions between a pairing knife, large chef knife, boning knife, bread knife etc. In that case it might make more sense to duplicate or branch rather than configure. In the case of shoe sizes, I would think configurations would make more sense. There will be some things that just scale, and others that are size invariant (shoe lace holes, some tags).
It also depends a lot on how the designs/products will evolve over time. If you have something in a small, medium and large, will you someday need to add an extra small or extra large? Will you go back and need different internal mounting features that are being updated across the product line all at once?
Simon Gatrall | Product Development Specialist | Open For Work
The short answer is yes. I have seen some amazing examples of complex and robust surface modeling from the Onshape Forums and from Onshape commercial customers. Complex surfaces are defined by curves which are in turn defined by parameters. The underlying math of how the surfaces are formed is complex, but it isn't random.
There are some good examples in the surfacing courses in the Learning Center.
Maybe not a laundry detergent bottle, but here is a very cool bottle from Evan Reece.
I also love this video:
Other examples:
@euan_dykes I'm trained as an Industrial Designer, and have done a lot of professional projects in Onshape with teams of people at my former agency. Yes it is definitely possible. There's huge value in a unified team. I'm always surprised when teams allow the ID to be thought of as so separate from the rest of the product development as if there's nothing valuable for them to do once an initial concept is together. As any product developer knows, the process is a series of experiments, pivots, and compromises, and I believe that a good Industrial Design culture encourages the ID people to steward their design vision though that gauntlet up to production.
A side note on configuring aesthetic surfaces: I always look for subtle geometric relationships I can establish, so my curves/surfaces can update in a pleasing way even for sizes I've not planned for. But as your ID friend says, there are subjective aspects to surfacing in ID (sometimes you gotta eyeball it), however I don't think that precludes configuration. You can position points by eye, even if they are driven by configured dimensions.
The Onsherpa | Reach peak Onshape productivity
www.theonsherpa.com
@Matt_Shields I literally forgot about that bottle model lol. The surfacing actually wasn't mine. I just demoed the Attractor Pattern on someone else's model since they asked about it on the forum. Cool model though.
The Onsherpa | Reach peak Onshape productivity
www.theonsherpa.com
Here's a use case that I see great benefit in having robust model. We have this bottle design, and there's like 20 size versions of it. There's also versions that are PP and PE, slight geometric changes to improve mould-ability of each material. I can already see the common design features. E.g. the upper label bumper on the right 3 is almost the same height. The handle is the same size and form for each bottle.
There's also a level of CAD that is above robust, I call it adaptive. Our suppliers do this, they can deform the surfaces to adapt to how the moulding shrinks. Those front faces might be perfect for labelling in CAD, the bottles that come out can cause issues.
Onshape has had a bunch of surfacing (and other) improvements over the last year that would improve the workflow demonstrated in this video, but I had to share it because it looks a lot like your green bottle :)
Perhaps entity attribute tagging and query variables would then go beyond the robust category and fall into the adaptive category. But yes, totally possible if you set it up correctly.
Query Finder. This is a very different approach from using the built in
Query Variables as query variables are actually saving queries not just tagging entities with an attribute. I found that both have their perks and their limitations.
Fyi the easiest way to tag and find entities with attributes is currently with the
Tagging an entity with an attribute, will keep that attribute with the entity even if the entity splits into several different parts or has a boolean applied to it, which can be useful downstream. However, it does not search for a query so new edges will not be found. This way can also let you pass selections through derives, which is not possible with a query variable at the moment.
Query variables or just queries are excellent because the query search can be re-evaluated on use, so it would even find new edges. But if the edges are not within the original query like with a split or boolean for example, it may not find the edges since its a different query all together.
Learn more about the Gospel of Christ ( Here )
CADSharp - We make custom features and integrated Onshape apps! Learn How to FeatureScript Here 🔴
@MichaelPascoe You can leverage the qAdjacent filters in my latest iteration of Query Variable+ to march across a selection and grab new vertexes, edges, or faces that might not get grabbed by re-evaluating a standard query variable. I'm planning to add other functions like the distance or size queries to this feature so that they can be combined with features already integrated like the qGeometry filters to push query variables into that truly adaptive category for cases where the entire seed might change or disappear.
Like hypothetically if someone had a part studio that switched between a dozen configurations of import geometry and wanted to make stable mate connectors and references to faces in an assembly when they switch configs.
Derek Van Allen | Engineering Consultant | Meddler@Derek_Van_Allen_BD That looks legit! I don't see a link to your feature anywhere in that post. Can you post a link?
Oh wait, I found it. Just wasn't called Query Variable+.
Learn more about the Gospel of Christ ( Here )
CADSharp - We make custom features and integrated Onshape apps! Learn How to FeatureScript Here 🔴
@MichaelPascoe updated the thread so there's a link at the top, I just realized my first link in that post was to the old query explorer which isn't the most helpful thing for discovery lol
Derek Van Allen | Engineering Consultant | MeddlerHi Evan. Thanks for the comments here. It's good too have some ID experts out there challenging the status quo. Do you believe that many ID put robust modelling in the too hard basket? I agree with the curvature constraining comment. I've been learning G3 level curvature and it's quite easy to define because it's just maths, just needs a vector and a magnitude and you have your curve regardless of distance. Ok you do need some ramping in space. That's why I think it's best to design the smallest the design will be and configurate to be bigger.
Thanks Simon. Actually you've explained something that's been a bit of a blind spot for me with Onshape. Version vs configuration. In robust modelling you would have a version and then make the configurations. Then if a new version is needed, the configurations are already established, so all the configurations inherit the design change.
Well I'm sure I can't speak for the whole profession, but I guess some people are less inclined to CAD mastery. I've learned this stuff because it's fun and makes my life easier when designing (I'd rather build a good model than debug a bad one). But the bigger thing is that there's a whole spectrum of technicality and focus within ID: Some focus hard on user research, design thinking, problem definition, ergonomics, beautiful forms, marketing renderings etc, which are insanely valuable, and some of these designers are truly great, but all of those things can be done without deep knowledge of stable production-level CAD. On the other end of the spectrum some ID people become more technical and find themselves straddling design, engineering, and sourcing. I suspect those folks are going to also gravitate to CAD more readily. To my thinking, the best organizations don't fight this. They know what sub-skills of ID they're looking for and the job descriptions they fill are specific enough to hire what they need. "Industrial Designer" is nearly as broad a title as "engineer". If I were building an ideal product dev team I think I'd have one of each focus of ID (research-centric and product-dev-centric) plus at least one mechanical engineer to create continuity from big problem definition to design details and through functionality and DFMA.
The Onsherpa | Reach peak Onshape productivity
www.theonsherpa.com
At Trek Bikes we leverage the power of Onshape's configuration options for our Carbon Fiber and Alloy bike designs. Think about modeling a bicycle frame that comes in many sizes. Our long standard has been develop one size first, typically a middle size, and really get the ID dialed. Then duplicate, change the centerline geometry to the next size, update, and reshape.
The problem occurs once we get to the bookend sizes. Often times due to tighter angles on size small, and really stretched shapes of an XL we miss something that causes the manufacturing process already determined to not work, or additional size specific parts are needed. In order to keep the manufacturing process the same across the size run and keep sku's down we then end up going back and updating the already completed first sizes so they can meet those needs. It's a lot of rework as you can imagine.
I'm working to master all sizes together. Meaning I configure our CL geometry sketch, and hang supporting sketches and features on it so in a way as the CL is changes between sizes the entire model updates.
I think @EvanReese says it well above;
"A side note on configuring aesthetic surfaces: I always look for subtle geometric relationships I can establish, so my curves/surfaces can update in a pleasing way even for sizes I've not planned for. But as your ID friend says, there are subjective aspects to surfacing in ID (sometimes you gotta eyeball it), however I don't think that precludes configuration. You can position points by eye, even if they are driven by configured dimensions."
Expanding on that,
I leverage sketch constrains and helper geometry over dimensions. Think of in terms of proportion and position things accordingly. IE this point should be 2/5's along the length on a line across the size range. There are a lot of tools in Onshape that use percentages for placement, use them instead of static dimensions.
You gotta eyeball them, so where needed configure a dimension, do this as little as you can get away with.
The bridging curve is your best friend. Its robust, and does a great job holding consistent shape across changing parameters within reason. The magnitudes can further be configured for outlier sizes.
I haven't dug into Query Variables yet, shame on me - too busy!, but before they existed I have been using @lana's Assign Identity FS to configure selected vertices, edges, or faces. This FS helped almost eliminate the need for whole size specific subsets of feature's needed to generate a specific size where the original set could not get it done.
There's more to it than this for really complex models, hope to share more later.
The past few projects I have been able to configure the full size run up to our first release. This has saved loads of time later in the production phase as we are able to foresee most issues from the start. Once these products are made public, I'll be sure to share more.
Man, great writeup, Nick!
The Onsherpa | Reach peak Onshape productivity
www.theonsherpa.com
Thank you 🙏
These are great insights. I'm not a ID, I was a web designer than turned mechanical engineer. I found I had so many overlapping skills with IDs. I think I've had the privilege to watch ID from the side. The engineer in me is interested in the system of design. And I see the output being prototypes, screen renderings etc. And then people managing the business are suppose to take these outputs and make decisions. From my 20 years experience doing this I've boiled it all down to risk management. A design is only as good a positive result from trustworthy consumer insights study, because the risk of it being not accepted by the target consumer has evidence to prove it's ok. People feel safer to say lets go with that. Business decisions are a leap of faith because even engineers can guarantee anything. Can get lucky on the market, or it flops and it had nothing to do with how well the design process was done.
I see robust modelling as an enabler for collaborative design. And on Onshape it's even more enhanced. In web design, content management systems liberated the designer from becoming a "web master" and having to update lines of HTML code. I used CMS systems to enhance the design process, my clients could start with the content and I could use the content to design a better website. Some years later when I was a CAD tech in SW, I had a boss who always wanted to see the design first and then make changes. This is when I developed robust CAD methods. My web development experience helped, as I was driving the design with central variables. I could sit with my boss and go, what configuration do you want? It was real time collaboration with the decision maker. I was reducing risk too, because if he got new information to a design better, he knew it was a quick change on my end.
In startups adaptation was the best mindset to have. In large organisations with scale it's a different kind of risk management. Trying to sell the concept of robust modelling is proving challenging. Even if I do make a convincing demo, it's lost on them as it's so far ahead of what they are use to. The silo mindset = passing on troubles to other departments /outsource it. The worst part of it is this unfounded belief things can't be done differently. That's the worst because it suggests there's not much hope for it ever changing. Hence for being a bit lost and lonely in this quest and want to connect to some like minded experts. The Onshape team have also been great.
Thanks Nick. This is gold. I would love to have a call with our team to show them how you use robust modelling and how you go back up the design process in the development phase. This is something we are also missing. The fear of going back to the drawing board has us going down dead end roads and having to bush wack our way to market.
With scale-able sketches, I also developed this fractional relationship method. To developed it I had to break my models to develop the best constraint setup. And I wonder if many IDs don't like breaking things, barely getting things to work just to hit their deadlines. I've also seen many ID start projects last minute and doing all nighters to deliver. In mechanical engineering you need the time for the mental calculations, so starting early is critical.
My next question then for you and @EvanReese. Can robust modelling be taught? Is it upskill-able or is just something people have to develop by themselves?
BTW: I'm a keen mountain biker again. Heck I was even a pro mechanic at Johnsonville Cycles in Wellington, NZ for a year, best job ever. I loved to geek out about bike fit with the customers and regulars that frequented the shop. I moved to Germany some 12 years ago and stopped doing mtb, because of the lack of trails in the region. Turns out there are trails so I got a mtb. Orbea Laufe for 999 euros this year. Dropper post, through axles front and rear, 12x 1 by, super slack geo. Loving this new gen of mtbs. My last bike was a 26er before tubeless and droppers, so it's a big change to an aggressive 29 hardtail. Thanks for all the pioneering over the last 20 years and getting away from road bike influence. I'm riding down things I think I would have not on my old bike. Such a buzz.
@euan_dykes Pioneering configured frame models at Trek was a personal endeavor fueled from past projects that required rework. It wasn't until practiced enough to reveal in an actual production project that it was excepted. It certainly is a masterly level skill that requires understanding of other advanced skills; configurations, surfacing (Class A), clean modeling, for it be possible.
The process walks a fine line in terms of spending too much time modeling a full size run. Something I've had to monitor through a projects timeline. Modeling all sizes at once was not our long time process, so it is seen by some as time not well spent. This might depend on your business or product.
For our product we are able to leverage size run modeling in steps. This allows us to review at certain levels of model completeness. There are basically 3 levels that we could review at; Primary Surfaces, Intermediate Sections, and Final or Fully modeled. By building our model in these steps if crunched on time we can at least review at an earlier level of completeness. Primary surfaces are the main tube shapes, Intermediate Sections contain the heavier surfacing blends, and the Final combines the sections and we model in any features that span the whole part.
I think when you look at a lot of the training material out there robust modeling is something that is always at the forefront of any presenters demonstration. Examples for size run surfacing models could be pulled together. For me it would be a matter of time to review my work a distill an more simple example to get these ideas across.
That's awesome! Glad to hear you are enjoying mountain biking in Germany. Thanks for riding bikes and I hope you can consider a Trek in the future!